CONTACTLESS BIOMETRIC SENSOR AND ACQUISITION METHOD

Abstract

The invention relates to a contactless biometric sensor (1), comprising an upper surface (3) above which, in use, a part of the human body bearing dermatoglyphs is positioned, a light source configured to illuminate said part of the human body through the upper surface, and an imager configured to acquire an image (I) of said part of the human body through the upper surface. The upper surface of the sensor is provided with a mask of spatially variable light transmissivity. Figure for the abstract: Figure 1

Description

Title of the invention: CONTACTLESS BIOMETRIC SENSOR AND ACQUISITION METHOD technical field

[0001] The field of the invention is that of devices and methods for acquiring and processing biometric images of a human body part exhibiting dermatoglyphs. More particularly, the invention relates to the non-contact acquisition of such an image by positioning the human body part above a sensor. Previous technique

[0002] The MorphoWve™ biometric reader is a 3D fingerprint reader that can be used to control access to a secure area by means of a single, contactless wave of the hand over the reader. This reader, which is based on patented technology, notably through patent EP 3 312 771 Bl, uses a uniform lighting system in one color and a projection system of a target consisting of repetitive light patterns in at least one other color. Description of the invention

[0003] One objective of the invention is to provide a biometric reader that is less complex in design, in particular to reduce its cost, without degrading its performance.

[0004] To this end, the invention proposes, according to a first aspect, a contactless biometric sensor, comprising:

[0005] - an upper surface above which, in use, a part of the human body bearing dermatoglyphs is positioned; - a light source configured to illuminate said part of the human body through its upper surface; and - an imager configured to acquire an image of said part of the human body through the upper surface.

[0006] The upper surface is equipped with a mask with spatially variable light transmissivity. In use, the mask projects a pattern, referred to as the illumination projection, of a target exhibiting spatial modulation of light intensity onto the said part of the human body. The mask also projects a pattern, referred to as the imaging projection, of said target into the image acquired by the imager.

[0007] Some preferred but not limiting aspects of this sensor are as follows:

[0008] - the mask consists of the upper face on which an opacity pattern is produced; - the mask consists of a transparent support on which an opacity pattern is created, said support being attached to the upper face; - the spatially variable light transmissivity of the mask exhibits a periodic modulation along a principal direction; - Periodic modulation is either sinusoidal modulation or triangle modulation.

[0009] The invention also relates to a biometric measurement device comprising a sensor according to the first aspect of the invention and an image processing unit for the image acquired by the imager. The processing unit is configured to compensate for the image projection of said target in the image acquired by the imager, in order to provide a compensated image corresponding to the light reflected by said part of the human body subjected to the illumination projection.

[0010] The processing unit can further be configured to compensate for the projection of lighting from said target in the compensated image, in order to provide an image of the der-matoglyphs.

[0011] The processing unit can further be configured to deduce from the image compensated a depth map representative of the geometry of the dermatoglyphs.

[0012] The processing unit can further be configured to straighten the image of the dermatoglyphs using the depth map, in order to provide a flattened image of the dermatoglyphs.

[0013] The processing unit can further be configured to implement a biometric identity recognition method based on the image of the dermatoglyphs or on the flattened image of the dermatoglyphs.

[0014] The processing unit can further be configured to deduce from the compensated image an amplitude image of the modulation introduced by the lighting projection.

[0015] The processing unit can further be configured to implement, on the basis of said amplitude image, a fraud detection method aimed at authenticating that the part of the human body is an authentic part of a human body. Brief description of the drawings

[0016] Other aspects, objectives, advantages and features of the invention will become clearer upon reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the accompanying drawings in which:

[0017] [Fig-1] [Fig.1] is a diagram of a contactless biometric sensor according to a mode of possible realization of the invention;

[0018] [Fig.2] [Fig.2] is a diagram schematically illustrating steps put in work during the execution of a biometric measurement process according to a possible embodiment of the invention.

[0019] DETAILED DESCRIPTION OF SPECIFIC EMBODIMENTS

[0020] The invention relates to a contactless biometric sensor, an example of which is illustrated in [Fig. 1]. In this example, the sensor 1 comprises a console 2 having an upper face 3 above which, during use, a part of the human body bearing dermatoglyphs is positioned. The positioning of the human body part can be carried out dynamically by passing said body part over the upper face 3 in a sweeping motion (motional acquisition) or, conversely, statically above the upper face 3 (static acquisition).

[0021] The upper face can optionally be surmounted by a cap 4. The upper face and the cap 4 are separated by a distance of a few centimeters to define between them an acquisition zone, here open frontally and laterally, forming a passage for a user's hand.

[0022] The cap 4 may incorporate a screen 5, for example a touchscreen, intended in particular to indicate to the user how to position their hand. The cap 4 may also include a chasing light 6 consisting of light-emitting diodes arranged in a line and programmed to illuminate the diodes successively at a desired speed for the movement of the hand.

[0023] The cap 4 can also integrate a contactless access card reader 7 enabling multi-factor authentication.

[0024] The console 2 contains a light source configured to illuminate said part of the human body through the upper face 3 and an imager configured to acquire an image of said part of the human body through the upper face 3. The imager is generally placed vertically above the upper face 3 while the light source is inclined relative to the imager so that the direction of illumination and the direction of imaging form an angle between them preferably greater than 10°, even more preferably greater than 20°.

[0025] The control panel 2 may also include a sensor 8, for example an infrared barrier, enabling the detection of the entry of the hand into the acquisition area and the activation of the lighting source and the imager.

[0026] The light source is configured to provide uniform illumination of the upper surface 3 in a single color, visible or invisible (for example, in the infrared). The light source may thus comprise a single light-emitting diode (LED) associated with an optical component used to control the divergence of the light beam and direct it towards the acquisition area. Thus, unlike the light source of patent EP 3 312 771 Bl, the light source of the sensor according to the invention uses a It uses only one color of illumination and does not incorporate a test pattern projection system. The cost of the sensor is therefore reduced, while maintaining its performance since the amount of light collected by the imager is only slightly altered due to absorption by the mask.

[0027] According to the invention, the upper face 3 of the sensor 1 is provided with a mask with spatially variable light transmissivity. Since this mask is present in the illumination channel, during operation it projects, referred to as the illumination projection, a target exhibiting spatial modulation of light intensity onto the part of the human body within the acquisition zone. Furthermore, this mask is also present in the imaging channel (the imager views the part of the human body through the mask) and thus also projects, referred to as the imaging projection, the target onto the image acquired by the imager.

[0028] The imager can be a color camera or, preferably, a black and white camera which is more sensitive than a color camera and therefore able to compensate for the alteration related to the absorption of light by the mask.

[0029] In one possible embodiment, the sensor includes, at the periphery of the mask, a marker whose position can be controlled in each image acquired by the imager in order to compensate for possible material drifts, such as thermal expansion. This ensures that the image projection correction described below is always optimal. Optionally, the mask position can be used to compensate for distance information based on drift learning. In the presence of a visor 4, the screen 5 can be used to display a control image of the mask, thus allowing for the detection and correction of any mechanical drift.

[0030] The upper face 3 is typically transparent, and the mask may consist of the upper face 3 on which an opacity pattern is produced. Alternatively, the mask may consist of a transparent substrate on which an opacity pattern is produced, said substrate being attached to the upper face. By way of example, the opacity pattern may be produced with a halftone opaque element (for example, chrome on glass or black screen-printed or inkjet-printed ink) or with an element of variable density (gelatin, thermal sublimation ink, for example).

[0031] The opacity pattern exhibits spatially variable light transmittance such that in the most transparent areas, more than 90% of the incident light is preferably transmitted, while in the least transparent areas, at least 30% (more preferably at least 50% and even more preferably at least 70%) of the incident light is preferably transmitted. In particular, the opacity pattern does not have any completely opaque areas so as not to occlude the imaging system. The modulation ratio (ratio between maximum and The minimum transmission of the opacity pattern can be less than 50%, or even less than 30%. Minimum and maximum transmission are considered over a surface corresponding to the optical resolution of the imaging system at the mask level, provided the opacity pattern does not cause a completely black spot in the image acquired by the imager.

[0032] In a preferred embodiment, the opacity pattern exhibits a periodic modulation of its light transmissivity along a principal direction. In particular, the opacity pattern may be invariant along a first direction x (for example, along a direction of hand movement in the acquisition zone) and correspond to a periodic function along a second direction y orthogonal to the first direction x. The periodic modulation may, in particular, be sinusoidal or triangular modulation.

[0033] The periodic modulation of the light transmissivity of the mask can thus be, for example, of the form / x , / 2jr.y , ,), with T(y) the transmission Depending on y, a and b are two constants such that a - b > 0.3 (preferably greater than 0.5, even more preferably greater than 0.7) and a + b ≤ 1, p is the period, and a constant. The mask period is preferably chosen to be much larger than that of the dermatoglyphs (350 to 700 pm), for example between 3 and 5 mm.

[0034] To simplify the recognition of the target and eliminate the risk of confusion between two periods, an amplitude modulation b(y) can be added (the coefficient b being for example different for each period), a frequency modulation p(y) and / or a phase modulation 0(y), with p(y) and / or ÿ(y) a function of y (either a continuous function, or a constant function on each period and discontinuous between periods).

[0035] For reasons of manufacturing simplicity, a triangle function can, for example, be used instead of a sine function. A square wave function, on the other hand, should be avoided because it exhibits too much energy at higher frequencies and can interfere with the extraction of dermatoglyphs.

[0036] The invention also relates to a biometric measurement device comprising a contactless sensor as described above and an image processing unit for the image acquired by the imager. The processing unit may be integrated directly into the sensor or be external to the sensor. It may be embedded within the control panel or take the form of a remote unit connected to the sensor by a communication network, for example, wirelessly. It may also be composed of different parts, and different parts of the method described below may be implemented by different parts of the processing unit, which may be geographically separated and communicate with each other via a communication network.

[0037] With reference to [Fig.2], following the acquisition during an ACQ step by the imager of an image I, the processing unit is configured to compensate (during a COMP-PI step) the imaging projection of the target in the image acquired by the imager, in order to provide a compensated image corresponding to the light reflected by said part of the human body subjected to the lighting projection. This COMP-PI step makes it possible to correct the fact that the target is present in the imaging channel and from there to apply from the compensated image classic biometric image processing, such as for example those presented in patent EP 3 312 771 Bl, in application EP 3 726 423 Al or in the article "Complete 2.5D fingerprint reconstruction system: a feasibility study" by Laurent Condat and Vincent Roullier (RFIA 2012 - Pattern Recognition and Artificial Intelligence, Jan 2012, Lyon, France, pp.978-2-9539515-2-3. hal-00656576).

[0038] Considering (u,v) the coordinates of a pixel in the acquired image and I(u,v) the intensity measured at a pixel of the acquired image, this COMP-PI compensation can include the application of a predetermined correction function such that the intensity at a pixel of the compensated image Ic(u,v) can be expressed as Ic(u,v) = (I(u,v) - offset(y))*G / T(y) with G a constant, T(y) the transmissivity of the target at a point y corresponding to pixel (u,v), and offset(y) a shift function that can optionally be introduced to take into account the scattering of light by the mask.

[0039] In one possible embodiment, the correction function can be predetermined by knowing the opacity pattern and its projection in the acquired image. Alternatively, the correction function can be predetermined during a calibration phase using a calibration target of uniform reflectance placed directly against the mask on the upper surface of the sensor. For example, it is possible to acquire a first image with a uniformly light calibration target and a second image with a uniformly dark calibration target, which allows the calculation of gain coefficients g(u,v) and offset coefficients o(u,v) such that, after correction of the image projection, each of the first and second images is corrected. Images of calibration targets are aligned with the lighting projection:

[0040] I (u, v) - O (U, V) T (u, v)

[0041] This second approach has the advantage of compensating for all other defects (vignetting for example) related to the optical system (uniformity of lighting in particular) and / or to the making of the mask.

[0042] Still with reference to [Fig.2], the processing unit can also be configured to compensate during a COMP-PE step the projection of illumination of said target in the compensated image, in order to provide an image of the dermatoglyphs Id. Such compensation can for example be achieved by means of a band-stop filter, typically a filter centered around the spatial frequency of the target (for example 3mm) and whose bandwidth is adapted so that the filter does not alter the dermatoglyphs (for example a bandwidth of the order of 1mm).

[0043] The processing unit can also be configured to, during a Z-MAP step, derive a depth map representative of the dermatoglyphic geometry from the compensated image. To do this, each point of the compensated image can be mapped to the y-coordinate of the opacity pattern, which allows, through triangulation, the calculation of the distance between the imager and the part of the human body positioned within the acquisition area. The depth map can be established by phase calculation, for example in the Fourier plane, by Gabor filters, or by means of solutions based on deep learning. When the opacity pattern is periodic, the mapping is accurate to within one period. This ambiguity can nevertheless be resolved by exploiting geometric considerations (for example, a priori on the distance, the acquisition area being bounded between the upper face 3 and the cap 4).It is also possible to use a non-periodic opacity pattern (for example, a sinusoid with added amplitude and / or phase modulation) or to exploit a single reference point (an edge of the pattern, for example). Furthermore, it is possible to implement a technique based on the apparent frequency of the opacity pattern (when the illumination and the optical center of the imager are not equidistant from the plane containing the pattern) or on demodulation (the pattern becoming increasingly blurred the further one moves away from it). The width of the finger or a secondary distance sensor can also be used. Finally, an arbitrary choice can be made, resulting in approximately a simple scaling error on the finger that can be tolerated by the comparison algorithms.

[0044] As illustrated in [Fig. 2], the processing unit can further be configured to, during a FLAT step, straighten the image of the dermatoglyphs Id using the depth map, in order to provide a flattened image of the dermatoglyphs. This FLAT step can be carried out in accordance with the teaching of EP 3 312 771 B1 or by means of a deep learning solution.

[0045] The processing unit can further be configured to implement, during a BIO step, a biometric identity recognition method based on the image of the Id dermatoglyphs or, as shown in [Fig.2], based on the flattened image of the dermatoglyphs.

[0046] In one possible embodiment exploiting, for example, the teaching of the EP 3 726 423 Al request, the processing unit is further configured to, during an AM step, deduce from the compensated image the lam image of amplitude of the modulation introduced by the lighting projection.

[0047] This amplitude image lam can be used to help identify the fringes of the opacity pattern in the case of amplitude modulation thereof. Furthermore, based on said amplitude image lam, the processing unit can, during a FRAUD step, implement a fraud detection method aimed at authenticating that the part of the human body is a genuine part of a human body. This step may notably include comparing the amplitude image to a depth-dependent interval as determined by the depth map and determining whether the finger is genuine or counterfeit based on the proportion of points belonging to this interval.

[0048] It should be noted that for fraud detection, it can be advantageous to compensate for the projection blur of the target onto the finger. In the case of an approximately sinusoidal target, this can be done by correcting the modulation amplitude image lam based on knowledge of the target's modulation amplitude for a given distance. This knowledge can be obtained from calibration (including the acquisition of images of a uniform flat object at different distances from the sensor's upper surface to calculate the target's amplitude) or from calculation (using the distance between the target and the finger and a parameter corresponding to the angular aperture of the illumination).

[0049] In the foregoing, monochromatic lighting and a mask have been described. The invention, however, also extends to a mask exhibiting color variation which, while it may be more expensive to manufacture, has the advantage of offering better image quality and allowing the use of higher frequencies, resulting in better 3D reconstruction. An example of such a mask includes the superposition of three sinusoidal patterns with a 120° phase shift (one in cyan, another in magenta, and the third in yellow) or two green and blue patterns in opposite phase. Assuming the color of the finger is uniform and the diffusion of the finger in green and blue is similar, in the case of blue and green lighting, the image of the dermatoglyphs is obtained by a simple (weighted) sum of the blue and green images (j _ a[ + ), as well as an image exhibiting a maximum modulation amplitude as a difference (with the same weighting) of these same two images Q _ _ / ? / ,)• This last image is used to extract the phase (Z-MAP step) and the amplitude (AM step).

Claims

Demands

1. A contactless biometric sensor (1), comprising: - an upper face (3) above which, in use, a part of the human body bearing dermatoglyphs is positioned; - a light source configured to illuminate said part of the human body through the upper face; and - an imager configured to acquire an image (I) of said part of the human body through the upper face; characterized in that the upper face is provided with a mask with spatially variable light transmissivity, said mask, in use, introducing onto said part of the human body a projection, called the lighting projection, of a target having a spatial modulation of light intensity, said mask further introducing a projection, called the imaging projection, of said target into the image acquired by the imager.

2. Sensor according to claim 1, wherein the mask is constituted by the upper face on which an opacity pattern is made.

3. Sensor according to claim 1, wherein the mask consists of a transparent support on which an opacity pattern is produced, said support being attached to the upper face.

4. Sensor according to any one of claims 1 to 3, wherein the spatially variable light transmissivity of the mask exhibits periodic modulation along a principal direction.

5. Sensor according to claim 4, wherein the periodic modulation is sinusoidal modulation or triangle modulation.

6. Biometric measurement device comprising a sensor according to any one of claims 1 to 5 and an image processing unit for the image acquired by the imager, the processing unit being configured to compensate (COMP-PI) the imaging projection of said target in the image acquired by the imager, in order to provide a compensated image (the) corresponding to the light reflected by said part of the human body subjected to the projection of lighting.

7. Biometric measuring device according to claim 6, wherein the processing unit is further configured to compensate (COMP-PE) the illumination projection of said target in the compensated image, in order to provide an image of the dermatoglyphs (Id).

8. Biometric measuring device according to claim 7, wherein The processing unit is further configured to implement a biometric identity (BIO) recognition method based on the dermatoglyph image.

9. Biometric measurement device according to claim 7, wherein the processing unit is further configured to infer (Z-MAP) from the compensated image a depth map representative of the geometry of the dermatoglyphs.

10. Biometric measurement device according to claim 9, wherein the processing unit is further configured to flatten (FLAT) the dermatoglyph image using the depth map, in order to provide a flattened image of the dermatoglyphs.

11. Biometric measurement device according to claim 10, wherein the processing unit is further configured to implement a biometric identity (BIO) recognition method based on the flattened image of dermatoglyphs.

12. Device according to any one of claims 6 to 11, wherein the processing unit is further configured to deduce (AM) from the compensated image an image (lam) of amplitude of the modulation introduced by the projection of lighting.

13. Device according to claim 12, wherein the processing unit is further configured to implement, on the basis of said amplitude image (lam), a fraud detection method (FRAUD) aimed at authenticating that the part of the human body is an authentic part of a human body.

14. A biometric measurement method using a device according to any one of claims 6 to 13, comprising the steps: - of acquiring by the imager an image of a part of a human body positioned above the upper face of the sensor; - of processing the image acquired by the processing unit of said device.

15. Product computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the processing step of the process according to claim 14.

Images