WO2021209412A1

WO2021209412A1 - Method for detecting an attack by presentation for fingerprints

Info

Publication number: WO2021209412A1
Application number: PCT/EP2021/059492
Authority: WO
Inventors: Joannes FALADE; Sandra Cremer; Christophe Rosenberger
Original assignee: Imprimerie Nationale
Priority date: 2020-04-16
Filing date: 2021-04-13
Publication date: 2021-10-21
Also published as: FR3109457A1; EP4136565A1; FR3109457B1

Abstract

The invention relates to a method and a system for detecting an attack by presenting fingerprints on a suitable sensor in which a texture vector and a business vector are concatenated to build a model making it possible to distinguish real fingerprints from fake ones.

Description

DESCRIPTION

PRESENTATION ATTACK DETECTION PROCESS FOR

FINGERPRINTS

The invention relates to a method for verifying whether a fingerprint presented during a check is a real fingerprint or a dummy fingerprint. It is used in particular to detect presentation attacks (false fingers) on fingerprint sensors.

The expression "attack detection by presenting fingerprints" in the present invention means that we will detect whether a fingerprint presented on a suitable sensor is a real fingerprint or a dummy fingerprint, and thus avoid a fraudulent use of a person's identity.

[0003] The fingerprint is one of the most widely used biometric methods to secure access and the issuance of sovereign titles. This massive use of fingerprints has led to the emergence and proliferation of attacks on biometric systems. For example, an individual previously expelled from a country can re-enter the territory by replacing the prints of his right hand with those of his left hand, at an access control gate or by using a finger dummy.

We speak of "level 1" attack, by presentation on the fingerprint sensor, where the impostors deposit false fingers (silicone, latex, wood glue) or dead fingers, on the sensors of fingerprints in order to impersonate or change identity. These attacks take place at the level of the biometric sensor and the impostor will usurp the biometric data of another individual, or create a new biometric data, in order to access confidential information over which he has no rights.

[0005] It is therefore very important to set up an attack detection system by presentation, also called “anti-spoofing” in the prior art. The purpose of this system will be to generate an alert in the presence of a false finger, in order to avoid the issuance or the use of sovereign titles to impostors. Several solutions are proposed in the state of the art for the detection of attack by presentation. These solutions are of two types: the hardware approach and the software approach.

[0007] The hardware approach requires the integration of specific electronic components for fingerprint sensors. It is therefore very expensive, dependent on the types of false fingers when learning the system, and not very industrializable.

[0008] The software approach is the approach most explored in the prior art. It is available between a dynamic approach and a static approach.

[0009] The dynamic software approach consists of capturing multiple images of the fingerprint over a period of finger movement on the sensor, a rotation and a long press of the fingerprint lasting from zero to five seconds. These methods analyze the variations on several successive images. They have the disadvantage of being less precise and above all they require more time during the acquisition of the imprint, which can appeal to an impostor.

[0010] The static software approach is to use a single image of the fingerprint to determine whether it is a real finger or a dummy finger. This is the most popular state-of-the-art approach. Only one image is needed and the acquisition time is thus reduced.

The solutions known in the prior art consider a fingerprint image as any image on which we will apply methods of extracting conventional image texture descriptors before making a decision using a Previously trained “classifier or classifier”. The texture descriptors measure the local variations in intensity on each of the pixels of the image. The measurement of these variations, in a global way, gives the texture of the analyzed image. The texture descriptors are calculated for each of the pixels of the image and correspond to a redefinition of a pixel with respect to its local neighborhood. One of the best known descriptors is the “local binary pattern” known by the abbreviation LBP, acronym for “Local Binary Pattern”. The descriptors are then inserted into a classifier of the support vector machine type better known by the English abbreviation SVM (Support Vector Machine) or neural networks (NNET) which are models of machine learning by machine (or Machine Learning) to learn the factors discriminants on the descriptors. These models use notions of probability calculations to find the set of descriptors allowing the best possible separation between real fingerprints and dummy fingerprints. Once the model is learned on the basis of a chosen descriptor, when a new image arrives at the input of the system, we extract the same type of descriptor, then we submit it to the model in order to make a classification decision of the l 'picture.

[0012] Other descriptors called "deep" or "deep" descriptors are also learned by deep learning, usual in image classification. It should be noted that the “deep” descriptors are more precise and provide better results. However, they take a long time to set up because the model learned is complex.

In summary, the methods of the prior art have one or more of the following drawbacks:

[0014] They are expensive and not very easy to use,

[0015] The fingerprint acquisition time to determine an attack is too long for usual control applications.

The different types of descriptors generally have at least one of the following drawbacks:

- They do not allow the desired precision to be obtained,

- They are complex to implement and provide results that are difficult for humans to understand.

By way of illustration, the document by Xiaofei et al, entitled "Multi-scale local binary pattern with filters for spoof fingerprint detection", Information Sciences 268 (2014) 91-102, and the document by Kumar Abhishek et al , titled "A Minutiae Count Based Method for Fake Fingerprint Detection", Procedia Computer Science 58 (2015) 447 - 452, disclose fingerprint detection methods.

The idea of the present invention is to provide a new method for detecting attack by presentation which will use business descriptors, derived from knowledge of fingerprints, combined with conventional texture descriptors. In the remainder of the description, the expression “business descriptors” denotes descriptors which reflect the characteristics of a detail, which includes the global descriptor and the local descriptor of an imprint, and under the expression “texture or LBP descriptors” the descriptors associated with the texture of the image of the imprint. The minutiae are specific points of the imprint which materialize a particular deformation of a ridge and valley.

The idea of the method according to the invention is based on the exploitation of business descriptors based on statistical estimators of characteristic elements of a fingerprint as well as on the quality of these indices. The method will use minutiae extractors for fingerprints which provide several exploitable information and which will help in the construction of discriminating descriptors to discriminate the real fingerprints and the dummy fingerprints.

The object of the invention relates to a method for detecting an attack by presenting fingerprints comprising at least the following steps:

- Generate an M attack detection model by presentation by performing the following steps:

- Acquire one or more real or dummy labeled imprints using a sensor,

- Determine a file F _qm containing the quality values of an acquired fingerprint,

- Submit an image of said imprint (s) to a texture extractor in order to generate a texture vector Vi (Ibpi, ..., Ibp _m ),

- Submit said image to a minutiae extractor and extract a number n of minutiae M _n , a minutiae being characterized by at least its abscissa, x, its ordinate y, its type t, its orientation Q and its quality index q,

- Calculate the components vm _k forming a business vector V ₂ = (vm-i, ..., vm _k ) containing the result of the four statistical indicators:

The average

The standard deviation

The level of asymmetry of the values around the mean S =

Information on the flattening of the distribution of the variable w

performed for all the minutiae extracted on each of the following variables: the type of minutiae, its orientation, its quality index, the distance separating said minutiae to at least one neighboring minutiae, the number of peaks separating said minutiae M _j to said neighboring thoroughness considered,

- Determine the overall quality Q _g of the imprint by summing the quality values per area of the image of the acquired imprint and taking into account the number of occurrences of the frequency of appearance for each quality value, the quality frequency with index i indicating the number of occurrences i in the file F _gm containing the quality values of the acquired fingerprint and add this overall quality value to the components vm _k ,

- Concatenate the texture vector Vi with the business vector V ₂ containing the properties of the imprint, to form a vector V _c containing the characteristics of the k variables described by the set of minutiae acquired for a given imprint,

- Submit this vector V _c to a discrimination algorithm configured to generate an attack detection model by presentation of fingerprints,

- Submit a new fingerprint acquired by the fingerprint sensor to said fingerprint presentation attack detection model to verify whether said fingerprint is real or fake.

The method can further comprise the following steps:

- Calculate the values of the frequencies (Qi), f (Q ₂ ), f (Û3), f (C), f (Qs) of occurrence for overall quality values varying from 1 to 5 and add said values to the business descriptor vector V ₂ .

According to one embodiment, the method further comprises a step of calculating the overall variation of the directions of minutiae of the imprint by comparing the minutiae read with the variable DQ with the global directions of the imprint contained in a file. and adding the global variation value of the directions to the business descriptor vector V ₂ . The method may further include a step of determining the reading error frequency f ( _eri ect) on the directions of the minutiae and adding this value to the business descriptor vector V ₂ .

The method can also include a step of calculating the number of empty zones, N ( _zv ) defined by the number of occurrences of a value equal to "1" where the value "1" indicates a zone read as empty of l 'fingerprint, and adding this value to the business descriptor vector V ₂ .

[0026] According to an alternative embodiment, the statistical descriptors will be applied to the distance characteristic of a minutiae, considering the distance taken with respect to the three minutiae considered as the closest neighbors.

The invention also relates to an attack detection system by presenting fingerprints comprising a fingerprint sensor connected to a texture extractor configured to generate a texture vector Vi, a concatenation device, a configured discrimination algorithm. to generate a detection model by attacks, and a comparison device, characterized in that it further comprises the following elements:

- a minutiae extraction module configured for:

- execute the four statistical indicators:

The average

The standard deviation

The level of asymmetry of the values around the mean S =

Information on the flattening of the distribution of the variable w

performed for all the minutiae extracted on each of the following variables: the type of minutiae, its orientation, its quality index, the distance separating said minutiae to at least one neighboring minutiae, the number of peaks separating said minutiae M _j to said neighboring minutiae considered, in order to calculate the components viri _k , of a business vector V ₂ = (vmi, .. vm _k ), - Calculate the components vrri _k forming a business vector V ₂ = (vmi, vrri _k ) containing the result of the four statistical indicators,

Determine the overall quality Q _g of the imprint by summing the quality values per area of the image of the acquired imprint and taking into account the number of occurrences of the frequency of appearance for each quality value, the frequency of quality index i indicating the number of occurrences i in the file F _gm containing the quality values of the acquired fingerprint, and add this overall quality value to the components vm _{k in} order to generate a business vector V ₂ containing the properties of the impression,

- A module configured to concatenate the texture vector Vi with the vector V ₂ , to form a vector V _c containing the characteristics of the k variables described by the set of minutiae acquired for a given fingerprint,

- Said discrimination algorithm being configured to generate an attack detection model by presenting fingerprints,

- Said comparator being configured to compare a fingerprint to be verified with the attack detection model and decide whether the fingerprint is a real fingerprint or a dummy fingerprint.

The minutiae extractor module is, for example, configured to determine at least one of the following values:

- The values of the frequencies f (Qi), f (Q ₂ ), ί (<¾), f (Û4), f (Qs) of occurrence for values of overall quality varying from 1 to 5,

- The overall variation of the minutiae directions of the imprint,

- The reading error frequency on the directions of the minutiae f ( _er iect) and the number of empty zones, N ( _zv ), on the image of the imprint,

- The number of empty areas, N ( _zv ), on the image of the imprint,

- The addition of one or more of these values to the business descriptor vector V ₂ .

The discrimination algorithm used is for example an automatic machine learning model of SVM type or of neural networks type. Other characteristics, details and advantages of the invention will emerge on reading the description given with reference to the appended drawings given by way of non-limiting example and which represent, respectively:

[0031] [Fig.1] illustrates an example of an architecture allowing the implementation of the method according to the invention,

[0032] [Fig.2] a representation table of an imprint by all the minutiae extracted with the variables characteristic of its local behavior,

[0033] [Fig.3] an illustration of "holes" between the ridges and valleys of a footprint,

[0034] [Fig.4] an example of a file bearing the quality indices by area of the image of a fingerprint,

[0035] [Fig.5] a sequence of steps of the method according to the invention using a combination of texture descriptors and descriptors of descriptive business statistics in the same vector to differentiate dummy fingerprints from real fingerprints.

In order to better understand the method implemented by the invention, the following example is given to detect whether a fingerprint acquired by a fingerprint reader is a dummy fingerprint or a real fingerprint. The method is based in particular on the concatenation of conventional texture descriptors with business descriptors based on statistics of the biometric data. This will advantageously improve the precision of the classifier and therefore the control of the "veracity" of an imprint, real imprint or dummy imprint. For the construction of the model, the process will use labeled imprints, that is to say imprints of which we know whether they are real or fictitious. The method will use for this construction, a sufficient number of fingerprints, in the sense usually used for the construction of databases. Once the G model is built and learned by a system, it can be applied to a single footprint to decide whether it is real or dummy.

FIG. 1 illustrates an example of a system architecture according to the invention comprising a fingerprint sensor 10 connected to a processing module 20 of the data acquired by the fingerprint sensor. The processing module 20 comprises a first module 21, texture extractor, configured to determine image texture descriptors, a second module 22 configured to process the data of the acquired footprint, in order to define business descriptors complementary to the statistical descriptors as will be detailed below. This module 22 contains a minutiae extractor whose statistical indicators are used to produce business descriptors which will be combined by those skilled in the art with the texture descriptors.

The fingerprint sensor will allow the taking of labeled fingerprints for the construction of the model G, during a first phase I of the process, then the capture of a fingerprint whose authenticity is to be verified, during a second phase, phase II of the process.

The texture extractor 21 consists of descriptors with local binary patterns or LBP (Local Binary Pattern). The output of the texture extractor 21 can be written in the form of a texture vector LBP, Vi = (lbp-i, ... lbp _m ) with m = 59 for example. The output of the second module 22 corresponding to the business vector (local and global) will be written in the following form V ₂ = (vm-i,., Vm _n ) with n = 49, for example.

The texture vector Vi and the business vector V ₂ will be transmitted to a concatenation module 23 in order to generate a vector V _c resulting from the concatenation V _c = (vm-i, ....,. Nitΐ _h , Ibpi, .. ., lbp _m ). The concatenated vector is subjected to a discriminating algorithm 24 in order to generate a model for verifying the veracity of an imprint 25, phase I of the process.

The generated G model will be used to decide whether an imprint is a real imprint or a dummy imprint, phase II of the process.

For this, the system according to the invention comprises a comparator 26 taking as input data from a fingerprint acquired on the fingerprint sensor 20 and the data of the model 25 to detect whether the captured fingerprint is a real fingerprint. or a dummy imprint. The result can be displayed on a screen of an enrollment station or the result of the comparison will generate an alarm signal at an access control gate in the event of identity theft.

The detection system according to the invention can be implemented in the enrollment stations available in town hall to apply for a passport. These stations allow the capture of images of the ten fingerprints of the applicant. The image of each imprint can thus be processed by comparator 26. If the result of this comparator indicates that one of the imprints is dummy, then the town hall officer carrying out the enrollment will receive an alert in order to be able to interrupt the passport application process.

Algorithm 24 is a classifier of SVM (Support Vector Machine) type or of neural networks or NNET type which are machine learning models for learning the discriminating factors on the descriptors. Any supervised learning technique algorithm intended to solve discrimination problems can be used. These models use notions of probability calculations to find the set of descriptors allowing the best possible separation between false fingerprints and real ones. These algorithms are known to those skilled in the art and will not be detailed. The method according to the invention “injects” at the input of these discriminating algorithms, the vector V _c resulting from the concatenation of the texture vector Vi and the business vector V ₂ .

[0044] A footprint is comparable to an alternating surface of a set of ridges and valleys parallel to most regions in the footprint. The deformations between ridges and valleys form the minutiae which is the most stable representation used for comparison and identification of fingerprints. The minutiae represent local discontinuities and mark the positions where a ridge ends or branches off. On an imprint, it is possible to detect between [1, 150] minutiae knowing that fourteen minutiae are generally sufficient to perform a comparison.

A minutia m (x, y, t, Q, q, dst _1; nb_cr _1; dst ₂ , dst_cr ₂ , dst ₃ , bd_cr ₃ ) is characterized by its abscissa, its ordinate y, its type t, its orientation Q, the quality index q associated with the thoroughness on the impression. Two types of minutiae are used, ie, bifurcations and endings. The orientation of a minutia is the angle formed by the deviation of the ridge used to identify the minutia from the horizontal. The variables dsti, nb_cn, represent respectively the distance which separates the minutiae from its closest neighbor and the number of peaks which separate them. The indices 2 and 3 in the dsh nb_cn notation represent the same measurements for the second and third closest neighboring minutiae. We will generalize using the notations dsti, nb_cr _h i being the “rank” of the closest neighboring minutia with respect to the concerned minutia. Descriptors related to thoroughness are conventional descriptors used in the field of fingerprint biometrics. They are known to those skilled in the art and will therefore not be detailed.

Thus, an imprint / can be represented on the basis of the aforementioned variables characterizing a thoroughness, by considering three more closely related minutiae:

[0048] / = {m (x, y, t, e, q, dstl, nb_crl, dst2, dst_cr2, dst3, bd_cr3)}, with | / | G [1; 150] and t G {0.1}.

Figure 2 is a local representation of a fingerprint with ten minutiae and the variables described above. In the example, four statistical indicators are used: the mean w, the standard deviation E (w), the "skewness" S, the kurtosis K. The mean represents the indicator of central tendency of a distribution, the standard deviation indicates the fluctuation of different values around the mean value. Skewness indicates the level of skewness of values around the mean while kurtosis gives information on the flattening of the distribution.

We make the following assumptions: we consider each of the characteristics of a minutiae m as a variable, then we will calculate all of these four statistical indicators for all the minutiae of a fingerprint. For example, by taking the quality variable q for a minutiae for a given imprint, q indicates the average level of local quality observed over all of the minutiae detected on the imprint /, then E (q) determines the deviation of the indices individual quality of each minutia compared to the central tendency of quality value. On dummy imprints, the value of the standard deviation E (q) is generally low, because the imprint is homogeneous and its variations are small unlike a real imprint. Similar observations are made on the orientation variable Q which indicates the variation of directions on minutiae and show homogeneity on fake or dummy fingerprints unlike real fingerprints.

The method according to the invention will in particular use the following four statistical indicators, which it will apply selected descriptors w:

mean for w, The method will consider all of the n minutiae Mi, ..., M _n of the captured imprint, then each variable of a minutia (with the exception of x and y), t, Q, q , dsti, nb_cr ₁ dst ₂ , dst_cr ₂ , dst ₃ , bd_cr ₃ . The method will calculate for each of these variables the value of the four aforementioned statistical estimators.

The calculation generates a set of values for all of the n minutiae of the imprint and on each of the minutiae variables:

which form the components of a business vector V ₂ which will be concatenated with the texture vector Vi. The values V _m are linked to the j variables which constitute the minutiae of the imprint.

From the example representing the imprint with ten minutiae and eleven variables per minutiae, there are 36 values v _m1 , ..., v _m36 , forming this business vector V ₂ .

To generate the business vector V _2, the method uses in particular the following business descriptors w:

[0061] The number of minutiae of the imprint,

The frequency of termination type minutiae,

The frequency of appearance type minutiae,

[0064] The average quality of the minutiae taken from the impression,

The standard deviation calculated on the distribution of the minutiae of the imprint,

The "skewness" calculated on the distribution or on the quality parameter,

The average distance of the orientations on the minutiae,

The average distance with an nth nearest neighbor, for a minutiae, or more generally with a minutiae close to the minutia considered, the first neighboring minutiae, the second neighboring minutiae, and so on. At these values, the method will add information on the overall quality Q _g of the imprint.

Whatever the effort made by an attacker to reproduce a fingerprint, a homogeneous character of the peaks induces failures as regards the known normal quality for a fingerprint. In general, the overall quality Q _g of an actual impression will be higher than the overall quality obtained by a dummy impression.

[0071] The difficulty of positioning a finger evenly on the fingerprint sensor results in the appearance of small empty areas on the image of the fingerprint or "holes" in the image. A dummy fingerprint image typically has more "blank areas", 30, shown in Figure 3, than a fingerprint image obtained with an actual fingerprint.

[0072] To the previous parameters, the method can add additional business descriptors making it possible to better differentiate a real footprint and a dummy footprint, to improve and make the decision-making more reliable.

Thus, the method can use:

The quality frequency equal to zero on the overall quality (Figure 4) - f (Q ₀ ), The quality frequency equal to one on the overall quality - f (Qi),

The quality frequency equal to two on the overall quality - f (Q ₂ ),

The quality frequency equal to three on the overall quality - f (Q ₃ ),

The quality frequency equal to four on the overall quality - f (Û4),

The quality frequency equal to five on the overall quality - f (Qs),

The overall total quality of the imprint Q _tg , which corresponds to the sum of the qualities obtained for all the frequencies f (Q ₀ ), f (Qi), f (Q ₂ ), f (Û3), f (Û4), f (Qs),

The overall variation of the directions of minutiae on the recorded imprint, The frequency of reading error on the directions of the minutiae, f ( _eri ect),

The number of empty areas or holes present on an imprint, N ( _zv ),

The variation of the directions of thoroughness which includes: -Dq gives the variation linked only to the minutiae of the imprint,

Reading a file (.dm) which contains the overall directions of the imprint (peak and valley directions) and reading this file gives a variable of directions complementary to those read specifically on the minutiae with the DQ variable.

The four statistical indicators presented above will not be applied to these latter descriptors.

The output of the minutiae extractor 22 generates several files:

A file F _qm containing the quality values of the acquired fingerprint image. The quality will be read in this file, an example of which is illustrated in FIG. 4. This file contains quality values varying from 0 to 5 per zone of the image, the value of 5 being given by way of illustration. Thus, the overall quality of a print image is extracted by adding up all the quality values per area of the image. Then then, for each of the values between 0 and 5, the number of occurrences is counted which represents the frequency of occurrence for each quality value. The quality frequency equal to 2 indicates the total number of occurrences of "2" found in the file F _qm ;

A Fi _fm file which contains values "0" and "1" is representative of a map of the empty areas of the footprint or Low Flow Map. This file contains values of 0 and 1 only where 1 indicates an area read as empty in the footprint. By counting the number of occurrences of 1, we obtain the total number of empty areas of the imprint;

An F _hC m file representative of a high curvature map or High Curvature Map. This file makes it possible to count other additional singular points of the indentations called the “delta”, center and loop. Thus, the .hem file contains 0s and 1s with the frequency of 1s which indicates the presence of singular points on the imprint;

An F _{| Cm} file representative of a low contrast or Low Contrast Map and also contains “0” and “1”. The value 1 indicates the zones of strong contrasts indicating the presence of finger. In this file, the frequency of “1's” is counted to construct the contrast variable useful for identifying the total space of the imprint; The file F _d m file contains information on the direction of the ridges and valleys of the footprint. The global variation of the directions on the footprint, then the associated reading errors are read in the .dm file which stands for Direction Map. This file contains values from -1 to 15. The value "-1" indicates an inability to read the direction of the imprint. Thus, by counting all the occurrences of the values at “-1”, the descriptor of the frequency of the minutiae direction errors is formed. The remainder of values from 0 to 15 indicate a read direction value. Thus, by reading the file F _dm , if at a stage n the value of direction is equal to d and at stage n + 1 the value is different from d, then this variation corresponds to a change of direction and the method accounts for this change of direction. Otherwise there is no change of direction. By going through all the values of the file, we obtain the global change of direction on the imprint which forms the descriptor global variation of the directions of the imprint.

The method will count in the files F _{| Cm} , F _d m, Fi _fm , F _hC m, the digital values supplied by the minutiae extractor 22. For each reading of these files, line by line, a change of value corresponds to a variation. Frequency is obtained by counting the number of times a value appears. Direction errors correspond to values equal to "-1" in the file F _d m · The line-by-line scan takes into account the entire footprint, counting the frequencies of an entire file as well as their occurrence.

FIG. 5 illustrates a succession of steps implemented by the method according to the invention.

[0096] During phase I for the construction of a model from labeled fingerprints,

The first step 51 consists in acquiring several labeled fingerprints as defined above,

During a second step 52, an algorithm will extract a first texture vector, according to a known method such as LBPs for example.

During a third step 53 which is carried out in parallel with the second step 52, the second module will extract statistical descriptors from the local and global information supplied by a minutiae extractor in order to generate indicators allowing the construction of a business vector (local and global),

[0100] During a fourth step 54, the first texture vector and the business vector are concatenated to generate a combination result vector,

[0101] In a fifth step 55, the method executes an SVM-type algorithm for learning a decision model which will make it possible to discriminate a real fingerprint and a dummy fingerprint. To do this, the model will use a database containing several real footprints and several dummy footprints. For each of these imprints, steps 51, 52, 53 and 54 are carried out then at 55, all these extractions are automatically learned by the SVM which produces the separating model of a real imprint from a dummy imprint.

[0102] The sixth step 56 generates a model which will be used to perform a comparison, in a seventh step 57, with acquired fingerprints in order to determine whether they are dummy fingerprints or real fingerprints.

[0103] During phase II, the method will capture an imprint to be verified, i.e., which we seek to verify if it is a real or artificial imprint. The imprint is subjected to the model generated by the steps explained above, using a comparison technique known to those skilled in the art.

[0104] Detailed description of step 53

[0105] Step 53 for constructing the business vector comprises at least the following steps.

[0106] The four statistical indicators, the mean, the standard deviation, the "skewness", the "kurtosis" are performed, for all the minutiae extracted on each of the following variables:

The type of the minutia, its orientation, its quality index, its distance separating a minutia from a neighboring minutia, the number of peaks separating a minutia M from a neighboring minutia considered, in order to generate a business vector V _m ,

[0108] Determine an overall quality value of the imprint captured, Define a vector containing the result of the application of the statistical indicators and the value of the overall quality of the imprint, for the n minutiae for each of the j variables,

In order to improve the detection method, step 53 of constructing the business vector can add to the business descriptor vector V ₂ one or more of the following values:

The overall total quality of the print Q _tg , the quality frequency equal to zero over the overall quality f (Q ₀ ), or equal to one f (Qi), or equal to two f (Q ₂ ), or equal to three f (<¾), or equal to four f (C), or equal to five f (Q ₅ ), the global variation of the directions of minutiae on the recorded imprint, the frequency of reading error on the directions of the minutiae, f ( _er iect), the number of empty zones or holes present on an imprint, N ( _zv ).

[0112] The descriptor vector thus formed will be concatenated with the texture descriptor vector before being transmitted to the learning algorithm to generate an attack detection model by presenting fingerprints.

As described above, the method will generate a business vector which will be concatenated with the texture vector.

According to a first embodiment, to generate this business vector, the method considers the following parameters:

[0115] During a step of checking the "veracity" of a fingerprint, a person presents his fingerprint on the fingerprint reader, the fingerprint is analyzed according to the processing chain explained above. The result is submitted to the model in order to decide whether or not the imprint is true. If the fingerprint is considered an actual fingerprint, then the identity verification process can continue.

[0116] The method according to the invention has the following advantages in particular:

The information from a thoroughness extractor and the use of descriptors based on the estimators of descriptive statistics of mean, variance, skewness, kurtosis make it possible in particular to obtain more precision for validating or rejecting a fingerprint. as a real fingerprint or a dummy fingerprint, a quick check that can be used in real-time verification systems.

Claims

1. Attack detection method by presenting fingerprints comprising at least the following steps:

- Generate an attack detection model M (25) by presentation by performing the following steps:

- Acquire (51) one or more real or dummy labeled imprints using a sensor (10),

- Submit an image of said imprint to a texture extractor (21) in order to generate a texture vector Vi (lbp-i, .. lbp _m ), (52),

- Submit said image to a minutiae extractor (22) and extract (53) a number n of minutiae M _n , a minutiae being characterized by at least its abscissa, x, its ordinate y, its type t, its orientation Q and its quality index q,

- Calculate the components vm _k forming a business vector V ₂ = (vm ₁ , .., vm _k ) containing the result of the four statistical indicators:

The average

The standard deviation

The level of asymmetry of the values around the mean S =

Information on the flattening of the distribution of the variable w

- Determine the overall quality Q _g of the imprint by summing the quality values per area of the image of the acquired imprint and taking into account the number of occurrences of the frequency of appearance for each quality value, the quality frequency of index i indicating the number of occurrences i in the file F _qm containing the quality values of the acquired fingerprint and add this value of global quality with the components vm _k ,

- Concatenate (54) the texture vector Vi with the business vector V ₂ containing the properties of the imprint, to form a vector V _c containing the characteristics of the k variables described by the set of minutiae acquired for a given imprint,

- Submit (55) this vector V _c to a discrimination algorithm (24) configured to generate (56) an attack detection model M by presentation of fingerprints,

- Submit a new fingerprint acquired by the fingerprint sensor to said fingerprint presentation attack detection model in order to verify (57) whether said fingerprint is real or fictitious.

2. Method according to claim 1 characterized in that it further comprises the following steps: calculating the values of the frequencies (Qi), f (Q ₂ ), ί (<¾), - f (C), - f ( Qs) of appearance for global quality values varying from 1 to 5 and add said values to the business descriptor vector V ₂ ,

3. Method according to one of claims 1 or 2 characterized in that it further comprises a step of calculating the overall variation of the directions of minutiae of the imprint by comparing the minutiae read with the variable DQ to the overall directions of the footprint contained in a file and the addition of the global variation value of the directions to the business descriptor vector V ₂ .

4. Method according to one of claims 1 to 3 characterized in that it further comprises a step of determining the reading error frequency f ( _er iect) on the directions of the minutiae and adding this value. to the business descriptor vector V ₂ .

5. Method according to one of claims 1 to 4 characterized in that it further comprises a step of calculating the number of empty areas, N ( _zv ) defined by the number of occurrences of value equal to "1" where the value “1” indicates a zone read as empty of the footprint and the addition of this value to the business descriptor vector V ₂ .

6. Method according to one of the preceding claims, characterized in that the statistical descriptors are applied to the three distances of a minutia corresponding to its three closest neighbors.

7. Fingerprint presentation attack detection system comprising a fingerprint sensor (10) connected to a texture extractor (21) configured to generate a texture vector Vi, a concatenation device (23), an algorithm for discrimination (24) configured to generate an attack detection model M, and a comparison device (26), characterized in that it further comprises the following elements:

- A minutiae extraction module (22) configured for:

- Execute the four statistical indicators:

The average

The standard deviation

The level of asymmetry of the values around the mean S = _>

Information on the flattening of the distribution of the variable w

performed for all the minutiae extracted on each of the following variables: the type of minutiae, its orientation, its quality index, the distance separating said minutiae to at least one neighboring minutiae, the number of peaks separating said minutiae M _j to said neighboring minutiae considered, in order to calculate the components vm _k , of a business vector V ₂ = (vm ₁ , .. vm _k ),

- Calculate the components vm _k forming a business vector V ₂ = (vmi, .., vm _k ) containing the result of the four statistical indicators,

- Determine the overall quality Q _g of the imprint by summing the quality values per area of the image of the acquired imprint and taking into account the number of occurrences of the frequency of appearance for each quality value, the quality frequency of index i indicating the number of occurrences i in the file F _qm containing the quality values of the acquired fingerprint, and add this overall quality value to the components vm _{k in} order to generate a business vector V ₂ containing the properties of the imprint, - A module configured to concatenate (54) the texture vector Vi with the business vector V ₂ , to form a vector V _c containing the characteristics of the k variables described by the set of minutiae acquired for a given fingerprint,

- Said discrimination algorithm (24) being configured to generate (56) an attack detection model M by presenting fingerprints,

- Said comparator (26) being configured to compare a fingerprint to be verified with the attack detection model and decide whether the fingerprint is a real fingerprint or a dummy fingerprint.

8. Detection system according to claim 7 characterized in that the minutiae extractor module (22) is configured to determine at least one of the following values:

- The values of the frequencies f (Qi), f (Q ₂ ), ί (<¾), - f (Û4), - f (Qs) of occurrence for values of overall quality varying from 1 to 5,

- The overall variation of the minutiae directions of the imprint,

- The number of empty areas, N ( _zv ), on the image of the imprint,

- The addition of one or more of these values to the business descriptor vector V _2.

9. Detection system according to one of claims 7 or 8 characterized in that the discrimination algorithm (24) is a machine learning model of SVM type or neural networks.