EP3284065B1 - Method for verifying a security device comprising a signature - Google Patents

Description

The present invention relates to the field of security devices. It is known to produce a security device and to associate it with a sensitive document in terms of security, such as an identity document, in order to secure said document. An effective security device is characterized in that it is: difficult to produce or reproduce, and difficult to modify in an undetectable manner.

In known manner, an identity document comprises an image associated with the holder of the identity document, such as an identity photo. An identity check can thus compare an image comprising a photo of the holder, present on the identity document, with an acquisition carried out on the bearer of the identity document, in order to verify whether the acquisition corresponds biometrically, or not, to the image , in order to determine whether or not the bearer is the holder he claims to be.

Such a comparison is all the more convincing since the image present on the identity document actually represents the authorized holder. For this it is necessary that this image is indeed the one, authentic and original, arranged by an authority of issue, and that it could not have been modified since the issue.

So that a forger can neither replace nor modify the image on the identity document, in order, for example, to attempt to reproduce the appearance of a holder who differs from the holder, this image is advantageously accompanied by a security device. The security device is advantageously intimately linked to said image, so that the security and authentication characteristics of the security device also apply to the image.

The document WO 01/60047 A2 discloses the features of the preamble of claim 1.

The present invention proposes a multimodal verification mode capable of verifying a security device comprising an image, by making it possible to detect and discriminate between different possible counterfeits.

The present invention relates to a process for verification of a security device comprising an image comprising a signature, comprising the following steps: acquisition of the image according to a first optical spectrum to obtain a first representation, extraction of the signature, and verification of the signature, characterized in that the signature is colorimetric and comprises a particular orientation of a color plate, or the signature is frequency-based, the image comprising at least one reference spatial period.

According to another characteristic, the signature is frequency-based, the image comprising at least one reference spatial period, and the method further comprises the following steps: application of a spectral transformation to the first representation, to obtain a first transform comprising at least a first spatial period, verification that the value of the spatial period(s) correspond(s) to the value of the reference spatial period(s).

According to another characteristic, the image is visible according to the first optical spectrum and at least one second optical spectrum and the method further comprises the following steps: acquisition of the image according to the second optical spectrum to obtain a second representation, verification that the two representations are graphically substantially identical, verification that a distance between the two representations is less than a threshold.

According to another characteristic, the threshold is equal to 10 μm, preferably equal to 5 μm.

According to another characteristic, the distance between the two representations is determined by identifying, by means of a registration algorithm, a transformation for which one of the representations is an image of the other representation.

According to another characteristic, the first optical spectrum is located in the visible spectrum and/or the second optical spectrum is located in the infrared.

According to another characteristic, the method further comprises the following steps: application of the same transformation at the second representation, to obtain a second transform, verification that the first transform is substantially equal to the second transform.

According to another characteristic, the method further comprises a step of: verifying that the value of the spatial period(s) of the second transform corresponds(s) to the value of the period(s) s) spatial reference.

According to another characteristic, the spectral transformation is applied to at least part of the first representation and/or to the same at least part of the second representation.

According to another characteristic, the spectral transformation is applied to at least two parts of a representation, and the method further comprises a step of: checking that the transforms of the different parts are substantially equal.

According to another characteristic, the method further comprises a step of: checking that the two representations are colorimetrically different.

According to another characteristic, the image represents a part of the body, preferably the face, the eye, or the finger, of a holder associated with the security device and the method further comprises the steps of: acquiring an image of the part of the body with a wearer of the security device, verification that the acquired image corresponds biometrically to the first representation, and/or verification that the acquired image corresponds biometrically to the second representation.

According to another characteristic, the security device is associated with a digital storage means comprising a digital representation of the image, and the method further comprises the steps of: reading the digital representation of the image, checking that the digital representation is substantially identical to the first representation, and/or verifying that the digital representation is substantially identical to the second representation.

According to another characteristic, the method comprises another step of: verification that the acquired image corresponds biometrically to the digital representation.

The invention also relates to a verification apparatus comprising means for implementing such a verification method.

The invention also relates to a computer program comprising a series of logic instructions capable of implementing such a verification method.

The invention also relates to a computer data carrier comprising such a computer program.

• la figure 1 illustre un document identitaire comprenant une image associée à un dispositif de sécurité,
• la figure 2 illustre une étape du procédé de vérification, effectuant une comparaison entre deux représentations de l'image acquises selon des spectres optiques différents,
• la figure 3 illustre une autre étape du procédé de vérification, utilisant une transformation spectrale,
• la figure 4 illustre une possible contrefaçon, qu'une transformation spectrale permet de détecter.

Other characteristics, details and advantages of the invention will emerge more clearly from the detailed description given below by way of indication in relation to the drawings in which:

• the figure 1 illustrates an identity document comprising an image associated with a security device,
• the picture 2 illustrates a step of the verification method, performing a comparison between two representations of the image acquired according to different optical spectra,
• the picture 3 illustrates another step in the verification process, using a spectral transformation,
• the figure 4 illustrates a possible counterfeit, which a spectral transformation makes it possible to detect.

The figure 1 illustrates an identity document 20 comprising at least one image 2. The identity document 20 can, if necessary, comprise other elements 21. The image 2 is produced in such a way as to integrate a security device 1. According to one characteristic, the device security 1 consists in that the image 2 comprises a signature. A signature is a specific characteristic of the image 2 able to be detected, typically by an analysis tool. A signature is most often a consequence of the embodiment or of a machine used to produce the image 2. A signature can thus be intrinsically linked to the embodiment. Alternately a signature can be voluntarily introduced into image 2, in order to be able to be detected there for verification.

The nature of a signature can be very diverse. Several non-limiting examples will be described below.

The verification of such a security device 1 comprises the following steps. A first step carries out an acquisition of the image 2 according to the first optical spectrum to obtain a first representation 3.

Such an acquisition is carried out by illuminating the image 2 with lighting according to the desired optical spectrum and by producing the representation 3.4 by an acquisition, typically by means of an image sensor, sensitive in said desired optical spectrum. The result obtained, either a 3.4 representation is an image, which can be digitized and stored in a computer memory and conventionally organized in the form of an image, or a two-dimensional matrix of pixels.

An optical spectrum may be defined herein by at least one optical frequency band. An optical spectrum can thus be all or part of the infrared spectrum, all or part of the X spectrum, all or part of the ultraviolet spectrum, or even all or part of the visible spectrum, or any combination of the preceding.

Thus obtaining a 3.4 representation in an optical spectrum, such as for example the infrared optical spectrum, supposes illumination of the image 2 by a source covering at least the desired infrared optical spectrum and the simultaneous acquisition of representation 3.4 by means of a sensor, such as a camera, sensitive at least in the desired infrared optical spectrum. The representation obtained is an image, a two-dimensional matrix of pixels, where each pixel comprises a unique intensity, indicative of the optical radiation, in the optical spectrum considered, reflected by the image 2. Such a representation 3,4 generally has the form of a monochrome image.

In the particular case of an optical spectrum comprising at least partially the visible optical spectrum, a pixel may include several intensities, indicative of the elemental color intensities. A 3.4 representation then has the form of a polychrome image, ie the form of a superposition of several monochrome images, called component images.

During a second step, the signature is then extracted. The operating mode of this extraction step depends on the nature of the signature. During a third step, the signature is checked, to check that the signature extracted from the representation 3 resulting from the image 2 does indeed correspond to a signature, as it must be present, in that it has been introduced and inserted into the image 2 during the production of the image 2. The operating mode of this verification step still depends on the nature of the signature and is detailed further.

According to a first embodiment, the signature is colorimetric. This also covers numerous procedures, which are illustrated by non-limiting examples. A general idea of this type of signature is to take advantage of the technological advance, in terms of means of manufacture and means of verification, generally observed between manufacturers in the field of security devices and/or government offices issuing the documents identity, in relation to counterfeiters.

According to the first embodiment, the colorimetric signature uses the orientation of a given color board. Thus, in an offset printing process, each basic color (for example RGB(K) or CMY(W), typically 2 to 5 in number, is printed using a color plate. In order to avoid moiré effects, each such color plate is oriented at a different angle, so that each color plate is angularly spaced relative to the others, thus the angle of each color plate is characteristic of a printing machine.

A very precise measurement of this set of angles, or even a voluntary modification of at least one angle, can make it possible to identify and/or particularize a machine printing, and generalizing an issuing body. With precise verification tools, it is thus possible to use at least one angle of this set of angles as a signature.

A second unclaimed example of a color signature uses the precise hue of each color board. Each color board includes a base color. The different colors of the different color boards thus define a colorimetric base, at the same time as a vector base. The basic colors must include substantially distributed colors in order to have a good power of colorimetric expression. It is thus known to use an RGB base: Red Green (Green) and Blue, optionally supplemented by White (White) and/or Black (blacK). Another base is CMY: Cyan Magenta and Yellow (Yellow). But it is possible to define any n-tuple of base colors, or even starting from a classic triplet to slightly modify at least one of the base colors by shifting its hue by a few %. A precise measurement can thus make it possible to detect a printing machine with precision, relying solely on the inevitable dispersions from one machine to another or even by creating a deliberate shift. A voluntary shift is advantageous in that it can make it possible to particularize all the machines of the same entity and thus characterize a transmitter, such as a service or a state.

A third unclaimed example of a colorimetric signature is the use of a particular hue. Such a tint, a particular combination of basic colors can thus be used to produce a specific part of an image 2. It can, for example, be a frame, or even a particular point, produced with a shade definition, given absolute or relative, capable of being verified with great precision. The position of the point used may be part of the signature.

According to another embodiment, the signature is frequency-based. For this, image 2 comprises at least one reference spatial period. Here again several modes of realization are possible and some are illustrated further. The reference spatial period can be intrinsic in that it is introduced by the process for manufacturing the image 2 or it can still be artificial, in that it is added to the image.

The presence of at least one such reference spatial period constitutes a signature whose presence and quality can be verified. Due to the embodiment of the image 2, the period or periods 6.7 is (are) integrated in the entire surface of a representation 3.4, and must (must) be equal to the or the reference spatial period(s) as present in the security device 1 at the origin.

The extraction of the signature is then carried out by means of the following steps. A spectral transformation 8 is applied to the first representation 3. This makes it possible to obtain a first transform 9.

Such a spectral transformation 8 is characterized in that it highlights in the image/representation to which it is applied, due to a series breakdown of periodic functions, the spatial frequencies present in said image/representation. Such a spectral transformation 8 can be any transformation carrying out a decomposition according to a series of functions. A commonly used transformation of this type, in that it advantageously has an efficient and fast numerical implementation, is a fast fourrier transform (FFT). Such a transformation can be one-dimensional. In the case of an 8-transform applicable to an image, there is a two-dimensional version of this transformation (two-dimensional fast fourier transform, FT2), which transforms a 3,4 representation, homogeneous to an image, into a spectrum/transform 9 ,10, itself homogeneous to an image. A point of strong intensity, represented by a black point in the figures, is indicative of a spatial period 6.7, present in the representation 3.4.

A verification step is then carried out absolute, verifying that the value of the spatial period(s), at least the most remarkable, of reference correspond(s) to the value of the period(s) 6 of the first transformed 9.

This correspondence is checked by agreeing a tolerance in order to take into account any measurement and/or calculation errors. It is thus verified that a point of the transform θ, representing a spatial period, indeed corresponds to a reference spatial period, to within a tolerance.

The value of this tolerance must be able to be configured in order to take account of the performance of the optical sensor used. A tolerance equal to 50 µm can be used for an inefficient sensor. However, this tolerance is chosen as small as possible. A tolerance preferably equal to 30 μm, and even more preferably equal to 10 μm, is retained if the performance of the sensor allows it. In the case of the use of a mobile sensor, such as the camera of a smartphone, the value of the threshold can be adapted as a function of the distance, which is variable, of shooting.

This frequency verification step makes it possible to verify that the image 2 corresponds to the original image as produced by the issuing organization of the security device 1, in that it indeed includes the reference frequencies present at the 'origin. This can make it possible to discriminate against a counterfeit attempting to modify all or part of the image 2 without respecting said reference frequencies.

According to another characteristic, the image 2 is produced in such a way as to be visible according to a first optical spectrum and at least one second optical spectrum. The first optical spectrum and said at least one second optical spectrum are advantageously separated, two by two.

There will be detailed further several embodiments making it possible to obtain such a characteristic of the image 2. It should be noted that what characterizes the security device 1 is that, by construction, a same constituent component of the image 2 is visible according to a first optical spectrum and according to at least a second optical spectrum.

It can also be noted that such a characteristic allows the security device 1 to be intimately linked with the image 2, thus making any dissociation almost impossible. Such a security device 1, if it is verified, thus authenticates with relative certainty, its authenticity and its origin, and thus the authenticity and origin of the image 2.

The verification of such a security device 1 comprises the following steps, illustrated with reference to picture 2 . A first step acquires image 2 according to the first optical spectrum to obtain a first representation 3. A second step acquires image 2 according to the second optical spectrum to obtain a second representation 4.

Such an acquisition is carried out by illuminating the image 2 with lighting according to the desired optical spectrum and by producing the representation 3.4 by an acquisition, typically by means of an image sensor, sensitive in said desired optical spectrum. The result obtained, either a 3.4 representation is an image, which can be digitized and stored in a computer memory and conventionally organized in the form of an image, or a two-dimensional matrix of pixels.

An optical spectrum may be defined herein by at least one optical frequency band. An optical spectrum can thus be all or part of the infrared spectrum, all or part of the X spectrum, all or part of the ultraviolet spectrum, or even all or part of the visible spectrum, or any combination of the preceding.

Thus obtaining a 3.4 representation in an optical spectrum, such as for example the infrared optical spectrum, supposes illumination of the image 2 by a source covering at least the desired infrared optical spectrum and the simultaneous acquisition of the representation by means of a sensor, such as a camera, sensitive at least in the spectrum desired infrared optics. The representation obtained is an image, a two-dimensional matrix of pixels, where each pixel comprises a unique intensity, indicative of the optical radiation, in the optical spectrum considered, reflected by the image 2. Such a representation 3,4 generally has the form of a monochrome image.

In the particular case of an optical spectrum comprising at least partially the visible optical spectrum, a pixel can comprise several intensities, indicative of the intensities of elementary colors. A 3.4 representation then has the form of a polychrome image, ie the form of a superposition of several monochrome images, called component images.

It has been seen that, by construction, the same constituent component of the image 2, forms the image 2 and is visible according to the different optical spectra. This characteristic is used for verification, which compares the two representations 3.4 in order to verify that the two representations 3.4 are graphically substantially identical. Moreover, during a second step, it is verified that the two representations 3.4 are not offset relative to each other, in that a distance 5 between the two representations 3.4 remains lower at a threshold.

Thus, as illustrated in figure 2 , it is verified that the first representation 3 represents a first pattern which is substantially identical graphically to a second pattern represented by the second representation 4.

This first step verified, it is possible to determine a distance between the first pattern and the second pattern and to verify that this distance is less than a threshold.

It follows that the security device 1 is verified if and only if the two preceding tests are validated: the first pattern is graphically substantially identical to the second pattern, and the distance between the two patterns is less than the threshold.

As the security device 1 is designed, the same component of the image 2 is visible according to the first spectrum optical and according to said at least one second optical spectrum. Also a shift or distance between the two 3.4 representations is theoretically zero. In order to take account of measurement and/or calculation inaccuracies, a tolerance is introduced in the form of said threshold. However, this threshold can be chosen to be very small. In order to allow discrimination between an authentic device, where the visible image according to a first optical spectrum is produced jointly and simultaneously with the visible image according to a second optical spectrum, and a possible counterfeit which would produce, in two stages, a first image visible according to a first optical spectrum and a second image visible according to a first optical spectrum, aligned with the first image, said threshold should be lower than the alignment capabilities (in English: registration) of current technologies and production machines. A threshold equal to 10 μm, preferably equal to 5 μm, meets this need, in that such alignment performance is unattainable regardless of the technology used.

It has been seen that a first verification step consisted in comparing the first representation 3 with the second representation 4 and in testing the graphic identity of the two representations. Numerous image processing techniques are applicable to carry out such a comparison.

According to an illustrative embodiment, the identity between the two representations 3,4 can be verified by identifying, by means of a known registration algorithm, a transformation making it possible to pass from one representation 3 to the other representation 4. In this case the verification is acquired if said transformation is sufficiently close to the identity transformation. An advantage of this approach is that the identification of the transformation still provides, as a modulus of this transformation, the distance between the two 3,4 representations, which can then be compared to the threshold.

In the case where at least one of the representations 3.4 is a polychrome image, the comparison can be applied to any of the component images of said polychrome image, or even after preprocessing of the polychrome image in order to make it monochrome, by any method whatsoever (average, saturation, etc.).

The two optical spectra can be arbitrary, as soon as one has a component, visible simultaneously according to these two optical spectra and capable of entering into the production of image 2.

Advantageously, in order to allow certain tests with the naked eye, one of the optical spectra is situated in the visible spectrum. An optical spectrum included in the visible spectrum also has the advantage of simplifying the illumination of image 2 when performing the acquisition, since it can be performed by daylight or even by any type of light. usual artificial lighting.

The use of the visible spectrum is further advantageous in that it makes it possible to obtain a polychrome representation. As described above, polychromy can provide additional verification.

Alternatively, one of the optical spectra can be located in the ultraviolet, UV.

Alternatively, one of the optical spectra can be located in the infrared, IR.

Such optical spectra, not located in the visible, improve security in that their use is not necessarily detected by a counterfeiter. They slightly complicate the verification step in that lighting and a specific means of acquisition are necessary. However, it should be noted, in the case of an identity document 20, that the control offices, such as border posts, are most often already equipped with scanners capable of carrying out an IR or UV acquisition.

The embodiments of the image 2, allowing it to be visible according to at least two optical spectra, are detailed further.

Some of these embodiments contribute, intrinsically or artificially, to providing the image 2 with a frequency signature, so that it includes at least least one space period.

It was seen previously that the frequency signature of an image 2 can be checked absolutely.

When the image 2 is visible according to at least two optical spectra, it is still possible to apply a relative verification. For this, the same transformation 8 is again applied to the second representation 4. This makes it possible to obtain a second transform 10.

From these transforms 9,10, it can be verified that the first transform 9 is substantially equal to the second transform 10.

This equality can be tested by many methods. If the 9,10 transforms are images, it is possible to apply all the image comparison methods to them, such as the method previously described for comparing the representations and verifying that they are identical (identification of the registration).

In all cases, the 9,10 transforms represent characteristic points of the remarkable periods. It is possible to use methods extracting a set of the most remarkable p periods for each of the transforms 9,10 and to compare the p periods of each of the sets. It is considered that two transforms are equal if at least some parts of the remarkable periods of a transform 9 are found in the set of remarkable periods of the other transform 10.

If an equality is found, the verification step is positive and the security device 1 is deemed verified and therefore valid. Otherwise, the verification step is negative and the security device 1 and/or its authenticity are in doubt.

The previous verification step is relative in that it compares the respective 9,10 transforms of the two 3,4 representations. This makes it possible to verify that the image 2 has indeed been produced jointly, for its part 3 visible according to a first optical spectrum and for its part 4 visible according to at least a second optical spectrum, and that one finds substantially the same frequency spectra in the two representations 3.4, indicative of the presence of the same original frequency signature 5.

The absolute verification step, carried out for the first transform 9, can also be applied to the second transform 10, in order to verify that the period(s), at least the most remarkable of reference are indeed present in the (or the) period(s) 7 of the second transform (10). This second step of frequency verification makes it possible to verify that the particular periodicity of the image 2 corresponds to that carried out by the body issuing the security device 1.

According to a first embodiment, the spectral transformation 8 is applied to the whole of the first representation 3 and/or, likewise, to the whole of the second representation 4.

Alternatively, according to another embodiment, the spectral transformation 8 is applied to at least part of the first representation 3 and to the same at least part of the second representation 4. Each of the partial transforms can then be compared, with a partial transform of the other representation, for example with the corresponding partial transform, this comparison being able to be carried out part by part, but not necessarily, and/or with another partial transform of the same representation.

An interest of a verification using a spectral transformation 8 will now be illustrated in relation to the figure 4 .

It is assumed that an image 2 is counterfeited in order to modify at least a part l1 of it. Thus, as illustrated in figure 4 , a modified part 11 aims to modify the eyes on a photo ID. While the original image 2 and therefore its representation 3 comprises a frequency signature 5, the modified part 11, whether by addition or by replacement, whatever the technology used, has every chance of presenting a frequency signature 5' different from the original frequency signature 5, y including the case where no 5' frequency signature is present. Also a comparison of the 9,10 spectral transforms, carried out on all or part of a 3,4 representation necessarily reveals a detectable difference.

There will now be described several embodiments making it possible to obtain an image 2 comprising a security device 1 visible according to a first optical spectrum and according to at least a second optical spectrum.

According to a first embodiment, a security device 1 can be, in known manner, an image 2 produced by monochrome laser engraving. Such a safety device 1 is known and widely used in the technical field. The principle is to arrange a layer sensitive to the laser, in which it is possible to carry out, by means of a laser beam, a localized carbonization. It is thus possible, by means of a laser, to draw and produce an image 2. This embodiment makes it possible to produce an image, necessarily monochrome, such as an identity photo. It is known that a point of image 2, blackened by the laser, is visible in a first optical spectrum: the visible spectrum and that moreover a point of image 2 is still visible according to a second optical spectrum: the infrared spectrum.

It should be noted here that this property of visibility according to at least two optical spectra is known and exploited by the controllers. It is verified, for an image obtained by monochrome laser etching, that the image is visible in the visible optical spectrum and that, moreover, the image is visible in the IR optical spectrum. This allows the controller to verify that it is indeed in the presence of an image produced by monochrome laser engraving. However today, this verification is only human and qualitative: the controller visually verifies that an image can be seen, according to the two optical spectra. However, the prior art does not check either that the two representations 3.4 are identical, or that their distance is less than a threshold. The invention, which provides a quantitative approach, advantageously allows these two operations to be carried out automatically, with much more precision, including decision making.

According to another embodiment, a security device 1 can be an image 2 produced by color laser engraving. For this, a security device 1 comprises an arrangement comprising a color matrix. The color matrix is an array of pixels, each pixel comprising at least two sub-pixels of advantageously elementary and different colors. According to a first embodiment, the color matrix is sensitive to the laser, a laser firing allowing selectively, for each pixel, to express a hue by combining the elementary colors of the sub-pixels. According to another embodiment, the color matrix is insensitive to the laser, and said arrangement comprises at least one layer sensitive to the laser. Said at least one sensitive layer is arranged above and/or below the color matrix. Laser etching, according to the monochrome technology described above, makes it possible to produce, in said at least one sensitive layer, a monochrome mask, allowing each pixel to selectively express a hue by combining the elementary colors of the sub-pixels.

These two embodiments allow the production of a color image by laser engraving. Here again, the point carbonized by laser constituting image 2 is simultaneously visible in the visible optical spectrum and in the IR optical spectrum. It is therefore the same component, which is thus necessarily located at the same place in the first representation 3 or in the second representation 4.

According to yet another embodiment, a security device 1 can be an image 2 produced by a printing technique. The printing technique can be any printing technique: offset, screen printing, retransfer, sublimation, inkjet, etc..., as long as it uses an ink comprising at least one visible component according to the first optical spectrum and the second optical spectrum. This component, integrated in the ink, thus determines according to which optical spectra the image 2 can be seen. An image 2 can thus be invisible in the visible spectrum, but be visible in the IR and in the UV. The printing of image 2 creates image points which are simultaneously visible according to the at least two optical spectra. Here again, an image point is a single component, necessarily located at the same place in the first representation 3 or in the second representation 4.

A simplifying technique of counterfeiting consists in producing an image 2 in monochrome. Thus a counterfeiter may be tempted to produce a monochrome image 2, which is simpler to manufacture or requires simpler tools. Thus a polychrome print can be replaced by a monochrome print. Similarly, a counterfeiter can be equipped with a monochrome engraving laser, and master this technology, which is already quite old, and be tempted to replace a 2 color image created by laser engraving, whose very recent technology is still not widely available and probably difficult to access. to a counterfeiter, by a monochrome image 2 created by laser engraving.

Also, and provided that the authentic security device 1 includes a color image and that at least one of the optical spectra is the visible spectrum, the verification method can advantageously include an additional step verifying that the two representations 3,4 are colorimetrically different. Thus, typically, one of the representations represents a polychrome acquisition of the image 2 and the other representation, for example because it is visible in an optical spectrum situated outside the visible spectrum, represents a monochrome acquisition. This verification step checks an effective presence of color in one of the representations. The 3.4 representations are here colorimetrically different, even if they are graphically identical (same pattern).

The colorimetric difference can be checked by any colorimetric processing method. According to a possible embodiment, the 3,4 representations can be modeled according to a CIE Lab colorimetric model. he can it can then be verified that the representation reputed to be in color actually has generally high values of the coefficients a,b, whereas the representation reputed to be monochrome, is gray, and has low values of the coefficients a,b. An analogous approach could use a conversion of the 3,4 representations according to an HLS model, and an observation of the value of the saturation S.

At least three embodiments of a security device 1 visible according to at least two optical spectra have been seen: monochrome laser engraving, color laser engraving and printing with special ink.

An image 2 made by monochrome laser engraving includes a frequency signature 5, because the laser shots are made according to a shot matrix. Such a firing matrix, for example rectangular, is advantageously periodic. It therefore appears, spatially, at least one period 6.7, per dimension. In the case of a rectangular matrix, there may thus appear a period 6.7 along a first axis and a second period 6.7 along the other axis of the matrix.

Also if we apply a spectral transformation 8 to a 3.4 representation resulting from such an image 2, the transform 9 of representation 3 is equal to the transform 10 of representation 4. This spectral transformation 8 reveals, and this for the two optical spectra, at least the two periods 6.7. If the rectangular matrix is oriented parallel to image 2, and the spectral transformation 8 is an FFT2, there will appear at least a first point 6.7 on the ordinate axis, representative of the period along the abscissa axis and at least one second point on the abscissa axis, representative of the period along the ordinate axis.

An image produced by color laser engraving intrinsically comprises, most often, a frequency signature 5 in that the arrangement making it possible to engrave such an image 2 in color comprises a color matrix. Although this is not an obligation, in order to facilitate the engraving, the pixels and the sub-pixels comprising the colors are advantageously arranged in said color matrix periodically. It is thus possible to find, according to at least one dimension, a main period 6.7 corresponding to the distance between the pixels. In addition, each pixel comprises a number n, at least equal to 2, and conventionally equal to 4 (Cyan, Magenta, Yellow, Black), of sub-pixels each comprising a basic color. These n colors are advantageously spatially equitably distributed, thus forming a secondary spatial period n-sub-multiple of the main period 6.7.

According to one embodiment, the color matrix is arranged in lines, for example horizontal, alternating according to a sequence advantageously identically repeated the n colors.

The color matrix is theoretically visible only in the visible optical spectrum. However, dots produced by laser etching are visible on the one hand in the visible optical spectrum and on the other hand in the infrared, IR optical spectrum. Also, in an engraved image 2, the engraved points being necessarily arranged according to the color matrix, will make it possible to show the main 6.7 and secondary spatial periods of the color matrix. This characteristic assumes that the density of engraved points is sufficient. This is the case for a complex image and particularly for a photograph. The main 6.7 and secondary spatial periods appear, both in the first transform 9 resulting from a representation 3 according to a first optical spectrum, here the visible spectrum, and in the second transform 10 resulting from a representation 4 according to a second spectrum optical, here the IR spectrum.

For an authentic security device 1, the same frequency signature 5 from the color matrix is revealed and highlighted by the engraved dots and the two transforms 9,10 must be substantially identical. Moreover, the periods 6.7 highlighted by the spectral transformation 8 must correspond to the main periods and where applicable, the secondary reference of the frequency signature 5, as manufactured.

An image 2 produced by a printing process does not necessarily include a frequency signature 5. However, certain production methods can induce a periodic arrangement of the points which then forms a frequency signature 5, of which at least one spatial period 6.7 is the distance between the points. This periodic pattern thus forms a frequency signature 5 which can then be used to verify the security device 1 by applying a spectral transformation 8.

According to another embodiment, it is also possible to include in the image 2 an additional frequency signature, voluntarily added, by printing a periodic pattern. It is thus possible to insert a frequency signature 5, in an image 2, by replacing certain points or lines, advantageously periodically arranged, with a given color. Thus, like a color matrix capable of allowing the production of a color image by laser engraving, or even to attempt to simulate such a matrix, it is possible to modify an image 2 by replacing a line on p by a black line. This modifies the image 2 sufficiently little for it to remain usable, while giving it a frequency signature 5 usable for the purposes of verification after application of a spectral transformation 8 .

If moreover an image 2 is printed with a special ink, it is possible to verify the presence, the identity and the distance of the two representations 3,4 resulting from acquisitions according to at least two optical spectra. If the image 2, or at least said additional frequency signature 5 is printed with a special ink, the frequency signature 5 thus produced is visible according to at least two optical spectra and must be present in the two transforms 9,10 resulting from these two representations 3.4, these two transforms then being equal.

According to another characteristic, image 2 represents a part of the body of a holder associated with the security device 1. The verification method may further comprise the following steps. A first step consists of acquiring an image of said part of the body from the wearer of the security device 1. A second step verifies that this acquired image corresponds biometrically to the image 2 of the security device 1. image 2 of security device 1 is deemed to be a representation of the authorized holder. Also if a biometric match can be verified between a direct acquisition from the bearer accompanying the security device 1, it can be assumed that the bearer is indeed the holder he claims to be.

If the image 2 is visible according to two optical spectra, the verification can be doubled, by verifying that the acquired image 13 corresponds biometrically to the first representation 3, and/or by verifying that the acquired image 13 corresponds biometrically to the second representation 4.

The term biometric correspondence is used here because such a step, comparing a direct acquisition from the wearer and an image 2, associated with the security device 1, resulting from an acquisition having been carried out during the issue, can be relatively old, and the wearer's appearance may have changed, is necessarily more complex than verifying identity between two images. Biometric matching techniques are assumed to be known.

This applies for example to the case where the part of the body is the face, the image 2 then representing an identity photograph of the bearer of an identity document 20 associated with said security device 1. According to another embodiment , it can still be the eye, one of the fingers or any other part of the body.

The verification process thus combines several verification steps targeting different aspects of a control. It is checked that the image 2 is authentic, and could not be modified since the delivery of the security device 1. It is also checked that the bearer corresponds to the holder. The guarantees provided by each of these checks reinforce the security of the security device 1.

According to another characteristic, the security device 1 is associated with a digital storage means comprising a digital representation of the image 2. Such a storage means is typically a secure device (in English: secure device, SD) offering services access to an internal memory, in a secure manner, such as a microcircuit. The digital representation of the image 2 has been stored beforehand, in a controlled manner, by the issuing authority of the security device 1. It is therefore deemed to be a representation of the holder. Securing guarantees that it has not been modified.

Such a characteristic makes it possible to redundant the security device 1 and to complete the verification method by adding another verification by means of the following steps. According to a first step, the digital representation of the image 2 is read from the storage means. According to a second step, the method compares the digital representation with one and/or both representations 3,4. Verification is deemed acquired if the digital representation is substantially identical to all the representations 3.4 with which it is compared.

If an acquisition of an image of the wearer is carried out, it is still possible to add another verification by testing a biometric correspondence between said image acquired from the wearer and the digital representation of the image 2 coming from the storage means.

The different characteristics of the verification process having been detailed, the description will now be completed by means of use cases, making it possible to highlight the discriminating capacities of each of the verifications.

Use case A - genuine device

An authentic identity document 20 comprising an image 2 representing an identity photo produced by color laser engraving and a microcircuit containing a representation digital photo ID is checked.

The verification method carries out an acquisition, advantageously in color, of the image 2 according to a visible spectrum to obtain a first representation 3, a monochrome acquisition of the image 2 according to an IR spectrum to obtain a second representation 4, a direct acquisition, advantageously in color, of the wearer's face and extracts a digital representation of the microcircuit.

A first verification confirms that the first representation 3 (visible) is graphically identical and not far from the second representation 4 (IR).

A second verification confirms that the direct acquisition corresponds biometrically to the first representation 3 (visible), and corresponds biometrically to the second representation 4 (IR).

A third check confirms that the digital representation from the microcircuit is identical to the first representation 3 (visible), is identical to the second representation 4 (IR), and biometrically corresponds to the direct acquisition.

A fourth verification applies a spectral transformation 8 to the representation 3, advantageously made monochrome, and to the representation 4, compares the two transformed 9,10 obtained to verify their equality and verifies that the spatial periods 6,7 detected are the periods of the frequency signature 5 of the color matrix used. The presence of the 5 frequency signature of the original color matrix, visible both in the visible spectrum and in the IR spectrum, ensures that the two 9,10 transforms are equal and that their 6,7 periods correspond to the periods of the color matrix. original.

A fifth verification verifies that representation 3, in color, differs colorimetrically from representation 4, monochrome.

Use case B - counterfeit device 1

A forged identity document 20 in that it includes an image 2 made by printing.

Image 2, printed here, has no visibility in the IR. Also the second representation 4 is a null image. The printed image has no frequency signature 5.

The first check fails in that it detects a difference between the first representation 3 (visible) and (the absence of content of) the second representation 4 (IR).

It can be assumed that the counterfeiter has made an image 2 representing a photo of the wearer. Also the second check succeeds in that a biometric match is found for the first representation 3 (visible). However it fails for the second representation 4 (IR).

Provided that the counterfeiter was able to modify the digital representation in the microcircuit, the third verification succeeds in that an identity is found for the first representation 3 (visible) and a biometric correspondence is found with the direct acquisition. However it fails for the second representation 4 (IR). If the counterfeiter has failed to alter the digital representation in the microcircuit, all checks fail.

Due to the absence of frequency signature 5 in the counterfeit printed image 2, the fourth verification can find an equality between the two transforms 9,10 (absence of significant spectrum) but fails in that it does not find the periods of the color matrix, neither in the transform 9 resulting from the visible spectrum, nor in the transform 10 resulting from the IR spectrum.

The fifth check succeeds in that image 2 is in color.

Use case C - counterfeit device 2

A counterfeit identity document 20 in that it comprises an image 2 produced by monochrome laser engraving.

Image 2, here laser-engraved, is visible in the visible and in the IR and presents two identical and superimposed (not distant) 3.4 representations. The engraved image monochrome does not have a frequency signature 5.

The first verification succeeds in that it detects a representation 3 (visible) identical and superimposed with the second representation 4 (IR).

It can be assumed that the counterfeiter has made an image 2 representing a photo of the wearer. Also the second verification succeeds in that a biometric match is found, both for the first representation 3 (visible) and for the second representation 4 (IR).

Provided that the counterfeiter was able to modify the digital representation in the microcircuit, the third verification succeeds in that an identity is found for the first representation 3 (visible), for the second representation 4 (IR) and a biometric match is found with direct acquisition.

Due to the absence of frequency signature 5 in the counterfeit engraved image 2, the fourth verification can find an equality between the two transforms 9,10 (absence of significant spectrum) but fails in that it does not find the periods of the color matrix, neither in the transform 9 resulting from the visible spectrum, nor in the transform resulting from the IR spectrum. In the particular case where a frequency signature is present, it bears no resemblance to a frequency signature of a color matrix and the spectral verification fails.

The fifth check fails in that image 2 is monochrome.

Use case D - counterfeit device 3

A counterfeit identity document 20 in that it comprises an image 2 produced by printing, said printing including lines simulating a frequency signature 5 of a color matrix.

Image 2, printed here, has no visibility in the IR. Also the second representation 4 is a null image. The printed image has a convincing frequency signature, but only in the visible.

The first check fails in that it detects a difference between the first representation 3 (visible) and the absence of content of the second representation 4 (IR).

It can be assumed that the counterfeiter has made an image 2 representing a photo of the wearer. Also the second check succeeds in that a biometric match is found for the first representation 3 (visible). However it fails for the second representation 4 (IR).

Provided that the counterfeiter was able to modify the digital representation in the microcircuit, the third verification succeeds in that an identity is found for the first representation 3 (visible) and a biometric correspondence is found with the direct acquisition. However it fails for the second representation 4 (IR).

If the printed frequency signature is sufficiently well produced to simulate a frequency signature 5 in the visible, the fourth verification may succeed in that it finds an acceptable transform θ in the visible. However, the fourth verification fails in that the transform 10 in the IR is not acceptable (absence of significant spectrum) and is not equal to the transform 9 (visible) either.

The fifth check succeeds in that image 2 is in color.

Claims (15)

1. A method for verifying a security device (1) comprising an image (2) including a signature, comprising the following steps:

   - acquiring the image (2) according to a first optical spectrum in order to obtain a first representation (3),

   - extracting the signature, and

   - verifying the signature,

   characterized in that:

   - the signature is colorimetric and comprises a particular orientation of a color chart, or

   - the signature is a frequency signature, the image (2) comprising at least one reference spatial period.
2. The method according to claim 1, wherein when the signature is a frequency signature, the method further comprises the following steps:

   - applying a spectral transformation (8) to the first representation (3), in order to obtain a first transform (9) comprising at least a first spatial period (6),

   - verifying that the value of the spatial period(s) (6) corresponds to the value of the reference spatial period(s).
3. The method according to claim 1 or 2, wherein the image (2) is visible according to the first optical spectrum and at least a second optical spectrum and wherein the method further comprises the following steps:

   - acquiring the image (2) according to the second optical spectrum in order to obtain a second representation (4),

   - verifying that the two representations (3, 4) are graphically substantially identical,

   - verifying that a distance between the two representations (3, 4) is less than a threshold.
4. The method according to claim 3, wherein the distance between the two representations (3, 4) is determined by identifying, by means of a reframing algorithm, a transformation for which one of the representations (3) is the image of the other representation (4).
5. The method according to claim 3 or 4, further comprising the following steps:

   - applying the same transformation (8) to the second representation (4) in order to obtain a second transform (10),

   - verifying that the first transform (9) is substantially equal to the second transform (10).
6. The method according to claim 5, further comprising a step of:

   - verifying that the value of the spatial period(s) (7) of the second transform (10) corresponds to the value of the reference spatial period(s).
7. The method according to claim 5 or 6, wherein the spectral transformation (8) is applied on at least part of the first representation (3) and/or on the same at least part of the second representation (4).
8. The method according to claim 5 or 7, wherein the spectral transformation (8) is applied on at least two parts of a representation (3, 4), and wherein the method further comprises a step of:

   - verifying that the transforms of the different parts are substantially equal.
9. The method according to claim 7 or 8, further comprising a step of:

   - verifying that the two representations (3, 4) are colorimetrically different.
10. The method according to any one of claims 3 to 9, wherein the image (2) represents part of the body, preferably the face, the eye or the finger, of a holder associated with the security device (1) and wherein the method further comprises the steps of:

    - acquiring an image (13) of the part of the body from a wearer of the safety device (1),

    - verifying that the acquired image (13) biometrically corresponds to the first representation (3), and/or

    - verifying that the acquired image (13) biometrically corresponds to the second representation (4).
11. The method according to any one of claims 3 to 10, wherein the security device (1) is associated with a digital storage means comprising a digital representation of the image (2), and wherein the method further comprises the steps of:

    - reading the digital representation of the image (2),

    - verifying that the digital representation is substantially identical to the first representation (3), and/or

    - verifying that the digital representation is substantially identical to the second representation (4).
12. The method according to claim 11, further comprising a step of:

    - verifying that the acquired image (13) biometrically corresponds to the digital representation.
13. A verification apparatus characterized in that it comprises means for implementing a verification method according to any one of the preceding claims.
14. A computer program comprising instructions which lead the apparatus according to claim 13 to execute the steps of a verification method according to any one of claims 1 to 12.
15. A computer data medium characterized in that it comprises a computer program according to the preceding claim.

Images