HK1087317B - Live finger detection by four-point measurement of complex impedance - Google Patents

Description

Live finger detection with four-point complex impedance measurement

Technical Field

The present invention relates to a sensor assembly and method for determining the condition of a structure by measuring a property in close proximity to the surface of the structure, in particular confirming whether a fingerprint is measured on a live finger.

Background

Capacitive or impedance based fingerprint sensors are some of the most promising approaches to provide low cost, miniaturized devices in biometric identification. For this reason, such sensors become possible objects to be integrated in mobile phones and the like.

To enhance the trustworthiness of the fingerprint sensor, it is very important that any attempt to fool the system with a fake finger should be detected and rejected. A fake finger typically comprises a sheet of material having finger-like electrical properties and a fingerprint is engraved or molded on the surface of the sheet. In more extreme cases, it is conceivable to use a dead, cut finger.

Importantly, for live finger detection systems, both the probability of accepting a fake finger (false acceptance ratio-FAR) and the probability of rejecting a genuine finger (false rejection ratio-FRR) are extremely low. This is important to develop a method of identifying the actual nature and unique properties of a living finger that cannot be readily replicated by synthetic materials or by other biological substances that are not living tissues, and which are characteristic of most fingers in the human population.

For some types of impedance measurements of the properties of a finger, fingerprint sensors based on low-cost capacitance are desirable, since such sensors can most often be integrated directly on the device, using existing measurement structures, or adding several special electrodes.

Several different types of fingerprint sensors have recently been developed, from the two-dimensional matrix sensor described in US 5,953,441, via a sensor array for reconstructing a fingerprint image from a series of semi-overlapping partial images as described in US 6,289,114, to the line sensor described in EP 0988614, which scans the surface of a finger and uses the measured finger velocity to reconstruct an image of the finger.

Attempts to detect live fingers include both blood oxygenation and blood pulse measurements. However, because blood circulation in the finger is virtually non-existent in very cold fingers, these methods are far from as easy as "waterproof". Either of these principles is not easily implemented on low cost devices.

U.S. Pat. Nos. 6,175,641 and 5,953,441 and patent application US2001/0005424A1 all describe methods for investigating whether an object placed on a fingerprint sensor is a live finger based on different impedances.

Us patent 6,175,641, which relates to impedance sensing on an optical matrix sensor, describes two different methods of measuring the electrical properties of a finger: first, the dielectric constant is measured locally by applying an AC signal between two very small separated electrode comb structures on the sensor surface. The patent states that this measurement can distinguish living tissue (high dielectric constant) from commercial plastics (low dielectric constant).

Second, the sensor has electrodes for determining finger impedance, called dual-point electrodes, which presumably will give additional information that can be used to distinguish real fingers from fake ones. The patent also states that the use of several frequencies can increase the reliability of the measurement.

However, the method described in this patent has a number of disadvantages. While dielectric measurements may work well for dry fingers, for sweaty or wet fingers, the very small separated electrode comb structure is likely to be shorted by saline sweat, and no useful information is available.

In addition, the impedance of a live finger, as measured by a two-point system, may vary by at least an order of magnitude with the humidity of the finger. Therefore, it is difficult to identify a finger using its impedance as a criterion, and both the magnitude and phase of the impedance, and the variation with frequency, may be fooled by simple materials known in everyday life, such as peeled potatoes.

Patent 5,953,441 describes how to detect spoofing of an AC capacitive fingerprint sensor containing a matrix of capacitive sensing cells. The main idea here for live finger detection is to send an AC signal through electrodes around the edge of the sensor area and detect the phase of the signal on the sensor unit, which phase characterizes the live finger.

However, this method, while excluding many different fake finger materials, is also relatively easy to find materials with approximately the same phase as the finger, thereby fooling the system.

Patent application US2001/0005424A1 describes a process similar to that described in 6,175,641. The impedance of the finger (either between two electrodes or between one electrode and "infinity") is measured as a function of frequency. By comparing the curve with a reference curve, the living body characteristic of the finger can then be detected. However, this method adds only a small point to the above method. The absolute impedance and frequency response between different fingers and between different states of the same finger (e.g., versus humidity) are so different that the "live finger criteria" has a wide range and therefore the principle is easily fooled.

International patent application PCT/NO03/00157(WO03094724), incorporated herein by reference, describes another live finger detection principle based on a four-point measurement of complex impedance. The patent passes a current or applies a voltage between two electrodes while measuring the voltage drop between two other electrodes, all of which are in contact with the finger surface. The four-point principle is applied to finger impedance measurements, as can be seen in fig. 1. An AC current is sent through the finger via the outer electrode while the voltage drop is measured with a differential amplifier between the two inner electrodes.

Disclosure of Invention

The object of the present invention is to ensure a four-point measurement system for live finger detection that can be used to compensate for differences in finger characteristics between people, such as varying stratum corneum thickness.

To achieve this object, the present invention includes an impedance measurement system having an array of at least four electrodes. The electrodes may be in direct contact with the finger or capacitively coupled to the finger through an insulating layer. The capacitors are arranged such that they can be used in at least two different four-point electrode configurations corresponding to different relative placements between the current and voltage sensing electrodes.

The use of a four-point technique enables this technique to eliminate the series impedance of the stratum corneum and thereby directly measure the impedance inside the finger. The impedance of the stratum corneum is strongly related to the moisture content of the skin and to environmental conditions such as temperature. Due to skin moisture and environmental conditions, it is difficult to identify "narrow" criteria that can be used to separate real and fake fingers. In contrast, the humidity inside the finger remains substantially constant under varying environmental conditions. Thus, the impedance (living skin and tissue) inside the finger is much more constant and more reproducible from person to person.

Thus, the four-point principle makes it easier to obtain a "narrow" criterion that can be used to identify whether a finger is a real live finger.

Because of the layered structure of the skin, the four-point principle also gives an inherent "depth selectivity": by increasing the frequency, deeper parts of the living skin are made available for measurement. Thus, a depth-specific change can be measured in electrical properties by just frequency scanning.

The living tissue inside the finger also has a unique characteristic dispersion (variation of electrical properties with frequency) that can be used to identify a real finger with a high degree of reliability. These properties change after death or when the finger is cut from the hand, and it can also be determined whether the finger is alive.

A disadvantage of the principle proposed in PCT/NO03/00157(WO03094724) is that the measurement of the four-point impedance is only carried out with a set of electrodes, all of which have a fixed distance to each other.

The impedance measurement is in any case more or less influenced by the stratum corneum (horny layer) in relation to the relative position of the electrodes in the four-point structure.

In the very short distance limit between the current electrodes, the current will not penetrate into the living skin inside the finger at all, thus giving only a single stratum corneum measurement.

In another margin for large electrode distances, the measurement will be largely determined by the properties of the living tissue inside the finger.

Since different people have different stratum corneum thicknesses, a fixed electrode distance will give different results for different people, thus making it difficult to identify a live finger without using very wide criteria for fixed electrode measurements. But if the criteria are not very narrow, the principle is more vulnerable.

It is well known to the skilled engineer that different electrode configurations, i.e. different relative placements of the electrodes, correspond to measurements of different parts of the object in proximity to the electrodes. However, the portion of the finger that is measured by the four-point principle is determined not only by the distance between the two current sensing electrodes, and between the two voltage sensing electrodes, but also by the geometric layout of the electrodes and the relative placement of the voltage sensing electrodes with respect to the current electrodes.

Switching between different four-point electrode configurations is possible by activating different electrodes in the array, or by swapping the roles of the electrodes being used (voltage sensing or current).

By switching between a number of different electrode configurations within the array, measurements can be made on various parts of the finger, e.g. corresponding to different depths, so that variations in e.g. the thickness of the stratum corneum can be compensated for.

For example, the characteristic dispersion (complex impedance versus frequency) observed for one person using one electrode arrangement can be detected for another person using another arrangement.

To reveal the same dispersion, a person with a very thick stratum corneum may, for example, have a greater requirement for the distance between the current injection electrodes or the voltage sensing electrodes than a person with a thin stratum corneum.

The minimum criterion for accepting an object as a live finger may be that at least one characteristic impedance-related phenomenon is detected using at least one electrode arrangement.

It should be emphasized that the emphasis on "difference in stratum corneum thickness" is merely exemplary. The present principles apply to all properties of the finger, where a change in the electrode geometry helps to reveal certain impedance-related phenomena in a given frequency range.

Preferably, the number of possible electrode configurations may be more than two, for example 3-5, in order to enhance the reliability of detecting a live finger in a dominant population. Of course, the various configurations can be arranged in a separate array, which is advantageous in minimizing the required translation.

According to one aspect of the present invention there is provided a sensor assembly for determining a condition of a structure by measuring a property against a surface of the structure, in particular for confirming whether a fingerprint is measured on a live finger, the sensor assembly comprising: a current source; at least four electrodes located at selected positions relative to each other, said positions providing at least two relative distances between said electrodes, wherein a selected first pair of electrodes among said at least four electrodes constitutes a current supply electrode, a selected second pair of electrodes among said at least four electrodes constitutes a pickup electrode, and at least one electrode among a selected second pair of electrodes among said at least four electrodes does not constitute a current supply electrode; a measuring device coupled to said at least four electrodes for measuring the impedance between said selected pair of pick-up electrodes to provide a value indicative of the structure, storage means for storing a predetermined set of values indicative of a selected condition of the structure; and the sensor assembly further comprises: computing means for comparing the characteristic value from each of said at least one pair of pick-up electrodes with said set of predetermined values to detect whether said structure is within a certain condition, and said sensor assembly is adapted to alternately couple at least one current supply electrode and said measuring device to different pairs of electrodes having different distances between them to measure the characteristic value at different depths in said structure.

According to another aspect of the present invention there is provided a method of using at least four electrodes coupled to a surface of a structure and characterizing conditions of the structure in close proximity to the surface of the structure, such as electrical properties of two surface portions of the skin, namely the stratum corneum and the vital skin, in at least two different distances between the electrodes, wherein a selected first pair of electrodes among said at least four electrodes constitutes current supply electrodes and a selected second pair of electrodes among said at least four electrodes constitutes pickup electrodes, the method comprising the steps of: applying a current or voltage to the skin between said first pair of current-supplying electrodes; measuring the impedance between said second pair of pick-up electrodes, at least one of which is not a current-supplying electrode, and calculating an electrical characteristic associated with that electrode; successively varying the effect of the electrodes to apply a current between said second pair of current-supply electrodes and to measure at least two different distances between the pick-up electrode and/or the current-supply electrode, respectively; comparing the measured impedance value with a set of predetermined values characterizing at least one condition of the structure; and determining the condition of the structure based on the comparison of said measured impedance value with said set of predetermined values.

Drawings

The invention will now be described with reference to the accompanying drawings, which illustrate the invention by way of example.

Figure 1 shows an electrically equivalent two-layer structure interrogated with a sensor assembly of the present invention.

Fig. 2 shows one possible configuration for switching the effect of the multiple electrodes of the present invention.

Fig. 3 shows a multivariate model for distinguishing live fingers from other objects. It is noted that if only one variable (either variable 1 or variable 2) is used, some "fake fingers" (marked in red) are likely to be mistaken for real fingers.

Fig. 4 shows the measured impedance absolute values as a function of frequency for a first electrode configuration.

Fig. 5 shows the measured impedance phase as a function of frequency for a first electrode configuration.

Fig. 6 shows the measured impedance absolute values as a function of frequency for a second electrode configuration.

Fig. 7 shows the measured impedance phase as a function of frequency for a second electrode configuration.

FIG. 8 shows real and imaginary parts of measured four-point impedance with frequency as a parameter for a plurality of live fingers.

Detailed Description

In fig. 1, a finger surface 11 is placed over a plurality of sensors 10. The finger structure comprises two layers: the stratum corneum 12 of a living finger and living tissue 13. The stratum corneum (horny layer)12 constitutes impedances Z1, Z2, Z3, and Z4, respectively, on each of the four illustrated electrodes 10, while the living tissue exhibits an impedance Z0.

In a practical arrangement, the four-point measurement may be performed by an array of electrodes on the sensor surface, such as those specified in thin film, thick film, or printed circuit boards. These electrodes can provide galvanic contact with the finger, and can also provide purely capacitive coupling from the electrode to the finger with a thin dielectric passivation.

Typical size of the individual electrodes (both current and voltage electrodes) is 0.5-5mm²Typical minimum electrode spacing is 0.3-2 mm.

Figure 2 shows an example of how an 8 electrode array can be arranged to enable measurements at a number of different galvanic electrode distances. In the illustrated configuration, voltage measurements are always made between electrodes 4 and 5, while different combinations of electrodes 1, 2, 3(AC source) and electrodes 6,7, and 8(AC drain or ground) are implemented with two switches S1 and S2 to vary the distance of the current electrodes. In addition, the roles of the current and voltage sensing electrodes may be interchanged so that current is always sent between the innermost electrodes, while voltage measurements are switched between different combinations of the remaining electrodes. It is well known in the art that if a pair of voltage sensing is exchanged with a pair of current electrodes, the measured impedance remains substantially the same.

By selecting the range of distances between the electrodes to correspond to the variation in stratum corneum thickness of the population (or other variation giving a corresponding effect, such as a difference in humidity), a more direct comparison can be obtained, thereby "narrowing" the criterion for a real finger. This allows for a higher degree of certainty of identification of a live finger than any of the above methods.

In designing a readout system it is important to maximize the input impedance of the voltage sensing branch, since too low an impedance will result in parasitic input currents affecting the measurement principle. To minimize this effect of the input impedance, an amplifying coupling as described in US 4,956,729 may be employed. The input current of the voltage attenuator can also be minimized ("active shielding") by shielding the input attenuator (pad) (and the traces connecting it to the amplifier) using electrodes having the same voltage as the input attenuator itself. Such voltages are easily obtained by simple voltage follower stages, in which the input voltage is fed back to the shielding electrode.

If the input impedance of the amplifier is sufficiently high (or equivalently if the input current is sufficiently low), the detected voltage will not be affected by the impedances Z2, Z3, Z1, and Z4 through the stratum corneum, but only by the impedance Z0 characterizing the interior of the finger.

It should be emphasized that the system disclosed in the figures is only one possible way of arranging the electrodes. In principle, all electrode arrangements that produce two or more different electrode configurations can be used. The sensing of the voltage or impedance may be with the fingerprint sensor unit itself.

Upon detection of a live finger, four-point complex impedance measurements may be obtained for each electrode arrangement at a single frequency or over a range of frequencies. For example, these properties may be measured continuously with a certain frequency or a number of different dispersion frequencies. The frequency spacing is preferably on the order of 1-3 of the amplitude.

The reactance X0 and resistance R0 of the complex impedance Z0 ═ R0+ jX0 can be determined for each frequency by measuring the current through the finger and the differential voltage at least two different times in a signal cycle, although other techniques for detecting the complex impedance component are possible.

The live finger data is preferably recorded just before, just after, or more preferably during the fingerprint image capture process. This makes it difficult to fool the system first with a real finger and then with a fake finger with the correct fingerprint pattern. In some systems live finger detection and fingerprint imaging cannot be performed at the same time due to conflicting signals. In this case, the fingerprint imaging may be paused for a short period of time and live finger detection may be performed in that time frame. It is important that the time for live finger detection is short enough not to significantly affect the quality of the image.

For the scanning type sensor described in EP 0988614, live finger detection can be accomplished as follows: for example, one or two image data lines are skipped and a live finger is detected during that time. In the above-mentioned application, the solution comprises a plurality of sensor units for measuring the impedance between the excitation electrode and the sensor units. According to the invention, the effect of the sensor unit can be changed during one measuring cycle or a few measuring cycles of the measuring finger. The live finger detection mode is not visible in the resulting fingerprint image because the scheme described in the above application allows oversampling and rejection of unnecessary data.

In this case, it is also important that the geometric area used for live finger detection overlaps the area used for fingerprint imaging in order to convince the person that the detected live finger is indeed the same as the object being imaged.

As previously mentioned, the criterion for accepting an object as a live finger may be based on the measurement of at least one impedance parameter associated with at least one electrode configuration.

The parameter may for example be a value or a combination of values related to the measured impedance, such as phase, amplitude, resistance or reactance, or may be a variation of some value with frequency. The parameter may also be some derived value, such as the frequency with which a certain parameter obtains a certain value.

In a preferred embodiment, at least one of these parameters is related to the observed phase change in the measured 4-point finger impedance, which occurs over a frequency range of between about 10kHz and 1 MHz. In this frequency range, the phase of the impedance may be considered to have undergone a phase shift of 50-90 degrees as the dominant portion of the impedance changes from capacitive to resistive.

The change in impedance from predominantly capacitive to resistive can also be viewed as a change in the frequency derivative of the impedance magnitude, i.e., the frequency derivative changes from negative to near zero as the frequency passes through a typical frequency.

Fig. 4-7 show impedance magnitude and phase measurements for a plurality of live fingers of different persons at two different electrode configurations. Fig. 4 and 5 show one electrode configuration, while fig. 6 and 7 show another electrode configuration.

In both configurations a large positive phase shift is observed and the impedance curve is also observed to be substantially flat at the frequency of the phase transition.

No similar phenomenon was observed for any other substance we tested. Such a large phase shift is not observed when measuring the finger with only two-point impedance measurements. This can therefore be used as a criterion for determining the validity of the finger.

The nature of this frequency change, e.g. its amplitude and its transition frequency, can be characterized in many different ways. For example, the measured complex impedance may be plotted as a function of frequency, i.e. phase versus amplitude versus frequency, or imaginary versus real with frequency as a parameter, see fig. 8. The imaginary and real parts Z of the measured impedance are shown in fig. 8_IAnd Z_RAnd indicates a possible analysis for identifying a live finger. As described above, the slope, center of gravity, or of these curvesLength is some possible parameter for identifying a live finger. In a preferred embodiment, the slope is used as a basis for live finger confirmation.

The advantage of the latter approach is that the curves look similar for different fingers even if the transition point occurs at a very different frequency.

Certain characteristics of these curves, such as the derivative of the curve, the length, "center of mass", the characteristic transition frequency, or the frequency when the curve approaches a certain value, can then be derived by the automatic calculation unit and used as an identification parameter for a live finger. It is well known to the skilled engineer that the same measured property may be represented in a number of mathematically equivalent ways.

It was also observed that the typical frequency of movement changes as the distance between the electrodes increases. This is because a greater distance generally gives a deeper measurement depth in the finger, and the electrical properties of the finger change with depth. This can be clearly seen by comparing the curves of fig. 4 and 5, which correspond to short distances, with those of fig. 6 and 7 (longer electrode distances). The typical transition frequency in fig. 4 and 5 is much lower than in fig. 6 and 7.

The measured transition frequency as a function of the electrode distance is an important property of a live finger and can be represented parametrically, and this parameter or these parameters can be used to improve the identification model of the live finger.

As can be seen from fig. 4-7, the actual transition frequency varies from person to person. This may be due to changes in humidity or stratum corneum thickness and may be corrected by measurements over a larger frequency interval, or by measurements over several different electrode distances. It can be seen from the graph that increasing the frequency has almost the same effect on the phase as increasing the electrode distance. Near the transition frequency, an increase in frequency or electrode distance will generally increase the phase of the live finger. This characteristic relationship between electrode distance and phase can be modeled mathematically and used as yet another criterion for identifying a live finger.

The criterion for accepting a live finger is preferably based on the measurement of more than one parameter. A set of related parameters or variables can be found by, for example, feeding the obtained impedance to a multivariate model (as shown in fig. 3). According to a preferred embodiment, the set of parameters is the impedance data shown in FIG. 8.

By performing a statistical analysis of the data measured by the live and fake fingers, the model will output a set of weighted, combined variables (typically two or three) that are optimized for distinguishing between real and fake fingers. Thus, the model includes selected deviation limits within the data set sufficient to distinguish between real, dead, and fake fingers.

The variables are preferably normalized by any available method to avoid the effects of changing sensor characteristics, etc., and the variables are preferably statistically independent.

The electrode configuration used to obtain the desired variables is preferably determined by the signal processing system based on measurements of several electrode configurations. This electrode configuration can be obtained with several different electrodes, for example as described above in the fingerprint scanner, but with other systems comprising a number of electrodes for making measurements on the skin.

One possible alternative to the variation of the sensor combinations discussed above is to measure one configuration at a time until the measurement meets a given criterion, and then infer or switch to a second configuration. It is also possible to combine measurements obtained from different electrode configurations.

Two-point impedance data or other measurements of the finger (e.g., temperature) may be used in combination with the four-point data to enhance selectivity for fake or dead fingers. Only objects where all unique variables fall within certain limits are considered live fingers. Other objects will be rejected.

Figure 3 schematically depicts a model with two variables, where only objects falling within the indicated elliptical areas (the data obtained are represented by triangles) can be considered live. Circles outside the ellipse correspond to rejected objects.

In summary, the preferred method requires not only a specific value, but also a set of variables within certain limits, which makes it extremely difficult to construct a "fake finger" material. On the other hand, a dead finger will be rejected due to the biological processes occurring after finger death changing the electrical parameters.

Claims (10)

1. A sensor assembly for determining a condition of a structure to confirm whether the structure is comprised of living tissue of skin by measuring a property value near a surface of the structure, the sensor assembly comprising:

a current source;

at least four electrodes located at selected positions relative to each other, said positions providing at least two relative distances between said electrodes, wherein at least a first pair of said at least four electrodes is coupled to said current source, thereby constituting current supply electrodes for supplying current to said structure, and at least a second pair of said at least four electrodes constitutes pickup electrodes and is coupled to a measuring device for measuring a characteristic value characterized by an impedance of said pickup electrodes,

storage means for storing a set of predetermined values characterizing a selected condition of the structure; and

computing means for comparing said property values from said second pair of electrodes with said set of predetermined values to detect whether said structure is within a certain condition,

the method is characterized in that:

said sensor assembly is provided with control means for at least locally interchanging the roles of said electrodes by coupling different pairs of said at least four electrodes to said current source and said measuring device and by changing the relative positions of said electrode pairs for measuring characteristic values with different pairs of pick-up electrodes and current supply electrodes.

2. The sensor assembly of claim 1, wherein said supply current oscillates over a selected frequency range.

3. The sensor assembly of claim 2, further comprising: measuring means for measuring the complex impedance of each pick-up electrode and wherein said calculating means comprises: comparing means for comparing the imaginary and real parts of the complex impedance signal as a function of applied frequency by determining the slope of a curve drawn by said complex impedance as a function of frequency, and comparing the slope with a predetermined set of slopes indicative of a live finger.

4. The sensor assembly of claim 1, wherein the distance between the first pair of supply electrodes and the first pair of pickup electrodes is less than 1 mm.

5. A sensor assembly according to claim 1, wherein the control means causes the roles of the pick-up electrode and the current supply electrode to be successively interchanged so as to change the relative position between the sensors and thereby change the value of the measured characteristic of the surface.

6. The sensor assembly of claim 5, further comprising: measuring means for measuring the phase of the signal at each pick-up electrode, and wherein said calculating means comprises: comparing means for comparing the distance between said pick-up electrode and said current supply electrode at a selected frequency corresponding to the phase of the signal and comparing these parameters with a predetermined set of parameters indicative of a live finger.

7. A sensor assembly according to claim 1, wherein the pick-up electrode is formed by a sensor element of a fingerprint sensor array.

8. A method of characterizing electrical properties of stratum corneum and vital skin proximate a surface of a structure using at least four electrodes coupled to the surface of the structure and at least two different distances between the electrodes, wherein a selected first pair of electrodes among said at least four electrodes constitutes current supply electrodes and a selected second pair of electrodes among said at least four electrodes constitutes pickup electrodes, the method comprising the steps of:

applying a current or voltage to the skin between said first pair of current-supplying electrodes;

measuring the impedance between said second pair of pick-up electrodes, at least one of which is not a current-supplying electrode, and calculating an electrical characteristic associated with that electrode;

successively varying the effect of the electrodes to apply a current between said first pair of current-supply electrodes and to take measurements at least two different distances between the pick-up electrode and/or the current-supply electrode, respectively;

comparing the measured impedance value with a set of predetermined values characterizing at least one condition of the structure; and

determining the condition of the structure based on the comparison of said measured impedance value with said set of predetermined values.

9. The method of claim 8, wherein the step of applying a current or voltage between two electrodes comprises: a signal of varying frequency is applied and a complex impedance is measured between one of the current-supplying electrodes and at least two pick-up electrodes positioned at a selected distance from the current-supplying electrode.

10. The method of claim 9 wherein the comparison of the measured impedances is performed by determining the slope of a curve drawn by the complex impedance as a function of frequency, the curve describing the relationship between the imaginary and real parts of the measured impedance signal as a function of applied frequency, comparing the determined slope with a predetermined set of values characterizing a live finger.