JP2009511094A - Biometric sensor - Google Patents

Abstract

A method for evaluating the authenticity of a sample presented for biometric evaluation is described. The sample is illuminated under different optical conditions. Light scattered from the sample is received. A plurality of images are formed, each image being formed from the received light for one of the optical conditions. A set of quality sensitivities is generated, each quality sensitivity being generated from one of the images. It is determined whether the generated quality sensitivity is consistent with a sample that is a genuine, unhidden biological tissue.

Description

(Citation of related application)
This application claims the benefit of each of the following filing dates: U.S. Patent Application No. 11 / 216,006 ("COMPARATIVE TEXTURE ANALYSIS OF TISSUE FOR BIOMETRIC SPOTATION DETECTION", filed September 1, 2005 by Robert K. Rowe). ), U.S. Patent Application No. 11 / 458,619 ("TEXTURE-BIOMETRICS SENSOR", filed July 19, 2006 by Robert K. Rowe), U.S. Patent Application No. 11 / 458,607 ("WHITE LIGHT SPECTRAL BIOMETRIC"). SENSORS ", Robert K. Rowe, et al., Filed July 19, 2006).

  The present application relates generally to biometrics. More specifically, this application relates to methods and systems for performing biometric measurements using spectral information.

  “Biometrics” generally refers to statistical analysis of biological properties. One category of biometrics generally includes “biometric identification” that functions in one of two ways and provides automatic identification of people or verifies their estimated identity. Biometric sensing techniques measure a person's physical or behavioral characteristics and compare such characteristics with similar pre-recorded measurements to determine if there is a match. Physical features commonly used for biometrics include the face, iris, hand shape, vein structure, and fingerprint patterns that are most prevalent among all biometric identification features. Current methods for analyzing collected fingerprints include optics, capacitive, high frequency, heat, ultrasound, and some other less common techniques.

  Most fingerprint collection methods rely on measuring skin properties at or near the finger surface. In particular, optical fingerprint readers typically rely on the presence or absence of a refractive index difference between a sensor platen and a finger placed thereon. If the fingerprint groove filled with air is above a specific location on the platen, total internal reflectivity ("TIR") occurs in the platen due to the refractive index difference between the air and the platen. Alternatively, if an appropriately reflective skin is in optical contact with the platen, the TIR at this location is “disturbed” and light can cross the platen-skin contact surface. A map of TIR differences across the area where the finger touches the platen forms the basis for conventional optical fingerprint reading. There are a number of optical arrangements used to detect this change in optical contact surface in both bright and dark field optical arrangements. In general, a single quasi-monochromatic beam is used to make this TIR basic measurement.

  There are also non-TIR optical fingerprint sensors. In most cases, these sensors illuminate the front, sides or back of the fingertip depending on the arrangement of quasi-monochromatic light and diffuse the light through the skin. The fingerprint image is formed by the difference in light transmission across the skin and platen boundary for the ridges and grooves. As is known to those skilled in the art, the difference in light transmission is due to the change in Fresnel reflection characteristics due to the presence or absence of an intermediate gap in the groove.

  Optical fingerprint readers are particularly susceptible to image quality problems due to non-ideal conditions. If the skin is excessively dry, the refractive index match with the platen is reduced, resulting in poor image contrast. Similarly, if the finger is very wet, the groove may fill with secretions, causing optical coupling throughout the fingerprint area and greatly reducing image contrast. Excessively fine features, such as too little or too much finger pressure on the platen, dirty skin or sensors, aging and / or worn skin, or certain ethnic or very young children A similar effect may occur when there is. These effects reduce image quality and thereby reduce the overall performance of the fingerprint sensor. In some cases, commercially available optical fingerprint readers incorporate a thin film of soft material such as silicone to reduce these effects and help restore performance. As a soft material, the membrane is susceptible to breakage, wear, and contamination, and the use of the sensor is limited without maintenance.

  Optical fingerprint readers based on TIR, as well as other aspects such as capacitance, RF, etc. typically produce images that are somewhat affected by non-ideal imaging conditions present during acquisition. Thus, the analysis of the texture characteristics of the resulting image is influenced by sampling conditions that can limit or obscure the patient's ability to observe the texture characteristics of the skin. The result is that the texture is limited in practicality in such a sensing aspect.

  Biometric sensors, particularly fingerprint biometric sensors, are generally prone to being defeated by various forms of camouflaged samples. In the case of fingerprint readers, various methods are known in the art to present a reader with a certified user fingerprint pattern incorporated into certain inanimate materials such as paper, gelatin, epoxy, latex, etc. It is. Thus, even if it can be considered that the fingerprint reader reliably determines the presence or absence of a matching fingerprint pattern, it ensures that the matching pattern is obtained from a real biological finger (many It may also be difficult to ascertain with common sensors), which is also important for overall system safety.

  Another way in which some biometric systems may be broken is through the use of replay attacks. In this scenario, an intruder logs the signal from the sensor when an authorized user is using the system. The intruder then gains access to the biometric secured system by bypassing the sensor itself by manipulating the sensor system so that a pre-recorded authorization signal is injected into the system.

  Common methods for making biometric sensors more robust, more reliable, and less error-prone are sometimes referred to in the art as “double”, “combined”, “layered”, “fused”, “multiplexed”. Combining biometric signal sources using a method called “biometric”, or “multifactor biometric” sensing. In order to provide enhanced safety in this way, different techniques can be defeated by measuring different parts of the body at the same time and using different samples or techniques that break down the different sensors combined. Biometric technology is combined in such a way as to withstand. When techniques are combined in a way that identifies the same part of the body, they are said to be “tightly coupled”.

  The accuracy of non-invasive optical measurements of physiological specimens such as glucose, alcohol, hemoglobin, urea, and cholesterol is adversely affected by changes in skin tissue. In some cases, it may be advantageous to measure one or more physiological analytes in conjunction with biometric measurements. Such dual measurements have potential interest and application in both commercial and law enforcement markets.

  Accordingly, there is a general need in the art for improved methods and systems for biometric sensing and analyte estimation using multispectral imaging systems and methods.

  Accordingly, embodiments of the present invention provide a method for assessing the authenticity of a sample presented for biometric assessment. The sample is illuminated under a plurality of different optical conditions. Light scattered from the sample is received. A plurality of images are formed, and each such image is formed from the received light for each of the plurality of different optical conditions. A plurality of quality sensitivities are generated, and each such quality sensitivity is generated from each of the plurality of images. It is determined whether the generated multiple quality sensitivities are consistent with a sample that is a genuine, unhidden biological tissue. In a particular embodiment, the quality sensitivity can comprise a degree of image contrast.

  Illumination of the sample under different optical conditions can be achieved by illuminating it with multiple lights having multiple different wavelengths, illuminating under multiple different polarization conditions, illuminating at multiple different illumination angles, etc. Can be achieved.

  In some embodiments, the quality sensitivity can be generated by performing a spatial moving window analysis of each of the plurality of images. In some cases, the plurality of quality sensitivities may comprise a degree of image contrast within a certain spatial frequency. For example, when the sample is illuminated with light having a plurality of different wavelengths, it can be confirmed that the image contrast under red illumination is lower than the image contrast under blue illumination. Alternatively, it can be confirmed that the image contrast under red illumination is less than a predetermined value. In some cases, the spatial moving window analysis can be performed by calculating a moving window Fourier transform for the plurality of images. Alternatively, it can be performed by calculating the moving window center and moving window variation of the plurality of images. For example, in one embodiment, the moving window centrality includes a moving window average value, and the moving window variation includes a moving window standard deviation. In another embodiment, the moving window centrality comprises a moving window average value, and the moving window variability comprises a moving window range.

  The authenticity of the sample can be determined by applying a multidimensional scaling method to map the plurality of quality sensitivities to points in the multidimensional space. Thereby, it can be determined whether or not the point is within a predetermined region of the multidimensional space corresponding to the sample that is a real biological tissue that is not hidden. In one embodiment, the multidimensional space is a two-dimensional space.

  In another set of embodiments, a method for performing a biometric function is provided. The part estimated to be the individual's skin is illuminated with illumination light. The part presumed to be the skin is in contact with a certain surface. Light scattered from a site estimated to be the skin is substantially received in a plane including the surface. An image is formed from the received light. Image quality sensitivity is generated from the image. The generated image quality sensitivity is analyzed to perform the biometric function.

  In an embodiment, the biometric function comprises an anti-spoofing function, and in such an embodiment, the image quality sensitivity is analyzed to determine whether the site presumed to be skin includes biological tissue. The In other embodiments, the biometric function comprises an identification function, and in such embodiments, the image quality sensitivity is analyzed to identify an individual's identity. In other embodiments, the biometric function comprises a demographic or anthropometric function, and in such embodiments, the image quality sensitivity is analyzed to estimate an individual demographic or anthropometric characteristic. The

  The surface may be the surface of an image detector, and light scattered from a site presumed to be the skin is received by the image detector. Alternatively, the light pattern can be translated from the plane to an image detector located outside the plane without substantial degradation or attenuation of the pattern, and the translated pattern is the image detection. The light is received at the instrument. In various embodiments, the light scattered from the presumed site of skin can be received at a monochromatic image detector or a color image detector.

  In some embodiments, the illumination light is white light. The image may include a plurality of images corresponding to various wavelengths. Therefore, the image quality sensitivity can be generated by performing a spatial moving window analysis of each of the plurality of images. For example, a moving window Fourier transform can be calculated for the plurality of images. Alternatively, the moving window centrality and moving window variation of the plurality of images can be calculated.

  In analyzing the generated image quality sensitivity to perform the biometric function, the generated image quality sensitivity can be compared to a reference image quality sensitivity. In some cases, the reference image quality sensitivity is generated from a reference image formed from light scattered from a reference skin site, and the site presumed to be skin is substantially different from the reference skin site. In one embodiment, in performing the biometric function, the spectral characteristics of the received light are compared with reference spectral characteristics.

  In yet another set of embodiments, a biometric sensor is provided. The sensor comprises a surface, an illumination subsystem, a detection subsystem, and a computing unit. The surface is configured to contact a site presumed to be skin. The illumination subsystem is arranged to illuminate the presumed part of the skin when the presumed part of the skin is in contact with the surface. The detection subsystem is arranged to receive light scattered from a site presumed to be the skin, and the light is received substantially in a plane including the surface. The arithmetic unit is linked to the detection subsystem and has instructions for forming an image from the received light. There are also instructions for generating image quality sensitivity from the image and for analyzing the generated image quality sensitivity to perform the biometric function.

  There are a number of different biometric functions that can be implemented by such sensors. In one embodiment, the biometric function includes an anti-spoofing function, and the arithmetic unit includes an instruction for determining whether the site estimated as the skin includes a living tissue. In another embodiment, the biometric function comprises an identification function, and the computing unit comprises instructions for identifying an individual's identity from the generated image quality sensitivity. In another embodiment, the biometric function includes a demographic or anthropometric characteristic function, and the arithmetic unit issues a command for estimating the demographic or anthropometric characteristic from the generated image quality sensitivity. Have.

In some cases, the biometric sensor further comprises an image detector.
In one such embodiment, the surface is the surface of an image detector, the detection subsystem comprises the image detector, and light scattered from a site presumed to be the skin in the image detector. Configured to receive light. In another such embodiment, the image detector is located outside the plane. The optical arrangement is configured to translate the pattern of light from the plane to the image detector without substantial degradation or attenuation of the pattern. The detection system includes the image detector and is configured to receive the translated pattern in the image detector. In various embodiments, the image detector can comprise a monochromatic image detector or a color image detector.

  In some cases, the illumination subsystem is configured to illuminate the presumed site with white light. The image may comprise a plurality of images corresponding to various wavelengths, and the instructions for generating the image quality sensitivity are for performing a spatial moving window analysis of each of the plurality of images. Provide instructions. For example, in one embodiment, there may be instructions for calculating a moving window Fourier transform based on the plurality of images, while another embodiment provides a moving window centrality and movement of the plurality of images. Has instructions for calculating window variability.

  In one embodiment, the instructions for analyzing the generated image quality sensitivity to perform the biometric function comprise instructions for comparing the generated image quality sensitivity with a reference image quality sensitivity. Such reference image quality sensitivity can be generated from a reference image formed from light scattered from a reference skin site, and the site estimated to be the skin is substantially different from the reference skin site. In one embodiment, the computing unit further comprises instructions for comparing spectral characteristics of the received light with reference spectral characteristics when performing the biometric function.

  In another set of embodiments, a biometric sensor is provided. The white light illumination subsystem is arranged to illuminate an area presumed to be an individual's skin with white light. The detection subsystem is arranged to receive light scattered from a site estimated to be the skin, and includes a color image device on which the received light is incident. The arithmetic unit is linked to the detection subsystem. The arithmetic unit has a command for extracting, from the received light, a plurality of spatially distributed images of the portion estimated to be the skin by a color image device. The plurality of spatially distributed images correspond to various amounts of a person's illuminated tissue. The computing unit also has instructions for analyzing the plurality of spatially distributed images to perform the biometric function.

  In one of these embodiments, the biometric function includes an anti-spoofing function, and the command for analyzing the plurality of spatially distributed images is that the site estimated to be skin is a living tissue. Instructions for determining whether to prepare are provided. In another of these embodiments, the instructions for analyzing the plurality of spatially distributed images to perform the biometric function include the plurality of instructions to estimate an individual demographic or anthropometric characteristic. Instructions for analyzing the spatially distributed image of In yet another of these embodiments, the instructions for analyzing the plurality of spatially distributed images to perform the biometric function determine a concentration of an analyte in an individual's blood. Instructions are provided for analyzing the plurality of spatially distributed images.

  In some embodiments, the biometric sensor can further comprise a platen in contact with the site presumed to be the skin, and the white light illumination subsystem is presumed to be the skin through the platen. Configured to illuminate the site. In other embodiments, if the white light illumination subsystem is not in physical contact with the biometric sensor, the white light illumination subsystem instead illuminates the presumed site. Composed.

  The white light can be provided in various ways in various embodiments. For example, in one embodiment, the white light illumination subsystem comprises a broadband white light source. In another embodiment, the white light illumination subsystem comprises a plurality of narrowband light sources and an optical arrangement for combining light provided by the plurality of narrowband light sources. The plurality of narrow-band light sources can provide light having a wavelength corresponding to each of a set of primary colors. In some cases, the portion estimated to be the skin and the illumination region where the portion estimated to be the skin is illuminated are relatively moving.

  Some embodiments utilize polarization by including a first polarizer in the illumination system that is arranged to polarize the white light. The detection system includes a second polarizer arranged to encounter the received light. The first and second polarizers may cross each other. In other embodiments, the first and second polarizers may be parallel. In some embodiments, the first polarizer may be omitted while retaining the second. In some embodiments, two or more of these polarization options may be combined in a single device. The detection system can sometimes comprise an infrared filter arranged to encounter the received light before the received light is incident on the color imaging device.

  In one embodiment, the site presumed to be the skin is a finger or a palm surface of the hand, and the biometric function comprises biometric identification. The command for analyzing the plurality of spatially distributed images is a command for extracting a surface fingerprint image or a surface palmprint image of a site estimated to be the skin from the plurality of spatially distributed images. Is provided. The surface fingerprint image or surface palm print image is then compared with a fingerprint or palm print image database to identify the individual. In other embodiments, wherein the biometric function comprises a biometric, the instructions for analyzing the plurality of spatially distributed images, instead, the plurality of spatially distributed images to identify an individual. For comparing to a database of multispectral images.

  In yet another set of embodiments, a method for performing a biometric function is provided. The part presumed to be the individual's skin is illuminated with white light. Light scattered from the site estimated to be the skin is received by a color image device on which the received light is incident. A plurality of spatially distributed images of a site presumed to be the skin are derived, and the plurality of spatially distributed images correspond to various amounts of individual illuminated tissue. The plurality of spatially distributed images are analyzed to perform the biometric function.

  In some of these embodiments, the biometric function comprises an anti-spoofing function, and the step of analyzing the plurality of spatially distributed images determines whether the estimated site comprises living tissue. Comprising steps. In other of these embodiments, the plurality of spatially distributed images are analyzed to estimate an individual demographic or anthropometric characteristic. In yet another of these embodiments, the plurality of spatially distributed images are analyzed to determine the concentration of the analyte in the individual's blood.

  The presumed site of skin can sometimes be illuminated by directing the white light through a platen in contact with the presumed site of skin. In some cases, the presumed skin area can be illuminated with a broadband white light source, while in other cases it may generate and combine multiple narrow-band rays that probably correspond to a set of primary colors. it can. The site presumed to be the skin sometimes moves relative to the illumination area where the site presumed to be the skin is illuminated.

  In one embodiment, the white light is polarized by a first polarization and the received light scattered from a site presumed to be the skin is polarized by a second polarization. The first and second polarizations may be substantially crossed with respect to each other or may be substantially parallel to each other. The received light is sometimes filtered at infrared wavelengths before the received light is incident on the color imager.

In some cases, the biometric function comprises a biometric.
For example, the portion estimated to be the skin can be a finger or a palm surface of a hand. Then, the analysis of the plurality of spatially distributed images includes deriving a surface fingerprint or palm print image of the part estimated to be the skin from the plurality of spatially distributed images, and the surface fingerprint or palm print image Or it can be processed by comparing with a palmprint image. In other embodiments, the plurality of spatially distributed images may be compared to a database of multispectral images to identify individuals.

  A further understanding of the nature and advantages of the present invention may be realized by reference to the remaining portions of the specification, and to the drawings, in which like reference designations are used throughout several drawings to refer to like components. . In some cases, a reference sign includes a numeric part followed by a Latin letter suffix, and a reference to only the numeric part of the reference sign collectively refers to all reference signs that have that numeric part but a different Latin letter suffix. For the purpose.

(1. Overview)
Embodiments of the present invention provide methods and systems that enable collection and processing of a wide variety of types of biometric measurements, including integrated multifactor biometric measurements in some embodiments. These measurements can provide an individual's identity as well as true and authentic certainty of the biometric sample being collected. In some embodiments, the sensor uses white light that penetrates the surface of the individual's skin and scatters into the skin and / or underlying tissue. As used herein, “white light” refers to light having a spectral composition adapted to separation into component wavelength bands, which may optionally have a primary color. Common primary colors used to define white light are red, green, and blue, but other combinations may be used in other cases as is known to those skilled in the art. For clarity, “white light” as used in this application may not appear white to human observers due to the precise wavelength distribution and intensity of the component wavelength bands, and the associated differences. It is emphasized that it may have a tint or color. In other cases, the white light may comprise one or more bands in the ultraviolet or infrared spectral region. In some cases, the white light may even be completely invisible to a human observer if it consists of wavelengths in the infrared and / or ultraviolet spectral region. Some of the light scattered by the skin and / or underlying tissue is used to eject the skin and form an image of the structure of the tissue at or below the surface of the skin. Due to the wavelength dependent nature of the skin, the image formed from each wavelength of light provided by the white light may be different from the images formed at other wavelengths. Thus, embodiments of the present invention collect images in such a way that characteristic spectral and spatial information can be extracted from the resulting image.

  In some applications, it may be desirable to estimate other parameters and characteristics of the body, either alone or in combination with biometric measurements. For example, in one such specific embodiment, the ability to measure an individual's analyte level simultaneously with the measurement of a fingerprint pattern is provided. Applications in law enforcement are found in embodiments in which the measured specimen comprises an individual's blood alcohol level, and such embodiments also allow for various commercial applications including limited vehicle access. As described above, the sample measurement based on the measurement and the identity of the individual can be closely linked.

  Skin composition and structure are very different, very complex and vary from individual to individual. Numerous assessments can be made by making optical measurements of the spatial spectral properties of the skin and underlying tissue. For example, a biometric function can be implemented to identify and verify who's skin is being measured, and a viability function is a tissue in which the sample being measured is viable and viable, Can be performed to verify that it is not a type of substance, and can provide estimates of various physiological parameters such as age, gender, ethnicity, and other demographic and anthropometric characteristics, and / or Measurements of the concentration of various analytes and parameters can be performed, including lucol, glucose, blood perfusion and degree of oxidation, bilirubin, cholesterol, urea, and the like.

  In various embodiments, the complex structure of the skin can be used to adjust aspects of the method and system for a particular function. The epidermis, the outermost layer of the skin, is supported by the underlying dermis and lower skin. The epidermis itself may have five distinct sublayers including the stratum corneum, the clear layer, the granular layer, the spiny layer, and the basal layer. Thus, for example, the skin below the uppermost stratum corneum has some properties related to surface topography, as well as some properties that vary with skin depth. The blood supply to the skin is in the dermis layer, but the dermis has a protrusion into the epidermis known as the “dermal papilla” that brings the blood supply close to the surface via capillaries. On the palm of the finger, this capillary structure follows the skin pattern on the surface. In some other parts of the body, the structure of the capillary bed is not very regular, but still exhibits certain locations and individual characteristics. Similarly, the topography of the contact surface between the various layers of the skin is very complex and exhibits skin location and individual characteristics. Although these sources of skin and underlying subsurface structures represent a significant source of noise in non-imaging optical measurements of skin for biometric judgment or analyte measurements, structural differences are advantageously compared throughout the embodiments of the present invention. It is manifested by the spatial spectral features that can be done.

  In some cases, inks, dyes, and / or other pigments may be present on the skin as a topical application or subsurface tattoo. These forms of artificial pigments may or may not be visible to the naked human eye. However, if one or more wavelengths used by the apparatus of the present invention are sensitive to dyes, in some embodiments the sensor may be present in addition to other desired measurement tasks, the amount of dye, and / or Or it can be used to verify the shape.

  In general, embodiments of the present invention provide a method and system for collecting spatial spectral information that can be represented in a multi-dimensional data structure having independent spatial and spectral dimensions. In some cases, the desired information is included in only a portion of the overall multidimensional data structure. For example, estimation of a uniformly distributed spectrally active compound may require only measured spectral characteristics that can be extracted from the entire multidimensional data structure. In such cases, the overall system design can be simplified to reduce or eliminate the spatial component of the collected data by reducing the number of image pixels to the limit of a single pixel. Thus, although the disclosed systems and methods are generally described in the context of spatial spectral imaging, the present invention encompasses similar measurements in which the degree of image is greatly reduced to a single sensing element. It will be recognized.

(2. Non-contact biometric sensor)
One embodiment of the invention is depicted by the schematic diagram of FIG. 1 showing a front view of a non-contact biometric sensor 101. The sensor 101 comprises an illumination subsystem 121 having one or more light sources 103 and a detection subsystem 123 comprising an image device 115. Although the figure depicts an embodiment where the illumination subsystem 121 comprises a plurality of illumination subsystems 121a and 121b, the invention is not limited by the number of illumination or detection subsystems 121 or 123. For example, the number of lighting subsystems 121 can be conveniently selected to achieve a certain level of lighting, to meet package requirements, and to meet other structural constraints of the sensor 101. The illumination light passes through the illumination optics 105 that shapes the illumination into a desired form, such as the form of flood light, light rays, light spots, and the like from the light source 103. The illumination optics 105 is shown as consisting of lenses for convenience, but more generally may include any combination of one or more lenses, one or more mirrors, and / or other optical elements. it can. The illumination optical system 105 may also include a scanner structure (not shown) that scans illumination light with a designated one-dimensional or two-dimensional pattern. In various embodiments, the light source 103 can comprise a point light source, a line light source, a region light source, or can comprise a series of such light sources. In one embodiment, the illumination light is provided as polarized light, such as by placing a linear polarizer 107 through which the light passes before striking the finger 119 or other skin site of the individual under consideration. Since the skin site to be imaged can be positioned to interact with light without contacting any solid surface, the embodiment as shown in FIG. 1 is referred to herein as a “non-contact” sensor. . In the “contact” biometric sensor described in detail below, the skin site to be imaged is in contact with some solid surface, such as a platen or photodetector.

  In some cases, in various embodiments, the light source 103 comprises a white light source that can be provided as a broadband source or as a collection of narrowband emitters. Examples of broadband sources include white light emitting diodes (“LEDs”), incandescent bulbs or glowbars, and the like. The collection of narrowband emitters can comprise a quasi-monochromatic light source having a primary color wavelength, such as embodiments that include a red LED or laser diode, a green LED or laser diode, and a blue LED or laser diode.

  An alternative mechanism for reducing direct reflected light utilizes an optical polarizer. Both linear and circular polarizers can be advantageously employed to make optical measurements that are more sensitive to a certain skin depth, as is known to those skilled in the art. In the embodiment illustrated in FIG. 1, the illumination light is polarized by a linear polarizer 107. The detection subsystem 123 may also include a linear polarizer 111 disposed with its optical axis being substantially perpendicular to the illumination polarizer 107. Thus, the light from the sample must undergo a plurality of scattering phenomena to significantly change its polarization state. Such phenomenon occurs when light penetrates the surface of the skin and is again scattered by the detection subsystem 123 after many scattering events.

  Conversely, the use of two polarizers 107 and 111 can also be used to increase the effect of directly reflected light by placing the polarizer 111 substantially parallel to the polarizer 107. it can. In some systems, two or more polarizations in a single device to allow collection of multispectral data collected under two different polarization conditions (ie, under cross-polarization conditions and parallel polarization conditions). It may be advantageous to combine the configurations. In other embodiments, either or both polarizers 107 or 111 can be omitted, allowing collection of substantially randomly polarized light.

  The detection subsystem 123 can incorporate detection optics comprising lenses, mirrors, phase plates, and wavefront coding devices, and / or other optical elements that form images on the detector 115. In addition, the detection optical system 113 can incorporate a scan structure (not shown) to sequentially transmit the entire image portion to the detector 115. In all cases, the detection subsystem 123 is configured to be sensitive to light scattered through the surface of the skin and subjected to light scattering in the skin and / or underlying tissue before being emitted from the skin.

  In embodiments where white light is used, the detector 115 may comprise a Bayer color filter array in which filter elements corresponding to a set of primary colors are arranged in a Bayer pattern. An example of such a pattern is shown in FIG. 2A for an arrangement using red 204, green 212, and blue 208 color filter elements. In some cases, the detector subsystem 123 can further comprise an infrared filter 114 arranged to mitigate detected infrared light. As can be seen from the color response curve for the exemplary Bayer filter array shown in FIG. 2B, there is generally an overlap within the spectral range of the red 224, green 232, and blue 228 transmission characteristics of the filter elements. In particular, as evident in the curves for the transmission characteristics of green 232 and blue 228, the filter array may allow transmission of infrared light. This is avoided by including an infrared filter 114 as part of the detector subsystem. In other embodiments, the infrared filter 114 may be omitted, and one or more light sources 103 that emit infrared light may be incorporated. In this way, all color filter elements 204, 208 and 212 allow light to pass substantially, resulting in an infrared image across the detector 115.

  Another embodiment of a non-contact biometric sensor is schematically illustrated by the front view of FIG. In this embodiment, the biometric sensor 301 includes an illumination subsystem 323 and a detection subsystem 325. Similar to the embodiment described in connection with FIG. 1, in some embodiments, there may be multiple lighting subsystems 323, and FIG. 3 is a specific embodiment having two lighting subsystems 323. Indicates. The white light source 303 provided by the illumination subsystem 323 can be any light source of white light, including broadband or a combination of the above narrow band sources. The light from the white light source 303 passes through the illumination optical system 305 and the linear polarizer 307 before entering the skin site 119. A portion of the light is diffusely reflected from the skin site 119 into the detection subsystem 325 comprising image optics 315 and 319, linear polarizer 311, and dispersive optical element 313. The dispersive element 313 can be a transmissive or reflective, prism, or other optical component known in the art that causes a bias in the path of light as a function of light wavelength, one or two dimensional. Can be provided. In the illustrated embodiment, the first image optical system 319 serves to collimate the light reflected from the skin site 119 for transmission through the linear polarizer 311 and the dispersive element 313. The spectral components of the light are angularly separated by the dispersive element 313 and separately focused on the detector by the second image optical system 315. As discussed in connection with FIG. 1, the polarizer 307 provided by the illumination and detection subsystems 323, 325 when the optical axes of the polarizers 307 and 311 are oriented substantially perpendicular to each other. Each of 311 serves to reduce detection of direct reflected light in the detector 317. The polarizers 307, 311 can also be oriented so that their optical axes are substantially parallel, which increases the detection of direct reflected light at the detector 317. In some embodiments, either polarizer 307 or 311 or both may be omitted.

  Thus, the image generated from the light received at the detector is an “encoded” image in a computed tomography image spectrometer (“CTIS”) method. Both spectral and spatial information are present simultaneously in the resulting image. Individual spectral patterns can be obtained by mathematical inversion or “reconstruction” of the encoded image.

  The depiction of the non-contact sensor of FIG. 1 described that a scanner structure can be provided to scan the illumination light. This is an example of a more general class of embodiments where there is relative motion between the illuminated area and the skin site. In such embodiments, the images can be constructed by building separate image portions that are collected during relative motion. Such relative motion may also be achieved in embodiments that constitute a card-ready sensor where the user is instructed to translate the skin site. An example of a card reading sensor is shown in top view by the schematic diagram of FIG. In this figure, the illumination area and the detection area 405 of the sensor 401 are substantially collinear. In some embodiments of the card reading sensor 401, there may be one or more illuminated areas. For example, there may be a plurality of illumination areas arranged on both sides of the detection area 405. In some embodiments, the illumination area 403 may partially or completely cover the detection area. Image data is collected by the sensor by translating a finger or other body part through the optically active area, as indicated by the arrows in FIG. In some implementations, the example can be used with a contact-type structure described in detail below, but the card-reading sensor can be implemented with any of the non-contact type sensor structures described above. Light received sequentially from separate parts of the skin site is used to construct images that are subsequently used in biometric applications.

  The above embodiments create a large number of spatial spectral data that can be used in biometric applications as described below. The present invention is not limited to any particular method of storing or analyzing the multiple spatial spectral data. For illustrative purposes, it is shown in the form of a data cube in FIG. The data cube 501 is shown broken down along a spectral dimension comprising a plurality of planes 503, 505, 507, 509, 511, each corresponding to a different part of the light spectrum and each containing spatial information. ing. In some cases, the multiple spatial spectral data may include additional types of information beyond spatial and spectral information. For example, different illumination conditions as defined by different illumination structures, different polarizations, etc. can provide additional dimensional information. More broadly, data collected under multiple engineering conditions is referred to herein as “multispectral” data, whether collected simultaneously or sequentially. A more complete description of aspects of multispectral data can be found in US patent application Ser. No. 11 / 379,945, filed Apr. 24, 2006, entitled “MULTISPECTRAL BIOMETRIC SENSORS” by co-pending same applicant. The entire disclosure of which is described and incorporated herein by reference for all purposes. Thus, the spatial spectrum data can be considered as a part of a certain type of multispectral data including different illumination wavelengths as different optical conditions.

In embodiments where illumination occurs under white light, the images 503, 505, 507, 509, and 511 correspond to images generated using light at, for example, 450 nm, 500 nm, 550 nm, 600 nm, and 650 nm. There is a case. In another example, there may be three images corresponding to the amount of light in the red, green, and blue spectral bands at each pixel location. Each imaging device represents the optical effect of light of a specific wavelength that interacts with the skin. Each of the multispectral images 503, 505, 507, 509, and 511 is generally different from the others due to the optical properties of the skin and skin components that vary with wavelength. Therefore, the data cube can be represented as R (X _s , Y _s , X _I , Y _I , λ), and when illuminated at the light source points X _s , Y _s , the data cube is seen at each image point X _I , Y _I. This represents the amount of diffusely reflected light having a wavelength λ. Various illumination configurations (floods, lines, etc.) can be summarized by summing the point responses over the appropriate light source point locations. A conventional non-TIR fingerprint image F (X _I , Y _I ) can be roughly represented as the multiple spectrum data cube for a given wavelength λ _{o and} can be summed over all light source positions.

逆に、スペクトルバイオメトリックデータセットＳ（λ）は、所与の波長λに対して測定された光の強度を、照明位置と検出位置との間の差

Conversely, the spectral biometric data set S (λ) represents the intensity of light measured for a given wavelength λ between the illumination position and the detection position.

と関連させる：

Relate with:

よって、前記データキューブＲは、従来の指紋画像およびスペクトルバイオメトリックデータセットの両方に関連付けられる。前記データキューブＲは、他の２つのデータセットのいずれもの上位集合であり、別々の２つのいずれにおいても失われ得る相関およびその他の情報を含む。

Thus, the data cube R is associated with both a conventional fingerprint image and a spectral biometric data set. The data cube R is a superset of any of the other two data sets and contains correlations and other information that can be lost in any of the two separate data sets.

  Light passing through the skin and / or underlying tissue is generally affected by various optical properties of the skin and / or underlying tissue at various wavelengths. Two optical effects in the skin and / or underlying tissue that are affected differently at different wavelengths are scattering and light absorption. Light scattering in the skin is generally a function wavelength that is smooth and relatively slowly changing. Conversely, light absorption in the skin is generally a function of strong wavelength due to the specific absorption characteristics of certain components present in the skin. For example, blood, melanin, water, carotene, bilirubin, ethanol and glucose all have fairly large light absorption characteristics within the 400 nm to 2.5 μm spectral region that can sometimes be included in a white light source.

  Due to the combined effect of optical absorption and scattering, different illumination wavelengths penetrate the skin to different depths. This effectively results in different spectral images having different complementary information corresponding to different amounts of illuminated tissue. In particular, the capillary layer close to the surface of the skin has different spatial properties, which can be imaged at wavelengths where blood is strongly absorbed. Due to the complex wavelength dependent properties of skin and underlying tissue, a set of spectral values corresponding to a given image location has distinct and different spatial properties. These spatial characteristics can be used to classify images collected on a pixel-by-pixel basis. This assessment can be done by generating a typical tissue spectral quality from a set of eligible images. For example, the spatial spectral data shown in FIG. 5 can be reordered as an N × 5 matrix, where N is the number of image pixels containing data from living tissue rather than from surrounding air. . Eigenvalue analysis or other factor analysis based on this set of matrices yields representative spectral features of these tissue pixels. The spectra of the pixels in subsequent data can then be compared to such previously established spectral features using numerical indicators such as Mahalanobis distance and spectral remainder. If not a few image pixels have spectral qualities that do not match the biological tissue, the sample is considered unauthentic and rejected, thus incorporating an anti-spoofing method in the sensor based on the viability determination of the sample Provide mechanism.

  Alternatively, skin texture characteristics can be used alone or in combination with spectral characteristics to determine the authenticity of the sample. For example, each spectral image can be analyzed in such a way that it can represent the magnitude of various spatial characteristics. Such methods include wavelet transform, Fourier transform, cosine transform, density co-occurrence and the like. The resulting coefficients from any such transformation represent the texture aspect of the image from which they are extracted. Thus, a set of such coefficients extracted from a set of spectral images results in an explanation of the color texture characteristics of the multispectral data. These characteristics can then be compared to similar characteristics of known samples to make biometric decisions such as camouflaging or viability decisions. The method for making such a determination is generally similar to the method described for spatial characteristics above. Applicable classification techniques for such decisions include linear and quadratic discriminant analysis, classification trees, neural networks, and other methods known to those skilled in the art.

  Similarly, in embodiments where the sample is a palm surface of a hand or finger, image pixels are classified as “ridges”, “grooves”, or “others” based on their spectral quality or their color texture. be able to. This classification can be performed using linear discriminant analysis, secondary discriminant analysis, principal component analysis, neural network, and other discriminant analysis methods known to those skilled in the art. Since the ridge and groove pixels are adjacent on a typical palm surface, in some cases, data from a local neighborhood around the image pixel of interest is used to classify the image pixels. In this way, conventional fingerprint images can be extracted for further processing and biometric evaluation. The “other” category may indicate image pixels that have a spectral quality different from that expected in a real sample. A threshold can be set for the total number of pixels in the image classified as “other”. If this threshold value is exceeded, the sample is judged to be unauthentic and an appropriate instruction can be given to take action.

  In a similar manner, multispectral data collected from regions such as the palm of the finger is defined as the “feature point” defined as the location where the bulge ends, branches, or undergoes other such topological changes. Can be analyzed to directly estimate the position of. For example, the color texture quality of the multispectral data set can be determined by the above method. These qualities can then be used to classify each image location as “end of bulge”, “lift divergence”, or “other” in the manner described above. In this way, feature point feature extraction is performed without the need for computationally tedious calculations such as image normalization, image binarization, image thinning, and feature point filtering, and techniques known to those skilled in the art. Can be achieved directly from the data.

  The biometric determination of identity can be made using the entire large number of spatial spectral data or using specific portions thereof. For example, an appropriate spatial filter can be applied to deposit a lower spatial frequency that is typical of deeper spectrally activated structures in the tissue. Fingerprint data can be extracted using similar spatial frequency separation and / or pixel classification methods as described above. Spectral information can be separated from the active portion of the image in the manner described above. These three portions of the multiple spatial spectral data can then be processed and compared with corresponding registration data using methods known to those skilled in the art to determine the degree of fit. Based on the strength of the fit of these characteristics, it is possible to make a judgment about the alignment of the sample with the registered data. Additional details about certain types of spatial spectral analysis that can be performed are described in Robert K. et al. Provided in US patent application Ser. No. 10 / 818,698, entitled “MULTISPECTAL BIOMETRIC SENSOR” filed April 5, 2004 by Rowe et al., The entire disclosure of which is hereby incorporated by reference for all purposes. Incorporated into.

  As previously noted, certain substances that may be present in the skin and underlying tissue have different light absorption characteristics. For example, ethanol has characteristic absorption peaks of about 2.26 μm, 2.30 μm, and 2.35 μm, and spectral troughs of 2.23 μm, 2.28 μm, 2.32 μm, and 2.38 μm. In some embodiments, non-invasive optical measurements are made at wavelengths in the range of 2.1 to 2.5 μm, more specifically in the range of 2.2 to 2.4 μm. In embodiments that include at least one peak wavelength and trough wavelength, the resulting spectral data provides an estimate of the alcohol concentration in the tissue as well as a biometric characteristic of the individual being examined. Analyzed using least squares, principal component regression, and other multivariate analysis methods known to those skilled in the art. A correlation to blood alcohol level can be generated by values determined using a subset of these wavelengths, but it is preferable to examine at least three spectral peak values, and seven spectral peaks and trough values are measured. If so, more accurate results are obtained.

  In other embodiments, non-invasive optical measurements are made at a wavelength in the range of 1.5 to 1.9 μm, more specifically in the range of 1.6 to 1.8 μm. In a specific embodiment, the optical measurements are made at one or more of the wavelengths of about 1.67 μm, 1.69 μm, 1.71 μm, 1.73 μm, 1.74 μm, 1.76 μm and 1.78 μm. . The presence of alcohol is characterized at these wavelengths by spectral peaks of 1.69 μm, 1.73 μm, and 1.76 μm, and by spectral troughs of 1.67 μm, 1.71 μm, 1.74 μm, and 1.78 μm. . Similar to the 2.1-2.5 μm wavelength band, alcohol concentration is characterized by the relative intensity of one or more of the spectral peaks and trough values. Also, a correlation to blood alcohol level can be generated with values determined using a subset of these wavelengths in the range of 1.5-1.9 μm, but at least three spectral peak values are examined. Preferably, more accurate results are obtained when seven spectral peaks and trough values are measured.

  In certain embodiments, a small spectrum alcohol monitoring device may be incorporated into various systems and applications. The spectral alcohol monitoring device can be provided to police personnel or for personal use by individuals such as electronic fobs, watches, cell phones, PDAs, or any other electronic device It can be configured as a dedicated system that can be integrated as part. Such a device can include a mechanism for indicating whether an individual's blood alcohol level is within defined limits. For example, the device can include red and green LEDs, and the electronics in the device illuminate the green LED if the individual's blood alcohol level is within a defined limit, otherwise the red LED Illuminate. In one embodiment, the alcohol monitoring device is typically positioned so that an individual can conveniently place tissue, such as a fingertip, on the device and can be mounted in a vehicle. In some cases, the device can only serve as an informational guideline for driving suitability, but in other cases the positive that an individual has a blood alcohol level below a prescribed level. Based on the determination, engine ignition of the automobile may be performed.

(3. Contact type biometric sensor)
The biometric sensor can be configured in a manner similar to that shown in FIGS. 1 and 3, but the skin site is configured to be placed in contact with the platen. Such a design has certain additional characteristics due to light interaction with the platen, sometimes allowing additional information to be incorporated as part of the aggregate spatial spectral data. .

  One embodiment is shown in FIG. 6, which provides a front view of contact biometric sensor 601. Like the sensor illustrated in FIG. 1, the contact sensor 601 has one or more illumination subsystems 621 and a detection subsystem 623. Each of the illumination subsystems 621 includes one or more white light sources 603 and illumination optics that shape the light provided by the light sources 603 into a desired shape. Similar to the non-contact arrangement, the illumination optics generally can include any combination of optical elements and sometimes can include a scanner structure. In some cases, the illumination light is provided as polarized light by placing a polarizer 607 through which the illumination light passes. Examples of white light sources 603 including broadband and narrow band sources are described above, and the light sources 603 can be configured to provide light sources having various shapes in various embodiments.

  The illumination light is directed by the illumination optical system 621 and passes through the platen 617 to illuminate the skin site 119. The sensor arrangement 601 and components can be advantageously selected to minimize direct reflection of the illumination optics 621. In one embodiment, such direct reflection is mitigated by relatively orienting the illumination subsystem 621 and detection subsystem 623 such that the amount of direct reflected light is minimized. For example, the optical axes of the illumination subsystem 621 and the detection subsystem 623 are installed at an angle such that a mirror installed on the platen 617 does not direct a considerable amount of illumination light into the detection subsystem 623. Can do. Also, the optical axes of the illumination and detection subsystems 621 and 623 can be placed at an angle relative to the platen 617 such that the acceptance angle of both subsystems is less than the critical angle of the system 601, such as The structure avoids significant effects due to the total internal reflectivity between the platen 617 and the skin site 119.

  The presence of the platen 617 does not detrimentally interfere with the ability to reduce direct reflected light through the use of polarizers. The detection subsystem 623 can include a polarizer 611 having an optical axis that is substantially perpendicular or parallel to the polarizer 607 provided by the illumination subsystem 621. Since light from the sample undergoes substantially many scattering phenomena and must change its polarization state before being sensed by the detector 615, the polarizers 611 and 607 are substantially relative to each other. Surface orientation at the contact surface between the platen 617 and the skin site 119 is reduced. The detection subsystem 623 can further incorporate detection optics that form an image of a region in the vicinity of the platen surface 617 on the detector 615. In one embodiment, the detection optical system 613 includes a scanning structure (not shown), and sequentially transmits the platen region portion onto the detector 615. In particular, in embodiments where the detector 615 is sensitive to infrared light, such as when a Bayer filter array is used, an infrared filter 614 may be included to reduce the amount of infrared light detected. Conversely, as described above, in some embodiments, the infrared filter 614 can be omitted, and in some embodiments, an additional light source 603 with infrared emission can be included.

  As with the other arrangements described above, the detection subsystem 623 is generally configured to be sensitive to light that penetrates the surface of the skin and is subject to light scattering within the skin and / or underlying tissue. The polarizer can sometimes be used to generate or enhance surface features. For example, if the illumination light is polarized in a direction parallel to the platen 617 (“P”), the detection subsystem 623 incorporates a polarizer 611 in the vertical direction (“S”) and the reflected light is The light is blocked by the extinction ratio of the pair of polarizers. However, light across the skin site at the ridge point is optically scattered, which effectively randomizes the polarization (but the skin itself has several characteristic features as known to those skilled in the art. (With polarization quality). This allows about 50% of the partially absorbed and re-emitted light to be observed by the S-polarized imaging system.

  A side view of one embodiment of the present invention is shown by the schematic provided in FIG. 7A. For clarity, this figure does not show the detection subsystem, but clearly shows the illumination subsystem 621. The illumination subsystem 621 in this embodiment has a plurality of white light sources 703 that are spatially distributed. As shown, the illumination optics 621 is configured to provide flood illumination, but in other embodiments, cylindrical optics, focusing optics, or other optical components known to those skilled in the art. By incorporating it, it can be arranged to provide a line, point, or other pattern of illumination.

  The arrangement of the white light sources 703 in FIG. 7A need not actually be planar as shown in the figure. For example, in other embodiments, an optical filter, fiber bundle, or fiber optic faceplate or taper transmits light from the light source at a convenient location to the illumination plane, and the light is re-imaged onto the skin site 119. The The light source can be controlled with the drive current on and off, as in the case of LEDs. Alternatively, if an incandescent light source is used, the step of turning off the light can be controlled using some form of spatial light modulator, such as a liquid crystal modulator, or an aperture, mirror, or other such optical element Can be achieved using microelectromechanical system ("MEMS") technology. Such a structure can allow the configuration of the sensor to be simplified. One embodiment is illustrated in FIG. 7B illustrating the use of electronic scanning of illumination sources such as optical filters and LEDs. An individual fiber 716a connects each of the LEDs located in the illumination array 710 to the imaging surface, and the other fiber 716b leads to an imaging device 712 that can comprise a photodiode array, CMOS array, or CCD array. Retransmit the reflected light. Thus, the set of fibers 716a and 716b define an optical filter bundle 714 that is used in transmitting light.

  Another embodiment of a contact biometric sensor is schematically illustrated by the front view of FIG. In this embodiment, the biometric sensor 801 includes one or more white light illumination subsystems 823 and a detection subsystem 825. The illumination subsystem 823 includes a white light source 803 that provides light passing through the illumination optics 805, and a polarizer 807 that is oriented on a platen 817 over which the skin site 119 is disposed. A portion of the light is diffusely reflected from the skin site 119 into a detection subsystem 825 that includes imaging optics 815 and 819, crossed polarizers 811, and dispersive optics 813. The first image optical system 819 functions to make the light reflected from the skin site 119 parallel to the transmission through the crossed polarizer 811 and the dispersion element 813. The separated spectral components are separately focused onto the detector 817 by the second image optics 815.

  Moreover, the contact-type biometric sensor as illustrated in FIGS. 6 to 8 is easily affected by a configuration in which the illumination region is in a relative motion state with the skin site. As previously described, such relative motion may be implemented by a mechanism for scanning the illumination light and / or by moving the skin site. The presence of the platen in the contact sensor embodiment generally facilitates movement of the skin site by limiting the surface of the skin site to a defined plane, and the freedom of movement is allowed in three dimensions. In form, additional difficulties may be due to the movement of the skin site outward in the imaging depth direction. Thus, the card-reading sensor can be implemented with a contact biometric sensor in the manner generally described above in connection with FIG. 4 but includes a platen that prevents movement of the skin site in one direction. . In some embodiments, the card-type reading sensor can be a fixed system, but the contact-type structure is implemented with a roller system in which the skin site is rolled over a roller structure that transmits white light. Make it possible. As will be appreciated by those skilled in the art, the encoder can help record the location information and synthesize a complete two-dimensional image from the resulting series of image pieces. Light received from a separate part of the skin site is used to construct an image.

  While the above description of non-contact and contact biometric sensors focuses on embodiments where white light is used, other embodiments combine other spectral combinations of light with similar structural arrangements. Can be used. In addition, other embodiments can include additional changes in optical conditions to provide multispectral conditions. Some descriptions of such multispectral applications are found in the same applicant by Robert K. US patent application Ser. No. 10 / 818,698, entitled “MULTISPECTAL BIOMETRIC SENSOR”, filed April 5, 2004 by Rowe et al., RobertK. U.S. Patent No. 11 / 177,817 entitled "LIVENESS SENSOR", filed July 8, 2005 by Rowe, Robert K. et al. Rowe and Stephen P. US Provisional Patent No. 60 / 576,364 entitled "MULTISPECTRAL FINGER RECOGNITION" filed June 1, 2004 by Corcoran, Robert K. US Provisional Patent Application No. 60 / 600,867 entitled “MULTISPECTRAM IMAGE BIOMETRIC” filed August 11, 2004 by Rowe, Robert K. US Patent No. 11 / 115,100 entitled “MULTISPECTRAL IMAGEING BIOMETRICS” filed April 25, 2005 by Rowe, entitled “MULTISPECTRAL BIOMETRIC IMAGEING” filed April 25, 2005 U.S. Patent Application No. 11 / 115,101, U.S. Patent No. 11 / 115,075 entitled "MULTISPECTAL LIVENCE DETERMINATION" filed April 25, 2005, Robert K. US Provisional Patent Application No. 60 / 659,024 entitled "MULTISPECTRAL IMAGEING OF THE FINGER FOR BIOMETRICS" filed Mar. 4, 2005 by Robert et al., Robert K. US Provisional Patent Application No. 60 / 675,776 entitled “MULTISPECTAL BIOMETRIC SENSORS” filed April 27, 2005 by Robert, Robert K. No. 11 / 379,945, entitled “MULTISPECTAL BIOMETRIC SENSORS” filed April 24, 2006 by Rowe. The entire disclosure of each of the prior applications is incorporated herein by reference for all purposes.

  In one embodiment, the non-contact and contact biometric sensors described above use white light imaging. The use of white light allows images to be collected in multiple colors simultaneously, and the overall speed of data collection is faster than embodiments where separate states are collected separately. This reduction in data collection time causes a reduction in motion artifacts as the skin site moves during data collection. By using a smaller number of light sources compared to using separate illumination sources for different colors, the overall sensor size can also be reduced and provided at a lower cost. Corresponding mitigation is also possible in electronic equipment used to support cooperative operation of the light source. Also, current color image devices are typically available at lower prices than single color image devices.

  The use of white light imaging also allows a reduction in the amount of data when the sensor is designed to use all pixels to achieve the desired resolution. For example, a typical design standard can provide a 1 inch field with a resolution of 500 dots per inch. This can be achieved with a monochromatic camera having 500 × 500 pixels. It can also be achieved with a 1000 × 1000 color camera when each color plane is extracted separately. The same resolution can be achieved by converting to {R, G, B} triplet using a 500 × 500 color imager and then extracting the monochromatic part of the image. This is a specific example of a more general procedure that uses a color imaging device by converting to the three primary colors, followed by extraction of a single color portion of the image. Such a procedure generally allows the desired resolution to be achieved more efficiently than other extraction techniques.

(4. Integrated multispectral / TIR sensor)
In some embodiments, the multispectral imaging enabled by the sensor can be integrated with a conventional TIR imaging system. One illustration of such integration is provided in FIG. 10, which shows a bright field fingerprint sensor integrated with multispectral analysis capabilities. The sensor includes a light source 953, a platen 961, and an imaging system 955. The imaging system 965 can include lenses, mirrors, optical filters, digital image arrays, and other such optical elements (not shown). The optical axis of the imaging system 955 is at an angle greater than the critical angle relative to the surface 961a. Light from the light source 953 passes through the platen 961 and strikes the diffuse reflective coating 963 that illuminates the platen surface 961a extensively. In the absence of the finger 969, light undergoes TIR at the surface 961a and a portion is collected to form a uniformly bright TIR image by the imaging system 955. When the appropriate refractive index skin is in optical contact with the platen surface 961a, the contact points form a relatively dark area on the resulting image. Variations of such a configuration using different light source 953 positions and / or imaging system 955 can achieve dark field TIR images. For example, when the platen 961 is illuminated by the light source 957 and imaged by the imaging system 955, a dark field image with relatively bright contact points is created. In some dark field embodiments, the coating 963 can comprise an optical absorber.

  The second imaging system 959 looks up the finger 969 through the facet 961b. The imaging system 959 has an optical axis that is less than the critical angle relative to the facet 961a. In some embodiments, the imaging system 959 is oriented substantially perpendicular to the facet 961a. The imaging system can include a lens, a mirror, an optical filter, and a digital image array (not shown). Thus, when the light source 953 is illuminated, a TIR image can be captured by the camera 955, while a direct image can be captured by the camera 959. Even if the TIR image is adversely affected by water, dirt, lack of contact, dry skin, etc., the image captured by the camera 959 is relatively unaffected and is generally useful bios that contain fingerprint patterns. Includes metric features.

  The imaging system 959 can further incorporate an optical polarizer (not shown) that can be a linear polarizer or an elliptical (eg, circular) polarizer. Similarly, other light sources 957 may be added to the system. The light source 957 may be a quartz tungsten halogen lamp or other incandescent light source commonly known in the art. The light source 957 may be a white light LED or other broadband source known in the art. The light source may be a simple color source, such as a semiconductor LED, organic LED, laser diode, or other type of laser or simple color source known in the art. The light source 957 can further include a lens, a mirror, an optical diffuser, an optical filter, and other such optical elements.

  The light sources 957 can be substantially the same or can provide various illumination wavelengths, angles, and / or polarization conditions. In the latter case, one of the light sources 957a may have an optical polarizer (not shown) that is oriented substantially perpendicular to the polarizer incorporated in the imaging system 959. Such optical arrangements tend to emphasize the features of the skin below the surface. One of the light sources 957b can incorporate a polarizer that is substantially parallel to the polarizer used in the imaging system 959 and tends to enhance the surface characteristics of the skin. Alternatively, one of the light sources 957b may omit the polarizer and produce random polarization. Such random polarization can be described as a combination of polarizations that are parallel and perpendicular to the polarizer in the imaging system 959, creating an image that combines features from two polarization conditions. The light sources 957 may be the same wavelength or various wavelengths (with or without a polarizer). The number and arrangement of light sources 957 may be different for various embodiments to adjust for form factor constraints, lighting level constraints, and other product requirements.

  In one embodiment, the light source 957 is oriented at an angle that is less than the critical angle relative to the surface 961a. In a preferred embodiment, the light source can be located at an angle and position such that direct reflection of the light source is not seen by the imaging system 959 or 955. Such direct reflections can also be greatly mitigated through the use of crossed polarizer configurations, but generally some image artifacts still exist when the light source is in the field of view. Furthermore, parallel polarized and unpolarized configurations are very susceptible to such back reflections.

  Attempts to impersonate such systems generally include (1) the use of a thin transparent film covering the finger or other tissue, (2) the use of a thin scattering film covering the finger or other tissue, and (3) the finger. Or it can be characterized according to three different categories: the use of thick samples covering other tissues. Embodiments of the present invention are more practical in the second and third examples, i.e. when the camouflage comprises a thick or thin scattering film. If the camouflage comprises a thin transparent film, ie is substantially non-absorbing and non-scattering, attempts to incorporate different fingerprint pattern topologies can be detected by comparing the traditional pattern with the TIR pattern. A more detailed description of how to make such comparisons is provided in US patent application Ser. No. 11 / 015,732, which is incorporated herein by reference.

  Also, camouflage using a thin scattering film, ie, a substantially non-absorbing film with a significant amount of light scattering, can incorporate different fingerprint pattern topologies in an attempt to deceive conventional sensors. The spectral imaging described in this application allows imaging of the underlying tissue characteristics even when the real fingerprint image is ambiguous. Conversely, a thick sample provides a combination of the sample's scattering and extinction characteristics that ensure that the detected light, if any, does not penetrate significantly into adjacent tissue. Attempts to incorporate different fingerprint pattern topologies to deceive conventional sensors detect according to embodiments of the present invention by evaluating the spectrum and other optical properties of thick samples and comparing them to real tissue be able to.

(5. Texture biometric sensor)
Another form of contact biometric sensor provided in embodiments of the present invention is a texture biometric sensor. “Image texture” generally refers to any of a number of numerical indicators that describe certain aspects of the spatial distribution of tonal characteristics of an image, some of which have been described above. For example, some textures commonly found in fingerprint patterns or wood grain are fluent and can be well represented by numerical indicators such as directionality and coherence. For textures with spatial regularity (at least locally), certain features of the Fourier transform and associated power spectrum, such as energy density, dominant frequency and direction, are important. Certain statistical moments such as mean, variance, asymmetry, and kurtosis can be used to represent the texture. Moment invariants, which are combinations of various moments that are invariant to scale, rotation, and other perturbation changes, can be used. A gray-tone spatial dependence matrix can be generated and analyzed to represent the image texture. The entropy on the image area can be calculated as a texture of image texture. Various types of wavelet transforms can be used to represent image texture aspects. Image textures can be represented using steerable pyramids, Gabor filters, and other mechanisms that use spatially bounded basis functions. In embodiments of the invention, these and other such quality sensitivities known to those skilled in the art can be used individually or in combination.

  Thus, image texture can be manifested by changes in pixel intensity across the image that can be used in embodiments of the present invention that implement biometric functions. In some embodiments, when such texture analysis is performed on different spectral images extracted from a multi-spectral data set, additional information can be extracted to create a color texture description of the skin site. These embodiments advantageously allow biometric functions to be performed by capturing a portion of the image of the skin site. The texture characteristics of the skin site are expected to be nearly consistent across the skin site, allowing biometric functions to be performed with measurements made at various parts of the image site. In many cases it is not even required that the parts of the skin site used in the various measurements overlap one another.

  This ability to use various parts of the skin site provides considerable flexibility in the structural design that can be used. This is also a result of the fact that instead of requiring a fit to a deterministic spatial pattern, a biometric fit can be made statistically. The sensor does not need to acquire an image over a specified spatial region and can be configured in a compact manner. The ability to provide a small sensor also makes it possible to make the sensor more economically than a sensor that needs to collect complete spatial information and perform a biometric function. In different embodiments, biometric functions can be implemented with pure spatial information, while in other embodiments, spatial spectral data is used.

  An example of a configuration for a texture biometric sensor is schematically illustrated in FIG. 10A. The sensor 1000 includes a plurality of light sources 1004 and an image device 1008. In some embodiments, the light source 1004 comprises a white light source, while in other embodiments, the light source comprises a quasi-monochromatic light source. Similarly, the image device 1008 can comprise a single color or color image device, an example of which is an image device having a Bayer pattern. Because the image is collected substantially in the plane of the skin site 119 being measured, the sensor 1000 is referred to herein as a “contact” sensor. However, it is possible to have various structures for operating the sensor, such that the imaging device 1008 is substantially in contact with the skin site 119, or the imaging device 1008 is provided for the skin site 119. Some are far from the plane.

  This is shown for two exemplary embodiments in FIGS. 10B and 10C. In the embodiment of FIG. 10B, the imaging device 1008 is substantially in contact with the skin site 119. Light from the light source 1004 propagates under the tissue at the skin site 119 so that light scattered from the skin site 119 and within the underlying tissue can be detected by the imaging device 1008. Another embodiment in which the imaging device 1008 is remote from the skin site 119 is shown schematically in FIG. 10C. In this figure, the sensor 1000 ′ includes an optical arrangement 1012 that translates an image in the plane of the skin site 119 to the imaging device, and with total internal reflectance along the fiber without substantial loss of intensity. A plurality of optical filters may be provided that translate individual pixels of the image. Thus, the light pattern detected by the imaging device 1008 is substantially the same as the light pattern formed in the plane of the skin site 119. Thus, the sensor 1000 'operates in substantially the same manner as the sensor 1000 shown in FIG. 10B. That is, the light from the light source 1004 is propagated to the skin site where it penetrates the skin site 119 and is reflected and scattered by the underlying tissue. Since the information is only translated substantially without loss, the image formed by the imaging device 1008 in such an embodiment is substantially the same as the image formed in the arrangement as in FIG. 10A. Are the same.

  In an embodiment that implements a biometric function using pure spatial information, spectral characteristics in the received data are recognized and compared to a spectral registration database. The tissue spectrum that results from a particular individual includes a unique spectral feature and a combination of spectral features that can be used to identify the individual once the device is tuned to extract the relevant spectral features. Extracting relevant spectral features can be done by a number of different techniques including discriminant analysis techniques. While not readily apparent in the visual analysis of spectral output, such analysis techniques can iteratively extract unique features that can be discriminated to perform biometric functions. An example of a specific technique is U.S. Patent No. 6,560,352, "METHODS TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TIMES OF TECHNOS US Patent No. 6,816,605 entitled "INDIVIDUALS USING LINEAR OPTICAL SPECTROCOPY", "APPARATUS AND METHOD FOR IDENTIFICATION OF INDIVIDUALS BY NEAR-INFRARED SP" US Pat. No. 6,628,809 entitled “ECTROSCOPY”, Robert K. US Patent Application No. 10 / 660,884 entitled “APPARATUS AND METHOD FOR IDENTIFICATION OF INDIVIDUALS BY NEAR-INFRARED SPECTROCOPY” filed September 12, 2003 by Robert et al., And Robert K. US Pat. No. 09 / 874,740 entitled “APPARATUS AND METHOD OF BIOMETRIC DETERMINATION USING SPECIALIZED OPTICAL SPECTROCOPY SYSTEM” filed June 5, 2001 by Rowe et al. The entire disclosure of each of the prior patents and patent applications is incorporated herein by reference in its entirety.

  The ability to perform a biometric function with image texture information, including biometric identification, can take advantage of the fact that a substantial portion of the signal from the organism is due to capillary blood. For example, if the skin region 119 comprises a finger, the finger capillaries follow an external fingerprint raised structure pattern as a known physiological characteristic. Thus, the contrast of the fingerprint features to the illumination wavelength is related to the spectral features of the blood. In particular, the contrast of images taken at wavelengths longer than about 580 nm is significantly reduced compared to images taken at wavelengths less than about 580 nm. Fingerprint patterns produced by other optical effects such as non-blood pigments and Fresnel reflectivity have different spectral contrasts.

  In various embodiments, the light scattered from the skin site 119 can be subjected to a variety of different types of comparative texture analysis. Some embodiments utilize a form of moving window analysis of image data extracted from collected light to generate a figure of merit, thereby assessing quality sensitivity or figure of merit. In some embodiments, moving window operations can be replaced with block-by-block analysis or tiled analysis. In some embodiments, a single area of the image or the entire area of the image can be analyzed at once.

  In one embodiment, the fast Fourier transform is performed on one or more regions of the image data. The in-band contrast figure of merit C is generated in such embodiments as the average or the ratio of DC power to in-band power. Specifically, for an index i corresponding to one of a plurality of wavelengths provided by white light, the contrast performance index is

である。
この式では、Ｆ_ｉ（ξ，η）は、指数ｉに対応する波長における画像ｆ_ｉ（ｘ，ｙ）のフーリエ変換であり、ｘおよびｙは画像に対する空間的座標である。Ｒ_ｌｏｗおよびＲ_ｈｉｇｈによって画定される値域は、指紋特徴の対象の空間周波数の限定を表す。例えば、一実施形態において、Ｒ_ｌｏｗは、約１．５縞／ｍｍとすることができ、Ｒ_ｈｉｇｈは、３．０縞／ｍｍとすることができる。代替的な式では、コントラスト性能指数は、２つの異なる空間周波数帯域における積分パワーの比率として定義することができる。上記に示される方程式は、帯域の１つがＤＣ空間周波数のみを備える特定の場合である。

It is.
In this equation, F _i (ξ, η) is the Fourier transform of the image f _i (x, y) at the wavelength corresponding to the index i, and x and y are the spatial coordinates for the image. The range defined by R _low and R _high represents the spatial frequency limitation of the object of the fingerprint feature. For example, in one embodiment, R _low can be about 1.5 fringes / mm and R _high can be 3.0 fringes / mm. In an alternative formula, the contrast figure of merit can be defined as the ratio of integrated power in two different spatial frequency bands. The equation shown above is the specific case where one of the bands comprises only the DC spatial frequency.

In another embodiment, moving window averages and moving window standard deviations are calculated for a large number of collected data and used to generate a figure of merit. In this embodiment, for each wavelength corresponding to the index i, the moving window average value μ _I and the moving window standard deviation σ _I are calculated from the collected images f _i (x, y). The moving window for each calculation may be the same size and may be conveniently chosen to cover approximately 2 to 3 times the fingerprint ridge. Preferably, the size of the window is large enough to cut out fingerprint features, but small enough to maintain background variation. The figure of merit C _i in this embodiment is calculated as the ratio of the moving window standard deviation to the moving window average value.

なお、別の実施形態では、同様のプロセスが行われるが、移動窓範囲（つまり最大値（画像値）−最小値（画像値））が移動窓標準偏差の代わりに使用される。よって、前の実施形態と同様に、指数ｉに対応する各波長について、前記移動窓平均値μ_Ｉおよび前記移動窓標準偏差σ_Ｉは、収集された画像ｆ_ｉ（ｘ，ｙ）から計算される。前記移動窓平均値の計算に対する窓のサイズは再び、好ましくは、指紋特徴を切り出すには十分大きいが、背景変動を維持するには十分小さい。場合によっては、前記移動窓平均値の計算に対する窓のサイズは、前記移動窓範囲の計算に対するものと同じであり、一実施形態における適切な値は指紋隆起の約２乃至３倍に及ぶ。この実施形態における性能指数は、前記移動窓平均値の比率として計算される。

In another embodiment, a similar process is performed, but the moving window range (ie, maximum value (image value) −minimum value (image value)) is used instead of the moving window standard deviation. Thus, as in the previous embodiment, for each wavelength corresponding to the index i, the moving window average value μ _I and the moving window standard deviation σ _I are calculated from the collected images f _i (x, y). The The window size for the moving window average calculation is again preferably large enough to cut out fingerprint features, but small enough to maintain background variation. In some cases, the window size for the moving window average calculation is the same as for the moving window range calculation, and suitable values in one embodiment range from about 2 to 3 times the fingerprint ridge. The figure of merit in this embodiment is calculated as a ratio of the moving window average value.

この実施形態および先行のものは、移動窓計算が収集されたデータについて行われ、移動窓中心度および移動窓変動度を計算するより一般的な実施形態の特定の場合であると考えることができる。特定の実施形態は、中心度が非加重平均値を備えるが、さらに一般的に、ある実施形態において、加重平均値または中央値などの任意のその他の種類の統計的中心度を備えることができる場合を例示する。同様に、特定の実施形態は、変動度が標準偏差または範囲を備えるが、さらに一般的に、ある実施形態において中央値絶対偏差（ｍｅｄｉａｎ ａｂｓｏｌｕｔｅ ｄｅｖｉａｔｉｏｎ）または平均値の標準誤差などの任意のその他の種類の統計的変動度を備えることができる場合を例示する。

This embodiment and the previous one can be considered to be a specific case of a more general embodiment in which moving window calculations are performed on collected data and moving window centrality and moving window variability are calculated. . Certain embodiments have a centrality with an unweighted average, but more generally, in some embodiments, can have any other type of statistical centrality, such as a weighted average or median. The case is illustrated. Similarly, certain embodiments have a standard deviation or range of variability, but more generally, in some embodiments, any other such as median absolute deviation or standard error of the mean The case where the kind of statistical variability can be provided is illustrated.

  In another embodiment that does not use explicit moving window analysis, wavelet analysis can be performed for each of the spectral images. In some embodiments, wavelet analysis can be performed in such a way that the resulting coefficients are approximately spatially invariant. This can be achieved by performing an undecimated wavelet decomposition and applying a dual-tree complex wavelet method, or other such method. Gabor filters, variable pyramids, and other such decompositions can also be applied to produce similar coefficients. Whatever decomposition method is selected, the result is a set of coefficients proportional to the magnitude of the variation corresponding to a particular basis function at a particular location in the image. To perform fake detection, the wavelet coefficients or the aggregation derived from the coefficients can be compared to the coefficients expected for a real sample. By comparison, if the result shows a sufficient approximation, the sample is considered authentic. Otherwise, the sample is determined to be camouflaged. In a similar manner, the currently measured set of coefficients can also be used for the coefficient biometric matching by comparing with a previously recorded set from the same person.

(6. Exemplary use)
In various embodiments, a biometric sensor (whether non-contact or contact), or any texture sensor of the type described above, can be operated by a computing system to implement biometric functionality. it can. FIG. 11 schematically illustrates how individual system elements can be implemented in a separate or more integrated manner. The computing device 1100 is shown to comprise hardware elements that are electrically coupled via a bus 1126 that is also coupled to the biometric sensor 1156. The hardware elements include a processor 1102, an input device 1104, an output device 1106, a storage device 1108, a computer readable storage medium reader 1110a, a communication system 1114, a processing accelerator 1116 such as a DSP or special purpose processor, and a storage device 1118. Including. The computer readable storage medium reader 1110a is further connected to a computer readable storage medium 1110b, the combination of which can be used to temporarily and / or more computer readable information in addition to remote, local, fixed, and / or removable storage devices. A comprehensive representation of storage media for permanent inclusion. The communication system 1114 may comprise wired, wireless, modem, and / or other types of interface connections, allowing data to be exchanged with external devices.

  The computing device 1100 is also software that is shown to be currently located in the working storage device 1120, including an operating system 1124 and other code 1122 such as a program designed to perform the method of the present invention. The device is provided. It will be apparent to those skilled in the art that substantial changes may be made in accordance with the specific requirements. For example, customized hardware may also be used and / or certain elements may be implemented in hardware, software (including portable software such as applets) or both. Furthermore, connections to other computer devices such as network input / output devices may be employed.

  As an example of an embodiment of the present invention, the multispectral condition can comprise a plurality of different wavelengths of illumination light. Quality sensitivity is the degree of image contrast generated for a certain spatial frequency of an image. In the present application, “image contrast” refers to a degree of tone variation with respect to an average tone value over a certain area of an image. In the case of an uncoated human finger, a living body imaged using visible light in a manner similar to that described herein, a significant portion of the fingerprint signal is due to capillary blood. This is due in part to the known physiological property that finger capillaries follow a pattern of external fingerprint ridges. Thus, the contrast of the fingerprint features to the illumination wavelength can be expected to be related to the spectral features of the blood. In particular, the contrast of images taken at wavelengths longer than about 580 nm is significantly reduced compared to images taken at wavelengths less than about 580 nm. Fingerprint patterns produced by other optical effects such as non-blood pigments and Fresnel reflectivity have different spectral contrasts.

  Such an absorbance spectrum of oxygen rich blood over the wavelength range of about 425 nm to about 700 nm is shown in FIG. In the exemplary embodiment, in this case, five different wavelengths, 445 nm, 500 nm, 574 nm, 610 nm, and 660 nm, are used to define the multispectral conditions, but the present invention uses any specific wavelength for the multispectrum used or used. It is not limited to the number of conditions. Each of these wavelengths is identified in the spectrum of FIG. As can be seen, the absorbance values for blood are maximum for exemplary wavelengths of 445 nm and 574 nm, smaller for 500 nm, and much smaller for 610 nm and 660 nm. Thus, real finger images taken under illumination wavelengths of 445 nm and 574 nm can be expected to provide relatively high contrast, and images taken at wavelengths of 610 nm and 660 nm provide relatively low contrast. However, an image taken at 500 nm has an intermediate contrast. In addition, effects from other absorbers such as melanin, changes in scattering from wavelength to wavelength, and various chromatic aberrations also affect the degree of contrast and / or other quality sensitivities, but these effects are small or other It is either a characteristic of a living human finger against possible disguised substances or methods.

  Various embodiments have examined data sets using 20 normal finger images collected in 13 people and using 10 camouflaged images collected by thin silicone camouflage. An example of an image that can be collected is shown in FIG. 13, where the drawing parts (a) to (d) show the result of being a real finger, and the parts (e) to (h) show the result of impersonation. The “plane” identified for each of the images corresponds to a different wavelength: plane 1 is 445 nm, plane 2 is 500 nm, plane 3 is 574 nm, plane 4 is 610 nm, and plane 5 is 660 nm. The plane 6 can be understood with reference to FIG. 1B. Plane 6 corresponds to the image collected using TIR illumination 153 but imaged by imaging system 159. The wavelength of the TIR illumination used in this experiment was about 640 nm. Thus, image comparison can be made between a real finger and camouflaged pair, ie, parts (a) and (e), both showing the result of plane [3 2 1], in which case [3 The notation 2 1] means that plane 3 is positioned at the red value in the color graphic, plane 2 is positioned at the green value, and plane 1 is positioned at the blue value. Similarly, the comparison shows that parts (b) and (f) both show the result for plane [5 4 5], both parts (c) and (g) both show the result for plane [5 3 1], both It can be performed between the parts (d) and (h) showing the TIR illumination result. [3 2 1] images of parts (a) and (e) were taken with blue and green illumination, indicating that the fingerprint ridge is relatively high contrast for both real tissue and camouflage. This is [5 4 5] in parts (b) and (f), taken under red illumination, showing a relatively low contrast for real finger fingerprint ridges but a relatively high contrast for camouflaged fingerprints. It is the opposite of the image result.

  Utilizing such differences is accomplished by the method of the present invention summarized by the flow diagram of FIG. One or more quality sensitivities are developed for each optical condition, and the resulting set of quality sensitivities are compared to the predicted values from a real sample. In block 1404, a sample presumed to be a tissue sample presented to enable biometric functionality is illuminated under various optical conditions. Examples of the various optical conditions include factors such as various wavelengths as shown in block 1424, various polarization conditions as shown in block 1428, and various light interaction angles as shown in block 1432. Can be included. Light scattered from the sample is collected at block 1408 and used in comparative texture analysis. Various embodiments utilize different forms of moving window analysis of multispectral image stacks extracted from collected light to generate a figure of merit in block 1412, thereby evaluating quality sensitivity or figure of merit. In some embodiments, the moving window operation 1412 can be replaced with block-by-block analysis or tiled analysis. In some embodiments, a single region of the image or the entire image can be analyzed at once. Although the drawings show three examples of contrast figure of merit that can be used in various embodiments, other contrast figure of merit as well as other textures will be apparent to those skilled in the art after reading this disclosure. This particular identification is not intended to be limiting, as degrees can be used in other embodiments.

  In one embodiment, a fast Fourier transform is performed on the multispectral image stack, as shown at block 1436. The in-band contrast figure of merit C is generated in embodiments such as average or DC power to in-band power ratio. Specifically, for optical condition i, the contrast performance index is

である。
この式では、Ｆ_ｉ（ξ，η）は、照明条件ｉ下で撮影される画像ｆ_ｉ（ｘ，ｙ）のフーリエ変換であり、ｘおよびｙは、画像に対する空間的座標である。Ｒ_ｌｏｗおよびＲ_ｈｉｇｈによって定義される範囲は、指紋特徴の対象の空間周波数の限定を表す。例えば、一実施形態において、Ｒ_ｌｏｗは、約１．５縞／ｍｍとすることができ、Ｒ_ｈｉｇｈは、３．０縞／ｍｍとすることができる。多重スペクトル条件が異なる照明波長によって全体的に画定される実施形態では、各状態ｉはそのような照明波長λの１つを表す。代替的な式では、コントラスト性能指数は、２つの異なる空間周波数帯域における積分パワーの比率として定義することができる。上記に示される方程式は、帯域の１つがＤＣ空間周波数のみを備える具体的な場合である。

It is.
In this equation, F _i (ξ, η) is a Fourier transform of an image f _i (x, y) taken under illumination condition i, and x and y are spatial coordinates for the image. The range defined by R _low and R _high represents the spatial frequency limitation of the object of the fingerprint feature. For example, in one embodiment, R _low can be about 1.5 fringes / mm and R _high can be 3.0 fringes / mm. In embodiments where multispectral conditions are generally defined by different illumination wavelengths, each state i represents one such illumination wavelength λ. In an alternative formula, the contrast figure of merit can be defined as the ratio of integrated power in two different spatial frequency bands. The equation shown above is a specific case where one of the bands comprises only a DC spatial frequency.

In another embodiment, as shown at block 1440, the moving window average and moving window standard deviation are calculated for the multi-spectral stack and used to generate a figure of merit. In this embodiment, for each optical condition i, the moving window average value μ _I and the moving window standard deviation σ _I are calculated from the collected images f _i (x, y). The moving window for each calculation may be the same size and may be conveniently selected to cover approximately 2 to 3 times the fingerprint ridge. Preferably, the size of the window is large enough to cut out fingerprint features, but small enough to maintain background variation. The figure of merit C _i in this embodiment is calculated as the ratio of the moving window standard deviation to the moving window average value.

なお、別の実施形態では、同様のプロセスが行われるが、移動窓範囲（つまり最大値（画像値）−最小値（画像値））が移動窓標準偏差の代わりに使用される。これは、一般的に、ブロック１４４４において示される。よって、前の実施形態と同様に、各光学条件ｉについて、前記移動窓平均値μ_Ｉおよび前記移動窓範囲δ_Ｉは、収集された画像ｆ_ｉ（ｘ，ｙ）から計算される。前記移動窓平均値の計算に対する窓のサイズは再び、好ましくは、指紋特徴を切り出すのに十分大きいが、背景変動を維持するには十分小さい。場合によっては、前記移動窓平均値の計算に対する窓のサイズは、前記移動窓範囲の計算に対するものと同じであり、一実施形態における適切な値は、指紋隆起の約２乃至３倍に及ぶ。この実施形態における性能指数は、前記移動窓平均値の比率として計算される。

In another embodiment, a similar process is performed, but the moving window range (ie, maximum value (image value) −minimum value (image value)) is used instead of the moving window standard deviation. This is generally indicated at block 1444. Therefore, as in the previous embodiment, for each optical condition i, the moving window average value μ _I and the moving window range δ _I are calculated from the collected images f _i (x, y). The window size for the moving window average calculation is again preferably large enough to cut out fingerprint features, but small enough to maintain background variation. In some cases, the window size for the moving window average calculation is the same as for the moving window range calculation, and suitable values in one embodiment range from about two to three times the fingerprint ridge. The figure of merit in this embodiment is calculated as a ratio of the moving window average value.

この実施形態および先のものは、移動窓計算が多重スペクトルスタックについて行われ、移動窓中心度および移動窓変動度を計算する、より一般的な実施形態の特定の場合であると考えることができる。前記具体的実施形態は、中心度が非加重平均値を備えるが、さらに一般的に、ある実施形態において加重平均値または中央値などの任意のその他の種類の統計的中心度を備えることができる場合を例示する。同様に、前記具体的実施形態は、変動度が標準偏差または範囲を備えるが、さらに一般的に、ある実施形態において中央値絶対偏差または平均値の標準誤差などの任意のその他の種類の統計的変動度を備えることができる場合を例示する。

This embodiment and the previous can be considered to be a specific case of a more general embodiment where the moving window calculation is performed on a multi-spectral stack to calculate the moving window center and moving window variability. . Although the specific embodiment has a centrality comprising an unweighted average, more generally, in some embodiments, it can comprise any other type of statistical centrality such as a weighted average or median. The case is illustrated. Similarly, the specific embodiment has a standard deviation or range of variability, but more generally any other type of statistical such as median absolute deviation or standard error of the mean in some embodiments. The case where a variation degree can be provided is illustrated.

  The results of the moving window analysis are evaluated at block 1416 to determine if they match a sample that is a real tissue sample. If so, a biometric function can be performed at block 1448 and typically includes further analysis to identify or verify the identity of the person to which the organization belongs. If the result is inconsistent with the sample that is a real tissue sample, it is identified as impersonation at block 1420. Such classification is accomplished by employing classification methods including linear discriminant analysis, quadratic discriminant analysis, K nearest neighbors, support vector machines, decision trees, and other such methods as known to those skilled in the art. be able to.

  Exemplary results for each of the above methods are provided in FIGS. 15A-15C provide results in which measure values are calculated from moving window fast Fourier transforms, and FIGS. 16A and 16B provide results in which measure values are calculated from moving window averages and moving window standard deviations. 17A and 17B provide a result in which the measure value is calculated from the moving window average value and the moving window range.

  The results shown in FIG. 15A show the in-band contrast performance index results collected for a real finger and derived from the moving window fast Fourier transform for each of multiple samples. The result for each sample is represented by a curve along the abscissa that identifies the illumination plane specified above, so the illumination plane number corresponds to the wavelength somewhat non-linearly, but the abscissa exemplifies wavelength dependence. Then you can think. The data for the individual samples are combined with a gray line and the average over all of the samples is shown with a thicker black line. As expected by the blood extinction characteristics illustrated in FIG. 12, most of the results from a real finger sample show a spectral blue-green color with a significant decrease in contrast at the red wavelength in the plane 4-5. In the region (planes 1 to 3), the characteristic expected “V” shape is shown. Although the contrast value for plane 6 does not match the other planes due to the different optical arrangements used to collect these images, the value from this plane provides additional discriminating capabilities.

  The results for both real fingers and camouflage are shown in FIG. 15B. In this case, the result of a real finger is shown as a diamond connected by a solid line, while the camouflage is shown as a circle connected by a dotted line. The camouflage was a real finger covered with a thin membrane. As is evident from the examination of the results in the illumination planes 4, 5 and 6, the camouflage shows a higher contrast under red illumination than would be expected for a real sample. In some cases, a relative comparison of contrast figure of merit, such as by evaluating whether the red contrast value is less than a certain coefficient α of the blue-green value, may be more robust than an absolute judgment. Various embodiments may utilize different values of α, and preferred values are expected to be about α = 0.5, α = 0.75, etc.

  Data can be processed using multi-dimensional scaling to provide a clear quantitative discrimination between real tissue and impersonation. This is illustrated in FIG. 15C which shows the multidimensional scaling result plotted for two coordinates. The coordinates in this case correspond to the eigenvector scores derived from the principal component analysis (“PCA”) of the contrast data. Specifically, “Factor 1” and “Factor 2” in this explanatory diagram are Massachusetts. Scores for the first two eigenvectors generated from data by the MatLab® program available from The Mathworks, headquartered in Natick State. Real finger derivation results are shown with asterisks, while circles are used to show camouflage derivation results. A clear division is evident between the two, the result of FIG. 15C can be used as a separation plot, and the line drawn in the graph is one potential boundary. An unknown sample that is analyzed in the same way that produces the inline and left side results can be identified as a real sample in block 1416 of FIG. 14, while the underline and right side results can be identified as impersonation. it can. The results here are shown for a two-dimensional space of different regions corresponding to different features of the sample, but more generally, such that different regions of any dimensional space have hyperplanes that define separation, etc. It can be used to distinguish the sample. More generally, as described elsewhere in this disclosure, a variety of different classification methods can be employed to classify the sample as a real finger or not.

  The results of FIG. 16A are similar to those of FIG. 15B, but were derived in an embodiment using the moving window average and moving window standard deviation as described above. Results from real fingers are again shown with diamonds connected by solid lines, while results from camouflage are shown by circles connected by dotted lines. As in the results of FIG. 15B, the camouflage shows a higher contrast under red illumination than the real sample. A separation plot similar to that of FIG. 15C is shown in FIG. 16B extracted using the results of FIG. 16A. In this case, the real finger result is again indicated by an asterisk, while the circle is used to indicate the impersonation result, and the result is again the first two eigenvectors provided by the MatLab® program. Was specifically derived as the eigenvector score from PCA. The part of the two-dimensional space to which the multidimensional scaling method results apply indicates the boundary that allows the camouflage to be distinguished from the real tissue. The general behavior shown in FIG. 16B is similar to that shown in FIG. 15C.

  The results presented in FIGS. 7A and 17B also correspond to those of FIGS. 15B and 15C or FIGS. 16A and 16B. In this case, the results were derived in an embodiment using the moving window average and moving window range of the multispectral stack as described above. Again, the result of the camouflage shown by the circle connected by the dotted line shows a higher contrast under red illumination than the result of the real sample shown by the diamond connected by the solid line in FIG. 17A. The separable plot of FIG. 17B again demonstrates the ability to define a boundary that divides the eigenvector score space into regions where the real tissue results (stars) are clearly separated from the camouflage results (circles). Show. As described above, such boundary capabilities allow unknown results to be classified in block 1416 of FIG. 14 according to which multi-spectral space the derived result applies to.

  In various embodiments, various alternative methods can be used to evaluate contrast data as shown in the drawings such as FIGS. 15A, 16A, and 17A. For example, in one such alternative embodiment, linear discriminant analysis (“LDA”) is used as a contrast discriminant. By way of example only, such an analysis may utilize two categories: a first class of contrast curves generated from human tissue and a second class of contrast curves generated by various pseudo samples.

  The various examples provided above focus on the use of fingers to provide tissue and the use of different illumination light wavelengths to provide multispectral conditions, but this is not intended to be limiting. In other embodiments, various polarization conditions can be used alternatively or additionally to define multispectral conditions. For example, the comparison can be made between two or more polarization conditions, which can include cross polarization imaging, parallel polarization imaging, random polarization imaging, and the like. In addition, incident light can have various polarization states such as linearly polarized light, circularly polarized light, elliptically polarized light, and the like. In yet other examples, various illumination angles can be used to provide multispectral conditions. The penetration depth of light varies as a function of the illumination angle, resulting in an image contrast that can be analyzed as described above in distinguishing between real tissue and camouflage.

  Further, the present invention is not limited to the specific figure of merit described above, and different or additional texture-based figure of merit can be used alternatively or additionally in various embodiments. For example, when deriving a figure of merit for discriminating a sample as described above, statistical moments such as moment invariants and concentration space dependence can be used.

  Spectral contrast is a relative metric. The advantageous result of this fact is that the above method is expected to be robust across a variety of different spectral sensors and under a variety of environmental conditions.

  An overview of additional functionality that can be implemented by the computing device is summarized by the flow diagram of FIG. In some embodiments, the site presumed to be skin is illuminated with white light as indicated at block 1804. This allows the biometric sensor to receive light from a presumed skin site at block 1808. As described above, the received light can be analyzed in a number of different ways when implementing a biometric function. The flow diagram shows how a certain combination of analyzes can be used in implementing the biometric function, but not all steps need to be performed. In other cases, some of the steps may be performed, additional steps may be performed, and / or the steps shown may be performed in a different order than that shown.

  At block 1812, a viability check is performed with the received light and verifies that the site presumed to be skin is not some sort of disguise, typically by verifying that it has biological tissue characteristics. be able to. If impersonation is detected, a warning can be issued at block 1864. The specific type of alerts that may be issued may depend on the environment in which the biometric sensor is deployed, while audio or visual alerts are sometimes issued near the sensor body, and silent alerts are security or Can be communicated to police personnel.

  Light that is scattered and received from a site that is estimated to be skin can be used in block 1820 to derive a surface image of the site that is estimated to be skin. If the site presumed to be skin is the palmar surface of a finger, such a surface image includes a depiction of finger ridges and groove patterns so that in block 1824 it can be compared to a conventional fingerprint database. . Additionally or alternatively, the received light can be used to derive a spatial spectral image at block 1828. This image can be compared to a spatial spectral database having an image associated with the individual at block 1832. In any case, the comparison allows the individual to be identified at block 1836 as a result of the comparison. It is generally expected that higher confidence identification can be made by using complete spatial spectral information and provide a comparison with the spatial spectral image. However, in some applications, there may be higher availability of conventional fingerprint data, with some individuals having their fingerprints stored in a large law enforcement fingerprint database but not stored in a spatial spectrum database. In such cases, embodiments of the present invention advantageously allow conventional fingerprint image extraction to be identified.

  The spatial spectral data may provide additional confidence in the identification, regardless of whether the identification is by comparison with a conventional fingerprint database or through comparison with spatial spectral information. Contains information. For example, as shown in block 1840, demographic and / or anthropometric characteristics can be estimated from the received light. If the database entry matching the image at block 1836 includes demographic or anthropometric information, a consistency check can be performed at block 1844. For example, the individual can be identified as a white male between the ages of 20 and 35 based on inferred demographic and anthropometric characteristics. If the matching database entry identifies the individual as a 68-year-old black woman, there is an apparent discrepancy that triggers a warning at block 1864.

  Other information, such as analyte concentration in block 1856, can also be determined from the received light. Depending on the determined analyte level, various measures can sometimes be taken. For example, if the blood alcohol level exceeds a certain threshold, a car engine ignition may be prohibited, or if a medical patient's blood glucose level exceeds a certain threshold, a warning may be issued. Other physiological parameters, such as skin dryness, can be estimated in other applications, and sometimes other measures are taken accordingly.

  FIG. 19 is a flow diagram similar to an exemplary application of a texture biometric sensor. The sensor is used at block 1904 by positioning an individual's skin site in contact with the detector. As mentioned above, the detector can be relatively small so that only a portion of the finger surface is in contact with the detector, and due to the nature of the texture biometric, it can be contacted during various measurements. Variations in specific parts of the surface that are placed will not be harmful. At block 1908, data is collected by illuminating the skin site and receiving light scattered from the skin site by the detector at block 1912.

  The flow diagram shows that different types of analysis can be performed. Each of these analyzes need not be performed in every case, and in fact it is generally expected that a single type of analysis will be used in most applications. One category of analysis, generally indicated at block 1916, uses purely spectral comparison of information. Another category of analysis generally indicated at blocks 1920 and 1928 is to determine image texture from the spatial spectral information of the light received at block 1920 and to compare the image texture to a database of texture biometric information at block 1928. By using image texture information. Either or both types of analysis perform a biometric function such as identification of an individual at block 1932.

  Thus, it will be appreciated by one skilled in the art that various embodiments, alternative constructions, and the like may be used without departing from the spirit of the invention, while describing several embodiments. Therefore, the above description should not be taken as limiting the scope of the invention which is defined in the following claims.

FIG. 1 provides a front view of a non-contact biometric sensor in one embodiment of the present invention. FIG. 2A provides an illustration of the structure of a Bayer color filter array that can be used in embodiments of the present invention. FIG. 2B is a graph illustrating a color response curve of a Bayer color filter array as illustrated in FIG. 2A. FIG. 3 provides a front view of a non-contact biometric sensor in another embodiment of the present invention. FIG. 4 is a top view of a sensor configuration that collects data during relative movement between a skin site and an optically active area of the sensor. FIG. 5 illustrates a multispectral data cube that can be used in certain embodiments of the present invention. FIG. 6 is a front view of a contact-type biometric sensor according to an embodiment of the present invention. FIG. 7A provides a side view of a contact biometric sensor in one embodiment. FIG. 7B provides a side view of a contact biometric sensor in another embodiment. FIG. 8 provides a front view of a contact biometric sensor in a further embodiment of the present invention. FIG. 9 provides a diagram of a multispectral biometric sensor in another embodiment of the present invention. FIG. 10A illustrates the structure of a contact-type texture biometric sensor according to an embodiment of the present invention. FIG. 10B provides a side view of a contact-type texture biometric sensor in one configuration. FIG. 10C provides a side view of a contact-type texture biometric sensor in another configuration. FIG. 11 is a schematic diagram of a computer system that can be used to manage the functionality of contact and non-contact biometric sensors according to embodiments of the present invention. FIG. 12 is a graph of the oxygen-containing blood absorption spectrum. FIG. 13 provides various images of real fingers and camouflage taken under lighting conditions. FIG. 14 is a flow diagram summarizing certain methods of the present invention. FIG. 15A provides a plot of the results collected in the first embodiment of the present invention. FIG. 15B provides the data comparison results of FIG. 15A for real fingers and camouflage. FIG. 15C is a separation plot resulting from the multidimensional scaling method of the data of FIG. 15A. FIG. 16A provides a comparison of the results collected in the second embodiment of the present invention between a real finger and a camouflage. FIG. 16B is a separable plot resulting from the multidimensional scaling method of the data of FIG. 16A. FIG. 17A provides a comparison of the results collected in the third embodiment of the present invention between a real finger and a camouflage. FIG. 17B is a separable plot resulting from the multidimensional scaling method of the data of FIG. 17A. FIG. 18 is a flow diagram summarizing a method of using contact and non-contact biometric sensors, illustrating a number of different biometric functions that can be implemented. FIG. 19 is a flow diagram summarizing a method of operation of a contact-type texture biometric sensor according to an embodiment of the present invention.

Claims (77)

A method for evaluating the authenticity of a sample presented for biometric evaluation, comprising:
Illuminating the sample under a plurality of different optical conditions;
Receiving light scattered from the sample for each of the plurality of different optical conditions;
Forming a plurality of images, each image being formed from the received light for each of the plurality of different optical conditions; and
Generating a plurality of quality sensitivities, each quality sensitivity being generated from each of the plurality of images; and
Determining whether the generated plurality of quality sensitivities are consistent with the sample being a real, non-hidden biological tissue.   The method of claim 1, wherein the quality sensitivity comprises a degree of image contrast.   The method of claim 1, wherein illuminating the sample under a plurality of different optical conditions comprises illuminating the sample with light having a plurality of different wavelengths.   The method of claim 1, wherein illuminating the sample under a plurality of different optical conditions comprises illuminating the sample with light under a plurality of different polarization conditions.   The method of claim 1, wherein illuminating the sample under a plurality of different optical conditions comprises illuminating the sample with light at a plurality of different illumination angles.   The method of claim 1, wherein generating the plurality of quality sensitivities includes performing a spatial moving window analysis of each of the plurality of images.   The method of claim 6, wherein each of the plurality of quality sensitivities comprises a degree of image contrast within a constant spatial frequency. Illuminating the sample under a plurality of different optical conditions comprises illuminating the sample with light having a plurality of different wavelengths;
Determining whether the generated plurality of quality sensitivities match a sample that is an unhidden biological tissue, confirming that an image contrast under red illumination is lower than an image contrast under blue illumination The method of claim 6 comprising: Illuminating the sample under a plurality of different optical conditions comprises illuminating the sample with light having a plurality of different wavelengths;
Determining whether the generated plurality of quality sensitivities matches a sample that is an unhidden biological tissue includes confirming that the image contrast under red illumination is less than a predetermined value. The method according to claim 6.   The method of claim 6, wherein performing the spatial moving window analysis comprises calculating a moving window Fourier transform for the plurality of images.   The method according to claim 6, wherein performing the spatial moving window analysis includes calculating moving window centrality and moving window variability of the plurality of images.   The method of claim 11, wherein the moving window centrality comprises a moving window average value, and the moving window variability comprises a moving window standard deviation.   The method of claim 11, wherein the moving window centrality comprises a moving window average value, and the moving window variability comprises a moving window range. Generating the plurality of quality sensitivities further comprises applying a multidimensional scaling method to map the plurality of quality sensitivities to points in a multidimensional space;
Determining whether the generated plurality of quality sensitivities match a sample that is a real, non-hidden biological tissue, determines whether the point is within a predetermined region of the multidimensional space. The method of claim 6 including the step of:   The method of claim 14, wherein the multidimensional space is a two-dimensional space. A method for evaluating the authenticity of a sample presented for biometric evaluation, comprising:
Illuminating the sample with light having a plurality of different wavelengths;
Receiving light scattered from the sample for each of the plurality of different wavelengths;
Forming a plurality of images, each image being formed from the received light for each of the plurality of different wavelengths; and
Generating a plurality of image contrast degrees, each image contrast degree being obtained from the one of the plurality of images by performing a spatial moving window analysis of one of the plurality of images. Generated steps, and
Determining from the plurality of image contrast degrees whether the sample has an image contrast under red illumination that is lower than an image contrast under blue illumination.   The method of claim 16, wherein performing the spatial moving window analysis comprises calculating a moving window Fourier transform for one of the plurality of images.   The method of claim 16, wherein performing the spatial moving window analysis comprises calculating moving window centrality and moving window variability of one of the plurality of images. The step of determining from the plurality of image contrast degrees whether the sample has an image contrast under red illumination that is lower than an image contrast under blue illumination is obtained by determining the plurality of image contrast degrees in a multidimensional space. Applying a multidimensional scaling method to map to
17. The method of claim 16, comprising determining whether the point is within a predetermined region of the multidimensional space. An apparatus for evaluating the authenticity of a sample presented for biometric evaluation,
Means for illuminating the sample under a plurality of different optical conditions;
Means for receiving light scattered from the sample for each of the plurality of different optical conditions;
Means for forming a plurality of images, each image being formed from the received light for each of the plurality of different optical conditions;
Means for generating a plurality of quality sensitivities, each quality sensitivity being generated from each of the plurality of images;
Means for determining whether the generated plurality of quality sensitivities are consistent with a sample that is a genuine, non-hidden biological tissue.   21. The apparatus of claim 20, wherein the means for generating the plurality of quality sensitivities includes means for performing a spatial moving window analysis of each of the plurality of images. Said means for generating an optical contrast measure further comprises means for generating a multidimensional scaling such that the figure of merit is mapped to a point in multidimensional space;
The means for determining whether the generated plurality of quality sensitivities are consistent with a sample that is a real, non-hidden biological tissue, wherein the point is within a predetermined region of the multidimensional space. The apparatus of claim 21 including means for determining whether. A method for performing a biometric function comprising:
Illuminating a site presumed to be an individual's skin with illumination light, wherein the site presumed to be skin is in contact with a surface;
Receiving light scattered from a site presumed to be the skin, wherein the light is received substantially in a plane including the surface;
Forming an image from the received light;
Generating image quality sensitivity from the image;
Analyzing the generated image quality sensitivity to perform the biometric function.   The biometric function includes an anti-spoofing function, and the step of analyzing the generated image quality sensitivity determines whether or not a site estimated as the skin includes biological tissue from the generated image quality sensitivity. 24. The method of claim 23, comprising steps.   24. The biometric function comprises an identity verification function, and the step of analyzing the generated image quality sensitivity includes determining the individual's identity from the generated image quality sensitivity. Method.   The biometric function includes a demographic function or an anthropometric function, and the step of analyzing the generated image quality sensitivity is a step of estimating an individual demographic characteristic or anthropometric characteristic from the generated image quality sensitivity. 24. The method of claim 23, comprising: The surface is a surface of an image detector;
24. The method of claim 23, wherein receiving light scattered from a site presumed to be skin includes receiving light scattered from a site presumed to be skin by the image detector. Receiving light scattered from the presumed site of the skin,
Translating a pattern of light from the plane to an image detector located outside the plane without substantial degradation or attenuation of the pattern;
24. receiving the translated pattern at the image detector.   24. The method of claim 23, wherein the illumination light is white light. The image comprises a plurality of images corresponding to various wavelengths,
30. The method of claim 29, wherein generating the image quality sensitivity comprises performing a spatial moving window analysis of each of the plurality of images.   The method of claim 30, wherein performing the spatial moving window analysis comprises calculating a moving window Fourier transform for the plurality of images.   The method of claim 30, wherein performing the spatial moving window analysis comprises calculating moving window centricity and moving window variability of the plurality of images.   24. The method of claim 23, wherein receiving light scattered from the site estimated to be skin includes receiving light scattered from the site estimated to be skin in a monochromatic image detector.   24. The method of claim 23, wherein receiving light scattered from a site presumed to be skin includes receiving light scattered from a site presumed to be skin by a color image detector. Analyzing the generated image quality sensitivity to perform the biometric function comprises comparing the generated image quality sensitivity to a reference image quality sensitivity;
The reference image quality sensitivity is generated from a reference image formed from light scattered from a reference skin site,
24. The method of claim 23, wherein the site presumed to be skin is substantially different from the reference skin site.   24. The method of claim 23, further comprising comparing spectral characteristics of the received light with reference spectral characteristics when performing the biometric function. A surface configured to contact the presumed site of the individual's skin;
An illumination subsystem arranged to illuminate the presumed site of skin when the presumed site of skin is in contact with the surface;
A detection subsystem arranged to receive light scattered from a site presumed to be the skin, wherein the light is received substantially in a plane including the surface;
In conjunction with the detection subsystem,
Instructions for forming an image from the received light;
Instructions for generating image quality sensitivity from the image;
And a computing unit having instructions for analyzing the generated image quality sensitivity to perform the biometric function. The biometric function includes an anti-spoofing function,
38. The instructions for analyzing the generated image quality sensitivity comprise instructions for determining whether a site that is estimated to be skin from the generated image quality sensitivity comprises a living tissue. The biometric sensor described in 1. The biometric function includes an identity confirmation function,
38. The biometric sensor of claim 37, wherein the instructions for analyzing the generated image quality sensitivity comprise instructions for determining an individual's identity from the generated image quality sensitivity. The biometric function includes a demographic function or an anthropometric function,
38. The biometric of claim 37, wherein the instructions for analyzing the generated image quality sensitivity comprise instructions for estimating an individual demographic characteristic or anthropometric characteristic from the generated image quality sensitivity. Sensor. An image detector,
The surface is the surface of the image detector;
38. The biometric sensor of claim 37, wherein the detection subsystem comprises the image detector and is configured to receive light scattered from a site presumed to be the skin at the image detector. An image detector disposed outside the plane;
An optical arrangement configured to translate the pattern of light from the plane to the image detector without substantial degradation or attenuation of the pattern;
38. The biometric sensor of claim 37, wherein the detection subsystem comprises the image detector and is configured to receive the translated pattern at the image detector.   38. The biometric sensor of claim 37, wherein the illumination subsystem is configured to illuminate a site presumed to be the skin with white light. The image comprises a plurality of images corresponding to various wavelengths,
44. The biometric sensor of claim 43, wherein the instructions for generating the image quality sensitivity comprise instructions for performing a spatial moving window analysis of each of the plurality of images.   45. The biometric sensor of claim 44, wherein the instructions for performing the spatial moving window analysis comprise instructions for calculating a moving window Fourier transform for the plurality of images.   45. The biometric sensor of claim 44, wherein the instructions for performing the spatial moving window method analysis comprise instructions for calculating moving window centrality and moving window variability of the plurality of images.   38. The biometric sensor of claim 37, wherein the detection subsystem comprises a monochromatic image detector and is configured to receive light scattered from a site presumed to be the skin in the monochromatic image detector.   38. The biometric sensor of claim 37, wherein the detection subsystem comprises a color image detector and is configured to receive light scattered from a site presumed to be the skin in the color image detector. The instructions for analyzing the generated image quality sensitivity to perform the biometric function comprise instructions for comparing the generated image quality sensitivity with a reference image quality sensitivity;
The reference image quality sensitivity is generated from a reference image formed from light scattered from a reference skin site,
38. The biometric sensor of claim 37, wherein the site presumed to be skin is substantially different from the reference skin site.   38. The biometric sensor of claim 37, wherein the computing unit further comprises instructions for comparing spectral characteristics of the received light with reference spectral characteristics when performing the biometric function. A white light illumination subsystem arranged to illuminate the presumed site of the individual's skin with white light;
A detection subsystem comprising a color imaging device arranged to receive light scattered from a site presumed to be the skin, and receiving the received light;
In conjunction with the detection subsystem,
Instructions for deriving, from the received light, a plurality of spatially distributed images of the site estimated to be the skin by the color image device, wherein the plurality of spatially distributed images Instructions corresponding to different amounts of illuminated tissue;
And a computing unit having instructions for analyzing the plurality of spatially distributed images to perform a biometric function.   The biometric function includes an anti-spoofing function, and the instruction for analyzing the plurality of spatially distributed images includes an instruction for determining whether the estimated site includes a living tissue. 52. The biometric sensor of claim 51.   The instructions for analyzing the plurality of spatially distributed images to perform the biometric function may include the plurality of spatially distributed images to estimate demographic or anthropometric characteristics of the individual. 52. The biometric sensor of claim 51, comprising instructions for analyzing.   The instructions for analyzing the plurality of spatially distributed images to perform the biometric function analyze the plurality of spatially distributed images to determine a concentration of an analyte in the individual's blood. 52. The biometric sensor of claim 51, comprising instructions for performing.   52. The bio of claim 51, further comprising a platen in contact with the estimated site of skin, wherein the white light illumination subsystem is configured to illuminate the estimated site of skin through the platen. Metric sensor.   52. The white light illumination subsystem is configured to illuminate the presumed site of skin when the presumed site of skin is not in physical contact with the biometric sensor. Biometric sensor.   52. The biometric sensor of claim 51, wherein the white light illumination subsystem comprises a broadband white light source.   52. The biometric sensor of claim 51, wherein the white light illumination subsystem comprises a plurality of narrowband light sources and an optical arrangement for combining light provided by the plurality of narrowband light sources.   59. The biometric sensor of claim 58, wherein the plurality of narrowband light sources provide light of a wavelength corresponding to each of a set of primary colors. The illumination subsystem comprises a first polarizer arranged to polarize the white light;
The detection system comprises a second polarizer arranged to encounter the received light;
52. The biometric sensor of claim 51, wherein the first and second polarizers intersect each other.   52. The biometric sensor of claim 51, wherein the detection system comprises an infrared filter positioned to encounter the received light before the received light is incident on the color imaging device. The site presumed to be the skin is the palm surface of a finger or hand,
The biometric function comprises biometric identification;
The instructions for analyzing the plurality of spatially distributed images are:
Instructions for deriving a surface fingerprint image or surface palmprint image of the site presumed to be the skin from the plurality of spatially distributed images;
52. The biometric sensor of claim 51, comprising: instructions for comparing the surface fingerprint image or surface palmprint image with a database of fingerprint or palmprint images to identify the identity of the individual. The biometric function comprises biometric identification;
The instructions for analyzing the plurality of spatially distributed images include instructions for comparing the plurality of spatially distributed images for identifying the individual's identity with a database of multispectral images. 52. The biometric sensor of claim 51, comprising: The illumination subsystem is configured to illuminate a site presumed to be the skin within an illumination area;
52. The biometric sensor according to claim 51, wherein the portion estimated to be skin and the illumination region are moving relatively. A method for performing a biometric function comprising:
Illuminating the presumed portion of the individual's skin with white light;
Receiving light scattered from a site estimated to be the skin by a color image device on which the received light is incident;
Deriving, from the received light, a plurality of spatially distributed images of the site estimated to be the skin by the color imaging device, wherein the plurality of spatially distributed images are illuminated by the individual; Steps corresponding to various amounts of the treated tissue;
Analyzing the plurality of spatially distributed images to perform the biometric function.   66. The biometric function comprises an anti-spoofing function, and analyzing the plurality of spatially distributed images includes determining whether the estimated site comprises biological tissue. the method of.   66. Analyzing the plurality of spatially distributed images comprises analyzing the plurality of spatially distributed images to estimate demographic or anthropometric characteristics of the individual. the method of.   Analyzing the plurality of spatially distributed images to perform the biometric function comprises analyzing the plurality of spatially distributed images to determine a concentration of an analyte in the blood of the individual. 66. The method of claim 65, comprising.   66. The method of claim 65, wherein illuminating the presumed site of skin comprises directing the white light through a platen in contact with the presumed site of skin.   66. The method of claim 65, wherein illuminating the person's estimated skin area with white light includes illuminating the individual's estimated skin area with a broadband white light source. Illuminating the presumed skin portion of the individual with white light,
Generating a plurality of narrowband rays comprising a plurality of narrowband light sources;
66. combining the plurality of narrowband rays.   72. The method of claim 71, wherein the plurality of narrowband rays have wavelengths that correspond to a set of primary colors. Illuminating the site presumed to be the skin includes polarizing the white light with a first polarization,
Receiving the light scattered from the site presumed to be the skin includes changing the received light with a second polarization;
66. The method of claim 65, wherein the first and second polarizations are substantially crossed with respect to each other.   Receiving light scattered from a site presumed to be skin includes filtering the received light at an infrared wavelength before the received light is incident on the color imaging device; Item 66. The method according to Item 65. The site presumed to be the skin is the palm surface of a finger or hand,
The biometric function comprises biometric identification;
Analyzing the plurality of spatially distributed images;
Deriving from the plurality of spatially distributed images a surface fingerprint image or surface palmprint image of the site presumed to be the skin;
66. The method of claim 65, comprising comparing the surface fingerprint image or surface palmprint image to a database of fingerprint or palmprint images to identify the individual's identity. The biometric function comprises biometric identification;
66. The step of analyzing the plurality of spatially distributed images comprises comparing the plurality of spatially distributed images to a database of multispectral images to identify the individual's identity. The method described. The step of illuminating the part of the person estimated to be the skin with white light is performed within an illumination area,
66. The method of claim 65, wherein the site presumed to be skin and the illumination area are moving relative to each other.

Images