METHOD AND DEVICE FOR SECURING ACCESS TO A MESSAGE

Description

TECHNICAL FIELD OF THE INVENTION

[0001] The present invention relates to a method and device for securing access to a message. It applies, in particular, to security documents and bank cards. STATE OF THE ART

[0002] There are many ways to protect content, or a message carried on a medium. For example, coding or encryption is applied to the data constituting this message. However, simply knowing the decoding or decryption key allows access to the message. However, these keys must be stored on media, for example, paper, plastic, or electronic media, or on servers. Whether these keys are carried by the user or stored on a computer system, a sufficiently equipped malicious third party can access them.

[0003] We know the patent application FR2957705 which discloses a means of reading an anti-copy mark.

[0004] These protections are therefore insufficient. SUBJECT OF THE INVENTION

[0005] The present invention aims to remedy all or part of these drawbacks.

[0006] To this end, according to a first aspect, the present invention relates to a device for reading a message, which comprises: a means for reading biometric data of a carrier of the medium, a means for capturing, on a medium, a coded message comprising elementary message units, said message having noise consisting of marking defects of the message on said medium which are random and unpredictable for each elementary message unit, a means for reducing noise of the captured message, as a function of biometric data and a means for decoding at least a part of the at least partially denoised message.

[0007] Thanks to these provisions, at least part of the coded message is not accessible in the absence of the bearer of the authorized medium whose biometric data are required. In addition, the device which is the subject of the present invention can be autonomous with respect to any remote source of information, for example a server on a computer network.

[0008] In embodiments, the decoding means implements redundancies of the denoised message to determine at least a portion of the encoded message.

[0009] Thanks to these provisions, the part of the message to be decoded can be accessed without calling on a remote information source.

[0010] In embodiments, the biometric data reading means is configured to read a fingerprint of the carrier of the medium.

[0011] In embodiments, the biometric data reading means is configured to capture an image of the face or hand of the carrier.

[0012] Thanks to these provisions, biometric data capture is carried out without contact between the holder of the support and the device, which reduces the need for cleaning and the risk of contamination.

[0013] In embodiments, the noise reduction means calculates a digest of biometric data and implements the digest to reduce noise in the captured message.

[0014] It should be remembered here that the condensate is better known to those skilled in the art under its English name of “hash”.

[0015] In embodiments, the noise reduction means implements a symmetrical condensate.

[0016] For example, a symmetric digest is derived from a symmetric key cipher.

[0017] In embodiments, the decoding means is configured to decode biometric data.

[0018] Thanks to these provisions, the reading device enables the display of biometric data, for example for visual confirmation of the identity of the carrier of the medium or for additional automatic verification of the identity of the carrier of the medium.

[0019] In embodiments, the decoding means is configured to implement an asymmetric key to access at least a portion of the message.

[0020] Thanks to these provisions, the reading device which is the subject of the present invention can be autonomous by having a set of keys, for example public keys from a public key infrastructure (“PKI”).

[0021] In embodiments, the reading device which is the subject of the present invention comprises a means for entering a secret code by the carrier of the medium, the decoding means being configured to implement said secret code to access at least part of the message.

[0022] Thanks to these provisions, the consent of the holder of the medium to read the message can be formalized by entering the secret code.

[0023] In embodiments, the reading device which is the subject of the present invention comprises a means for storing and communicating payment information.

[0024] Thanks to these provisions, payment is secure, the identity of the cardholder being verified by the use of biometric data.

[0025] According to a second aspect, the present invention aims at a method of reading a message, which comprises: a step of reading biometric data of a carrier of the medium, a step of capturing, on a medium, a coded message comprising elementary message units, said message having noise consisting of marking defects of the message on said medium which are random and unpredictable for each elementary message unit, a step of reducing noise of the captured message, as a function of biometric data and a step of decoding at least a part of the at least partially denoised message.

[0026] The present invention also relates to a device for marking a message on a support, according to claim 1.

[0027] The present invention also relates to a method of marking a message on a medium, according to claim 13.

[0028] The advantages, aims and characteristics of this reading method, of this marking device, of this marking method being similar to those of the reading device which is the subject of the present invention, they are not recalled here. BRIEF DESCRIPTION OF THE FIGURES

[0029] Other advantages, aims and particular characteristics of the present invention will emerge from the description which follows, given for explanatory and in no way limiting purposes, with reference to the appended drawings, in which: there figure 1 represents, schematically, a particular embodiment of the reading device which is the subject of the present invention, the figure 2 represents, in the form of a flowchart, steps of a particular embodiment of the reading method which is the subject of the present invention, the figure 3 represents, schematically, a particular embodiment of the marking device which is the subject of the present invention, the figure 4represents, in the form of a flowchart, steps of a particular embodiment of the marking method which is the subject of the present invention and the Figure 5 represents, schematically, a brand. DESCRIPTION OF EXAMPLES OF EMBODIMENT OF THE INVENTION

[0030] We note, from now on, that the figures 1 And 3 are not to scale.

[0031] We observe, in figure 1 , a device 100 for reading a message 105, which comprises: a means 110 for reading biometric data 115 of a carrier of the medium, a means 120 for capturing, on a medium 125, a coded message 130 comprising elementary message units, said message having noise consisting of defects in the marking of the message on said medium, said defects being random and unpredictable for each elementary message unit, a means 135 for reducing noise in the captured message, as a function of biometric data and a means 140 for decoding at least a part of the at least partially denoised message.

[0032] The biometric data reading means 110 is, for example, a reader of at least one fingerprint of the carrier of the support or a camera configured to capture a profile of a hand or the face of the carrier of the support. The biometric data reading means 110 provides data characteristic of the part of the body of the carrier of the support which is analyzed, for example the characteristic points of a fingerprint, the lengths and shapes of the fingers of the hand or the respective dimensions and positions of the components of the face. In embodiments, the biometric data reading means 110 captures the voice of the user.

[0033] The medium 125 is a document, for example an identity or payment card, a sheet of paper, a label. The marking of the message on the medium has the effect of modifying, locally, at least one physical characteristic of the medium, on the surface or in the thickness. For example, the marking is a print, with or without ink. The coded message 130 comprises elementary message units, said message having noise consisting of marking defects of the message on said medium, said defects being random and unpredictable for each elementary message unit. The error rate due to the marking noise is preferably greater than five percent, that is to say that at least five percent of the elementary units of the mark are sufficiently randomly modified so that their reading by the capture means is erroneous.

[0034] The means 120 for capturing the message 130 is, for example, a matrix image sensor or a scanner.

[0035] The means 135 for reducing noise of the captured message, based on biometric data, is configured to calculate a digest (or “hash”) of biometric data and to implement said digest to reduce the noise of the captured message.

[0036] Here we observe that, although the biometric data are different, between noise and denoising, since they correspond to two captures distant in time and made in different conditions, their processing, which extracts characteristics, for example by frequency analysis, frequency transformation or condensate extraction, provides data with a low variation, although not zero. By applying these characteristics, noisy because variable, to the message, we increase or reduce the noise with which this message is affected.

[0037] For example, starting from a digital trace, which is normalized in dimension and angles, we extract singular points (including minutiae), themselves normalized. By applying an "exclusive or" function to an image of this trace or of these singular points or to a condensate of these images, we produce a noise effect or a denoising of the message.

[0038] It is recalled here that a fingerprint or dactylogram is the result of placing a finger on a support after inking it. The design formed on the support is made up of dermatoglyphs. Fingerprints are unique and characteristic of each individual.

[0039] A "fingerprint", a more general term, is the result of contact between a finger and a medium, a fingerprint being an example of a fingerprint.

[0040] Papillary traces (or their prints) include digital traces (or their prints, of which automatic identification software identifies 150 to 200 characteristic points) and palmar traces (traces of the palms of the hand identified by 1,500 to 2,000 characteristic points).

[0041] By analogy, we speak of a voice print authenticated by speaker recognition techniques.

[0042] The digital trace can be: visible (or direct): it is said to be positive when material is applied and it is said to be negative when material is removed, latent (invisible to the naked eye): the trace comes from the deposit of sweat (secretion of sweat glands: 99% water which, when evaporating, leaves salts and amino acids in place) and / or the deposit of sebaceous secretions (sebaceous glands) and / or molded: the trace comes from the contact of a finger with a malleable surface (wax, putty, etc.).

[0043] Fingerprints can be divided into three main types of patterns: arcs (also called arches or tents), right- or left-hand loops, and whorls, whorls, or swirls. These three types of prints account for 95% of human fingers: 60% for loops, 30% for spirals, and 5% for whorls. 17 Each print belongs to a classified type: adelt, bidelt, tridelt (rare). Monodelts are divided into subgroups: normal, external, composite, and similarly for bidelts: concentric whorls, Z-whorls.

[0044] The patterns are differentiated from each other using “singular points” on the loops, arcs or coils: global singular points: core or center: place of convergence of the striae, delta: place of divergence of the striae, local singular points (also called minutiae): points of irregularity located on the capillary lines (terminations, bifurcations, islands - assimilated to two terminations, lakes).

[0045] It is estimated that there are more than one hundred points of convergence between two identical prints. In France, the law requires 12 points (called minutiae), 18 uncontradictory readings to authenticate a suspect's print. Between 8 and 10 points, a strong presumption is established thanks to algorithms. In Switzerland, a probabilistic system is used in comparisons: the probability of the pattern is calculated based on the statistics of the appearance of different minutiae: islands, divisions.

[0046] The probability of two people having the same fingerprint is 1 in 10 14< , which is very low on the scale of the human population. In addition, its random nature eliminates the risks of resemblances between individuals sharing the same genetic heritage: monozygotic individuals such as twins or quadruplets, for example, will each have a set of fingerprints that is unique to them and different from that of other individuals in the same family, the same goes for slightly different fingerprints between the left and right hands. The explanation is that genes are responsible for the general architecture of the prints (the three main patterns) while embryonic development and the environment of intrauterine life influence the singular points.

[0047] Preferably, noise reduction and noise reduction (means or steps) implement a symmetric hash. This principle is relatively easy to implement with biometric data such as iris because, by its nature, it is a print that is stable over time. By analogy, this biometric data behaves, when read, like a barcode.

[0048] Among the biometric data that can be implemented for sound effects and noise reduction, we can cite: the image of the iris of the eye, each fingerprint, data resulting from processing of a facial image, for example from facial recognition, data resulting from processing of a hand image, for example from recognition of the profile of the hand placed on a flat surface, data resulting from processing of voice capture, for example from voice recognition and data resulting from processing of a succession of positions, for example from recognition of movements, for example hand movements during a signature, body movements, eye movements.

[0049] The means 140 for decoding at least a portion of the at least partially denoised message implements redundancies of the denoised message to determine at least a portion of the coded message. For example, the means 140 knows the positions of the redundant elementary message units or the redundancy generation code, to exploit these redundancies, detect and correct errors constituting the noise. For example, the redundancies take the form of error correction codes (in English “check redundancy code” or redundancy verification code).

[0050] The decoding means 140 is, in embodiments, configured to decode biometric data. In embodiments, the device 100 comprises a means for displaying the biometric data read in the message. A third party can thus visually verify the identity of the carrier of the medium. In embodiments, the device 100 comprises a means for automatically verifying that the carrier of the medium corresponds to the biometric data read in the message.

[0051] In embodiments, the decoding means 140 implements an asymmetric key to access at least part of the message.

[0052] In the embodiment illustrated in figure 1, the device 100 further comprises a means 145 for entering a secret code by the carrier of the medium. For example, the means 145 is a keyboard. The decoding means 140 is then configured to implement the secret code to access at least part of the message. For example, this secret code is part of a key for decrypting the message.

[0053] In the embodiment illustrated in figure 1 , the device 100 comprises a means for storing and communicating payment information. For example, the device 100 addresses, with communication means 155 to a server of the bank of the holder of the medium, data from a store cash register 150.

[0054] Alternatively, a user challenge (a procedure known as a "challenge-response") is added regarding the transaction in order to secure the acquisition of biometric data at the hardware level. Thus, a data signature is introduced in order to render ineffective the attack by presentation of data already collected and possibly intercepted.

[0055] We observe, in figure 2 , a method of reading a message 200, which comprises: a step 205 of reading biometric data of a carrier of the medium, a step 210 of capturing, on a medium, a coded message comprising elementary message units, said message having noise consisting of marking defects of the message on said medium which are random and unpredictable for each elementary message unit, a step 215 of reducing noise in the captured message, based on biometric data and a step 220 of decoding at least part of the at least partially denoised message.

[0056] We observe, in figure 3 , a device 300 for marking a message on a medium, which comprises: a means 305 for reading biometric data of a carrier of the medium, a means 310 for creating noise in a coded message, based on biometric data, a means 315 for coding at least part of the message and a means 320 for marking, on a medium, the coded message comprising elementary message units, configured so that said mark presents noise consisting of marking defects of the message on said medium which are random and unpredictable for each elementary message unit.

[0057] In the case where the present invention is implemented to sign an action carried out by the bearer of the medium of the mark representing the message, the medium serves to authenticate the bearer. In the case where the present invention is implemented to sign a document, in addition to authenticating the bearer, the content of the document is authenticated. For example, the device 300 serves, when signing a contract, to associate the identity of the signatories with a trace, for example a summary of the complete content of the contract. Think about signing at the notary

[0058] We observe, in figure 4 , a method 400 for marking a message on a medium, which comprises: a step 405 of reading biometric data of a carrier of the medium, a step 410 of coding at least part of a message, a step 415 of adding noise to the coded message, as a function of biometric data, and a step 420 of marking, on a medium, the coded message comprising elementary message units, during which said mark presents noise consisting of marking defects of the message on said medium which are random and unpredictable for each elementary message unit.

[0059] We observe, in Figure 5 , a mark 500, which represents a noisy coded message based on biometric data and which comprises elementary message units 505. The mark 500 presents noise consisting of marking defects of the message on said support which are random and unpredictable for each elementary message unit.

Claims

1. Device (300) for marking a message on a medium, comprising: - a means (305) for reading biometric data of a holder of the medium; - a means (310) for encoding at least one portion of a message; - a means (315) for adding noise to the encoded message, by applying a mathematical function to the data of the encoded message, a function defined by biometric data; and - a means (320) for marking, on a medium, the noised encoded message comprising elementary message units, configured so that said marking has noise that consists of marking defects of the message on said medium which are random and unpredictable for each elementary message unit.

2. Device (300) according to claim 1, wherein the means (315) for adding noise calculates a biometric data hash and by applying a mathematical function to the data of the encoded message, a function defined by said hash to add noise to the message.

3. Device (100) for reading a message (105), characterised in that it comprises: - a means (110) for reading biometric data (115) of a holder of the medium; - a means (120) for capturing, on a medium (125), a message (130) encoded by a device (300) according to one of claims 1 or 2, an encoded message comprising elementary message units, said message having noise that consists of marking defects of the message on said medium which are random and unpredictable for each elementary message unit; - a means (135) for reducing the noise of the captured message, by applying a mathematical function to the data of the captured message, a function defined by biometric data; and - a means (140) for decoding and correcting errors of at least one portion of the message, said message being at least partly denoised.

4. Device (100) according to claim 3, wherein the decoding means (140) utilises redundancies of the denoised message to determine at least one portion of the encoded message.

5. Device (100) according to one of claims 3 or 4, wherein the biometric data reading means (110) is configured to read a fingerprint of the holder of the medium.

6. Device (100) according to one of claims 3 to 5, wherein the biometric data reading means (110) is configured to capture an image of the face or of the hand of the holder of the medium.

7. Device (100) according to one of claims 3 to 6, wherein the noise reduction means (135) calculates a biometric data hash and applies a mathematical function to the data of the captured message, a function defined by the hash to reduce the noise in the captured message.

8. Device (100) according to claim 7, wherein the noise reduction means (135) utilises a symmetrical hash.

9. Device (100) according to one of claims 3 to 8, wherein the decoding means (140) is configured to decode biometric data.

10. Device (100) according to one of claims 3 to 9, wherein the decoding means (140) is configured to utilise an asymmetric key to access at least one portion of the message.

11. Device (100) according to one of claims 3 to 10, which comprises a means (145) for inputting a secret code by the holder of the medium, the decoding means (140) being configured to utilise said secret code to access at least one portion of the message.

12. Device (100) according to one of claims 3 to 11, which comprises a means (155) for storing and communicating payment information.

13. Method (400) for reading a message on a medium, comprising: - a step (405) of reading biometric data of a holder of the medium; - a step (410) of encoding at least one portion of the message; - a step (415) of adding noise to the encoded message, by applying a mathematical function to the data of the encoded message, a function defined by biometric data; and - a step (420) of marking, on a medium, the noised encoded message comprising elementary message units, during which said marking has noise that consists of marking defects of the message on said medium which are random and unpredictable for each elementary message unit.

14. Method (300) for reading a message, characterised in that it comprises: - a step (305) of reading biometric data of a holder of the medium; - a step (310) of capturing, on a medium, a message encoded by utilising a method (400) according to claim 13, the encoded message comprising elementary message units, said message having noise that consists of marking defects of the message on said medium which are random and unpredictable for each elementary message unit; - a step (315) of reducing the noise of the captured message, by applying a mathematical function to the data of the captured message, a function defined by biometric data; and - a step (320) of decoding and correcting errors of at least one portion of the message, said message being at least partly denoised.