JP2013218393A - Imaging device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate an age by using human body detection, and to set various conditions by using an estimated age.SOLUTION: A digital signal processing circuit 18 detects a human body from an image signal obtained by photographing, and detects a face part. The detected face part is collated with a face image database so as to be recognized. A system control circuit 20 estimates the age of the human body from the ratio of the head to the shoulder of the human body, and when the age of the human body is equal to or less than a threshold age, outputs a message for encouraging the user to update the face image database.

Description

  The present invention relates to an imaging apparatus, and more particularly, to age estimation of a subject person.

  Conventionally, a technique for detecting a human body as a subject from an image obtained by photographing has been proposed.

  For example, in the following Patent Document 1, a template management unit that stores a template made of a closed curve representing a part of a contour of a human body model or a part of a human body, an image data receiving unit that inputs an image to be detected, and An object detection apparatus including a head position detection unit that detects a human body from an image by matching the input image using a plurality of templates is disclosed.

  Patent Document 2 discloses a contour extracting unit that extracts contour data of a determination target included in a two-dimensional image, and a shape value generating unit that calculates a ratio between a straight line portion and a curved portion of the contour from the extracted contour data. A human shape determination method including determination means for performing human determination by comparing a predetermined threshold value and a ratio of a straight line segment of a contour data calculated by a shape value generation means and a curve component is disclosed.

  Patent Document 3 discloses an edge image extraction unit that forms an edge image from an image, an edge direction of a predetermined pixel, an edge direction of an edge pixel existing in a neighboring area of the predetermined pixel, and a predetermined pixel and a neighboring area. An apparatus is disclosed that calculates the number of edge pixels defined by the spatial positional relationship with the edge pixels to be used as an image feature amount to improve the identification accuracy of a human image.

  Patent Document 4 discloses storing a standard face size and calculating an actual distance from this size and the photographed face size to the face of the subject.

  Furthermore, Patent Document 5 discloses a technique for determining whether or not a person is a person based on the ratio of the dimensions of the head and torso.

JP 2005-149145 A JP 2003-132340 A JP 2010-117772 A JP 2007-248698 A JP 2002-298142 A

  By the way, when a person or a person as a subject is included in an image obtained by an imaging device such as a digital camera, it is detected that the person or person is included in the image by various methods as described above. However, when a person or a person is detected, how to apply the detection information has not yet been sufficiently studied.

  For example, in the above-mentioned Patent Document 1, it is disclosed that a technique for detecting a human body from an image is used for security management of a facility. There is no mention.

  An object of the present invention is to provide an apparatus capable of detecting a human body included in a photographed image, estimating the age of a person as a subject using the detection information, and optimizing various conditions based on the estimated age. It is to provide.

  The present invention is an imaging apparatus, comprising: an optical system including a lens; an imaging unit that converts a subject image formed by the optical system into an electrical signal; and an edge pattern of an image signal obtained by the imaging unit. Detecting means for detecting a human body using, face detecting means for detecting a face portion from an image signal obtained by the imaging means, storage means for storing data of one or a plurality of face images, and a detected face portion By collating the face image data stored in the storage means, the face recognition means for recognizing the detected face portion, the age of the human body is estimated based on the size of the human body, and based on the estimated age Control means for executing processing for updating the face image stored in the storage means.

  The present invention also relates to an imaging apparatus, an optical system including a lens, imaging means for converting a subject image formed by the optical system into an electrical signal, and an edge of an image signal obtained by the imaging means It is characterized by comprising detection means for detecting a human body using a pattern, and control means for estimating the age of the human body based on the size of the human body and setting imaging conditions based on the estimated age.

  In one embodiment of the present invention, the control means estimates the age using the ratio of the length of the head and the length of the shoulder as the size of the human body.

  In another embodiment of the present invention, the control means outputs a message prompting the user to update the face image when the estimated age is equal to or lower than a threshold age. For example, when the estimated age is less than or equal to the threshold age, the control means sets the update timing of the face image corresponding to the face recognized by the face recognition means, and is recognized by the face recognition means When the elapsed time reaches the update timing, a message that prompts the user to update the face image is output.

  In another embodiment of the present invention, the control unit sets the shutter speed relatively high when the estimated age is equal to or lower than the threshold age.

  According to the present invention, the human body included in the photographed image is detected, the age of the person who is the subject is estimated using the detection information, and the face image database is updated and the photographing condition is optimized based on the estimated age. Can be planned.

It is a block diagram of the digital camera in an embodiment. It is a processing flowchart of an embodiment. It is another process flowchart of embodiment. It is a human body detection process flowchart of embodiment. It is a typical explanatory view of human body detection.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the following embodiment is an exemplification, and the present invention is not limited to the following embodiment.

  First, a basic configuration of a digital camera as an imaging apparatus in the present embodiment will be described.

  FIG. 1 is a block diagram showing the configuration of a digital camera according to this embodiment. The subject image is formed on the image sensor 14 through the lens 10 and the shutter / aperture 12. The image sensor 14 converts the subject image into an electrical signal and outputs it to the analog preprocessing circuit (analog front end) 16 as an analog image signal. The diaphragm is driven and controlled by an exposure control signal from the system control circuit 20 (exposure control: AE). The lens 10 is driven and controlled by a focus control signal from the system control circuit 20 (focus control: AF).

  The image sensor 14 is provided with an optical filter such as an IR cut filter, an optical low-pass filter, or a color filter array, and forms an object image on the image sensor 14. As the image sensor 14, a CCD image sensor or a CMOS image sensor is used.

  The analog preprocessing circuit (analog front end) 16 includes an analog amplifier, a gain controller, and an AD converter, amplifies the analog image signal from the image sensor 14, converts it into a digital image signal, and outputs the digital image signal to the digital signal processing circuit 18. .

  The digital signal processing circuit 18 executes gain correction (white balance adjustment), gamma correction, synchronization processing, RGB-YC conversion, noise reduction processing, contour correction, and JPEG compression on the supplied digital image signal. To do.

  The gain correction (white balance adjustment) is a process of correcting the RGB balance due to the light source color temperature, and adjusts the gains of the input R signal, G signal, and B signal, respectively. As a method of adjusting the gain, the user manually inputs the type of light source (sunlight, electric light), etc., and adjusts the gain based on the input light source. There are a method of arranging and correcting these by photographing with a camera, and a method of automatically determining and correcting a light source with a camera (auto white balance).

  The gamma correction is a process for adjusting the output characteristics of the image sensor 14 to the input / output characteristics (sRGB) of the display.

  The synchronization processing is calculated by calculating a signal of an insufficient color from the color signals of neighboring pixels because only one color signal exists in one pixel in a single-plate imaging method using a Bayer color filter. It is processing. As a method of synchronization processing, there are a method of averaging the values of neighboring pixels, a method of weighted averaging according to the distance from the target pixel, and the like.

The RGB-YC conversion process is a process of converting the R signal, the G signal, and the B signal that have been subjected to the synchronization process into a Y signal, a Cb signal, and a Cr signal. That is,
Y = 0.30R + 0.59G + 0.11B
Cb = BY
Cr = R−Y
To convert the luminance signal into a Y signal, a color difference signal as a Cb signal, and a Cr signal.

  The noise reduction process is a process for removing isolated points such as pulse noise using a median filter or the like. This process removes noise, but at the same time lowers the resolution, so it is usually executed for the color difference signals Cb and Cr.

  The contour correction processing is processing for correcting a decrease in MTF (Modulation Transfer Function) due to the influence of an optical low-pass filter or the like, and adds a contour signal to the original image signal by contour extraction processing and nonlinear processing. Normally, the contour correction process is performed on the luminance signal.

  In JPEG compression, each of the Y signal of the luminance signal, the Cb signal of the color difference signal, and the Cr signal is divided into blocks of 8 pixels × 8 pixels, and compression is performed by sequentially performing DCT conversion, quantization, and Huffman coding for each block.

  The digital signal processing circuit 18 stores the image signal compressed by the above processing in the buffer memory 28 via the data bus 22, reads out the image data stored in the buffer memory 28, and displays it on the liquid crystal monitor 26. . Alternatively, the image signal is stored in the memory card 24.

  The system control circuit 20 controls the operation of each unit based on signals input from various switches (SW) 19. For example, the operation of each unit is controlled based on the operation signal of the shutter button, and the image signal obtained by photographing is displayed on the liquid crystal monitor 26 or stored in the memory card 24. Further, at the time of shooting, exposure control (AE) and focus control (AF) are executed as described above. Regarding focus control, there are a contrast detection type AF and a TTL phase difference detection type AF. In contrast detection AF, the maximum contrast point of the captured image is used as the in-focus position. The focus is moved slightly from the current position, and when the contrast decreases, it moves in the opposite direction, and when the contrast increases, it moves further in the same direction. In this case, the position becomes the in-focus position when it decreases regardless of the direction of movement (so-called hill-climbing method). In the TTL phase difference detection AF, the lens transmitted light is measured by a distance measuring unit to determine the focus position of the lens, and the distance measurement unit determines an image according to the direction and amount of shift from the focus position. The in-focus position is determined using the fact that the lens moves left and right.

  In such a configuration, the digital signal processing circuit 18 executes each of the above processes, executes a human body detection process for detecting whether or not a human body is included in the obtained image signal, and outputs the detection result. Output to the system control circuit 20.

  In addition, the digital signal processing circuit 18 detects whether or not a face is included in the obtained image signal. If the face is included, the digital signal processing circuit 18 further recognizes the detected face. That is, the digital signal processing circuit 18 executes face detection processing (FD) and face recognition processing (FR). Face detection is performed using the contour of the face and the relative position and size of each part (eyes, nose, mouth) of the face. The face may be detected using color data (whether it is skin color). Face AF for detecting a face and controlling it to focus on the face, Face AE for detecting and controlling the face, and Face WB for adjusting the white balance by detecting the face are well known. The face detection algorithm used in the face WB can be applied as it is.

  Face recognition is performed by comparing the detected facial features with a face database stored in advance in the memory of the digital signal processing circuit 18 and identifying matching faces. That is, the faces of persons A, B, C,... Are registered in advance in the memory together with their names, and the face portion detected from the obtained image signal and the face image stored in the memory are stored. For example, if it matches the face of the person A best when it is checked against the database, the detected face is recognized as the face of the person A. Face recognition algorithms are known and are identified by extracting features that stand out from the face. For example, the contour of the face and the relative position and size of each part such as eyes, nose and mouth are used as features. A face image having a matching feature is searched using such a feature. Alternatively, standardized data may be created from a large number of face images, and each face image may be compressed only to the difference from the standard data, leaving only the data necessary for face recognition. The recognition algorithm may be either a method of directly comparing visual features geometrically or a method of statistically digitizing an image and comparing the numerical value with a template, such as linear discriminant analysis or elastic bunch graph. Matching, hidden Markov models, dynamic link matching with neuronal motivation, etc. can be used. Although the output form of the recognition result is arbitrary, for example, it is output by displaying the name of the recognized person A on the liquid crystal monitor 26 in the vicinity of the detected face as a tag.

  The system control circuit 20 estimates the age of the subject person using the human body detection information from the digital signal processing circuit 18. Specifically, the age of the subject person is estimated using the size of the human body included in the human body detection information. It is known that the size of each part of the human body changes with age. For example, the ratio of the entire head changes with age. The ratio of head size to shoulder size also changes with age. The system control circuit 20 stores the relationship between the size of the human body or the ratio of each part of the human body and the age in advance in a memory, and estimates the age of the subject person by referring to this table. Then, the system control circuit 20 sets various conditions based on the estimated age. For example, the digital signal processing circuit 18 recognizes the face by comparing the face detected as described above with a database. If the estimated age is less than a certain age, the face changes relatively greatly. In consideration of this, processing for updating a database stored in advance in the memory to the current face is executed.

  The human body size is more specifically the size of the upper body of the human body image on the imaging surface, and the upper body size includes the size of the head (head length, head width), shoulder Size (shoulder width) etc. are included. The size is defined by the number of pixels constituting the head and shoulders.

  FIG. 2 shows a processing flowchart of the present embodiment. First, when a subject image is captured (S101), the digital signal processing circuit 18 detects a human body from the obtained image (S102). Hereinafter, the human body detection process is referred to as HBD (Human Body Detection). Details of the HBD will be described later.

  Further, in parallel with the HBD or in tandem with the HBD, the digital signal processing circuit 18 detects a face from the obtained image (S103). Hereinafter, face detection is referred to as FD (Face Detection). Since HBD is a process that mainly detects the contour of the upper body of the human body, and FD is a process that detects the face of the human body, both are different processes. Of course, it is possible to improve the efficiency of the FD by using the detection result of the HBD in the FD. In this sense, it would be preferable to perform FD after HBD.

  After detecting the human body and face from the captured image, the system control circuit 20 estimates the age of the person using the HBD detection result, specifically, the detected human body size (S104). For example, the ratio between the length of the head of the human body and the length of the shoulder is calculated, the memory in which a table defining the correspondence between the ratio and the age is stored in advance, and the age corresponding to the calculated ratio is calculated. read out. The table that defines the correspondence between the ratio and the age is, for example, age 0-3 years, 3-6 years, 6-9 years, 9-12 years, 12-15 years, 15-15 years, 15-18 years , 18 to 21 years old, 22 years old and over, and define the average ratio for each category.

After estimating the age of the person, the system control circuit 20 determines whether or not the estimated age is equal to or less than a predetermined threshold age (S105). The threshold age can be set to 18 years old, for example. When the estimated age is 18 years old or less, it is considered that the face may change relatively greatly, and the face image database, which is basic data when the digital signal processing circuit 18 performs face recognition, is updated. Processing for prompting the user is executed (S106). Specifically, a message indicating that the latest face image of the currently photographed person should be registered in the database is displayed on the liquid crystal monitor 26 as a message to prompt the user. The message is not necessarily limited to letters, but can also be made by turning on an indicator. For example, an icon representing a database is blinked. When the user inputs the latest face image of the person being photographed according to this message (S107), the system control circuit 20 re-registers the input latest face image in the face image database of the digital signal processing circuit 18. (S108).

  On the other hand, when the estimated age exceeds 18 years old, it is considered that the face does not change significantly, and the face image database is not updated.

  After updating the face image database according to the estimated age in this way, the digital signal processing circuit 18 compares the face detected in S103 with the face image database stored in the memory to recognize the face. (S108). Hereinafter, face recognition is referred to as FR (Face Recognition).

  As described above, when the age is relatively young and the face is assumed to change relatively greatly, the accuracy of FR can be ensured by updating the face image database. That is, when the age is young, the face image database is sequentially updated according to the message, so that even if the face changes in a relatively short time, the detected face can be recognized correctly. On the other hand, if the face image database is not appropriately updated even when the age is young, the face and face image database are often detected when the age is young. Will not match and will not be FR or the FR will be inaccurate.

  In FIG. 2, the FR is executed in S109 after the face image data is updated in S108. However, the FR may be executed after the age is estimated, and the face image data may be updated after the FR is executed. it can. That is, after recognizing a face part detected by executing FR, a face image in the face image database corresponding to the face part is specified, so that the age of the face image is estimated. Therefore, the update timing is set when the estimated age of the face image is equal to or lower than the threshold age. After setting the update timing, the timer is activated to determine whether or not the update timing has been reached, and when the update timing has been reached, a message prompting the update is displayed on the liquid crystal monitor 26. For example, face image A, face image B, and face image C exist in the face image database, and the detected face portion corresponds to face image A as a result of FR. In this case, if the estimated age of the face image A is less than or equal to the threshold age, “update after 6 months” is set as the attribute of the face image A. Then, the timer is started together with the execution of FR, and when 6 months have passed, a message that the face image A should be updated is displayed. In this sense, it can be said that the update timing is set for each of a plurality of face images in the face image database by FR.

  In addition, when the face image database is urged to be updated in S106, it is preferable to notify the frequency together. That is, in general, the face changes relatively greatly as the younger, so the update frequency is set higher as the age is younger. For example, 3 to 6 years old is urged to update once every three months, and 15 to 18 years old is urged to update once a year.

  Further, when prompting the update of the face image database in S106, it may be prompted according to the result of FR instead of promptly prompting if the estimated age is equal to or lower than the threshold age. That is, when the estimated age is not more than the threshold age and face recognition cannot be performed by referring to the existing face database, the face image database is prompted to be updated. If FR is possible with reference to the existing face image database, the process for prompting the update of the face image database is not executed. In this case, the system control circuit 20 receives the FR result from the digital signal processing circuit 18 and cannot perform the FR (specifically, the face image database corresponding to the detected face portion does not exist and the tag is attached). The user is prompted to update the face image database.

  Furthermore, when prompting the update of the face image database in S106, if the estimated age is equal to or lower than the threshold age, it may be prompted according to whether the FR result is correct or not immediately. That is, when the estimated age is less than the threshold age and correct FR cannot be performed by referring to the existing face image database, the face image database is prompted to be updated. Whether or not the FR result is correct is determined by an input from the user. That is, the FR result is output by attaching a name tag to the vicinity of the detected face, but when the user determines that the tag is not correct, the fact is input from the operation switch SW19 of the digital camera. The system control circuit 20 prompts to update the face image database in response to an input from the user. As a result, it is possible to prevent a situation in which the system control circuit 20 prompts the update uniformly even though the user is updating the face image database. This is because it is assumed that the detected face can be correctly recognized when the user correctly updates the face image database.

  In FIG. 2, it is assumed that HBD and FD have been executed, but FD cannot be executed even if HBD can be executed, for example, because the subject person is wearing sunglasses or is not facing the front. There may be cases. In this case, the age can be estimated in S104 based on the HBD result, and the estimated age can be used for other processing, for example, setting of imaging conditions. Of course, even in this case, it is possible to promptly update the face image database. The adaptive setting of shooting conditions will be described later.

  FIG. 3 shows another processing flowchart of the present embodiment. First, a subject image is photographed (S201), and a human body is detected from the obtained image (S202).

  After detecting the human body from the captured image, the system control circuit 20 estimates the age of the person using the detected size of the human body (S203). Specifically, the ratio between the length of the human head and the length of the shoulder is calculated, and a memory in which a table that preliminarily defines the correspondence between the ratio and the age is accessed to correspond to the calculated ratio. Read age to do. The table that defines the correspondence between the ratio and the age classifies the age as, for example, 0-12 years, 12-15 years, 16 years or older, and defines an average ratio for each category.

After estimating the age of the person, the system control circuit 20 determines whether or not the estimated age is equal to or less than a predetermined threshold age (S204). The threshold age can be set to 12 years old, for example. If the estimated age is 12 years old or less, the shooting scene mode is set to the child mode (S205). When the photographing scene mode is set to the child mode, the system control circuit 20 adjusts the shutter speed and sensitivity so that the subject does not move even if the movement of the subject is relatively fast. On the other hand, if the estimated age is 13 years or older, the child mode is not set.

  When the estimated age is over 12 years old, it is further determined whether or not it is over 20 years old. If it is determined that the age is over 20, the shooting scene mode is set to the adult mode. Also good. In the adult mode, the system control circuit 20 highlights skin color, for example.

  FIG. 4 shows a process flowchart of human body detection, which is the process of S102 in FIG. 2 and S202 in FIG. First, the digital signal processing circuit 18 acquires an image by photographing (S301). Next, an edge is extracted from the obtained image (S302). In this edge extraction processing, the contour extraction result in the contour correction processing may be used as it is, or an edge may be extracted separately.

  After extracting the edge, the digital signal processing circuit 18 determines whether or not the extracted edge pattern matches the edge pattern of the upper body of a predetermined person (S303).

  The edge pattern of the upper body is stored in advance in the memory of the digital signal processing circuit 18 as a template. If the extracted edge pattern matches the edge pattern of the upper body, a human body is detected from the extracted edge (S304).

  FIG. 5 schematically shows processing for detecting a human body from the captured image 50. Assume that the human body 52 is reflected in the captured image 50. An arcuate edge 60 is present on the head of the human body. In addition, curved edges 62 and 64 also exist on the shoulders of the human body. These edges 60, 62, and 64 are stored in the memory as templates, and it is determined whether or not there is a pattern that matches the templates 60, 62, and 64 extracted from the captured image. Of course, since the sizes of the human body 52 in the photographed image 50 are various, it can be considered that the similar shapes match. Of course, a plurality of templates having different sizes may be prepared in advance. As described above, when both the face edge and the shoulder edge are detected, the human body can be detected from the subject. Alternatively, a combination 66 of a straight edge of the head and a straight edge of the shoulder may be prepared as a template.

  As described above, according to the present embodiment, the age of the human body obtained by human body detection (HBD) is estimated, and the imaging condition is adaptively set based on the estimated age. It can be applied effectively to obtain high-quality captured images.

  A technique for detecting the face of a subject and estimating the age of the subject from the detected face is known, but this embodiment can also be used as a technique for complementing such processing. In other words, the age of the subject is estimated from the detected face, and the age is estimated by the technique of the present embodiment. If both estimated ages match within an allowable range, the reliability of the age estimated from the detected face is estimated. Can be evaluated as high. In addition, when the age cannot be estimated from the detected face because the subject is not facing the front, it is also possible to compensate by estimating the age using the technique of the present embodiment.

  1 Digital Camera, 10 Lens, 12 Shutter / Aperture, 14 Image Sensor, 16 Analog Preprocessing Circuit, 18 Digital Signal Processing Circuit, 19 Operation Switch (SW), 20 System Control Circuit, 22 Data Bus, 24 Memory Card, 26 Liquid Crystal Monitor, 28 buffer memory.

Claims (7)

An imaging device comprising:
An optical system including a lens;
Imaging means for converting a subject image formed by the optical system into an electrical signal;
Detecting means for detecting a human body using an edge pattern of an image signal obtained by the imaging means;
Face detection means for detecting a face portion from the image signal obtained by the imaging means;
Storage means for storing data of one or more face images;
A face recognition means for recognizing the detected face by comparing the detected face with the data of the face image stored in the storage means;
Control means for estimating the age of the human body based on the size of the human body, and executing processing for updating the face image stored in the storage means based on the estimated age;
An imaging apparatus comprising: An imaging device comprising:
An optical system including a lens;
Imaging means for converting a subject image formed by the optical system into an electrical signal;
Detecting means for detecting a human body using an edge pattern of an image signal obtained by the imaging means;
Control means for estimating the age of the human body based on the size of the human body, and setting shooting conditions based on the estimated age;
An imaging apparatus comprising: In the imaging device according to any one of claims 1 and 2,
The image pickup apparatus, wherein the control means estimates the age by using a ratio of a head length to a shoulder length as the size of the human body. The imaging device according to claim 1,
The control device outputs a message that prompts the user to update the face image when the estimated age is equal to or lower than a threshold age. The imaging apparatus according to claim 4.
When the estimated age is less than or equal to the threshold age, the control means sets the update timing of the face image corresponding to the face recognized by the face recognition means, and has elapsed since being recognized by the face recognition means An image pickup apparatus that outputs a message prompting a user to update the face image when time reaches the update timing. The imaging apparatus according to claim 4.
The control means sends a message prompting the user to update the face image when the estimated age is less than or equal to a threshold age and the face recognition means is unable to recognize the face or is erroneously recognized. An imaging apparatus characterized by outputting. The imaging apparatus according to claim 2, wherein
The image pickup apparatus, wherein the control unit sets a shutter speed relatively fast when the estimated age is equal to or less than a threshold age.

Images