Detection of fraud for access control system of biometric type

1. A method of fraud detection for an access control system via biometric recognition, comprising the steps:

taking at least two images (S 2 ) of a face presented to the access control system by an individual (I) according to at least two positions of the individual in front of the access control system, or two angles of vision of the individual,

extracting the face from each of the images,

processing the at least two images to determine a score representative of a flatness of the face appearing on the images, and

as a function of the flatness score, detecting any fraud on the part of the individual, said detection conditioning the use of identification treatment of the individual by the access control system,

wherein the step for determination of a score representative of the flatness of the face comprises the steps:

creating a depth map of the face of the individual (I), estimating a median plane of the map of the face, and calculating the variance of the points of the map of the face relative to said median plane of the map of the face,

isolating from the images a block of the face centred on a nose of the individual, creating a depth map of said block, estimating a median plane of the map of said block, and calculating the variance of the points of the map of said block relative to said median plane of the map of said block, and

obtaining a flatness score of the face by linear combination of the variances of the points of the depth maps relative to the respective median planes.

2. The method according to claim 1 , also comprising a standardisation step of the depth map of the face so that the depth map of the face is independent of the angles of vision of the images of the face of the individual.

3. An access control system via biometric recognition, comprising:

at least one acquisition systems ( 12 ) adapted to capture at least two images of a face presented by an individual (I) according to at least two positions or viewing angles of the individual relative to the at least one acquisition system ( 12 ), and

a calculation platform ( 13 ) to which are sent the at least two images (S 2 ) by the at least one acquisition system ( 12 ), the calculation platform ( 13 ) being adapted to process the at least two images to determine a score representative of a flatness of the face appearing on the images, and to

isolate from the images a block of the face centred on a nose of the individual,

create a depth map of the face of the individual (I) and create a depth map of the block centred on the nose of the individual,

estimate, for each map, a median plane of said map,

calculate the variance of the points of each map relative to a median plane of said map, and

calculate a flatness score of the face by linear combination of the variances of the points of the depth maps relative to the respective median planes, and

from the score representative of the flatness of the face detect any fraud on the part of the individual, said detection conditioning the identification of the individual by biometric analysis of the face of the individual.

4. The access control system according to claim 3 , said system being arranged upstream of a regulated access door ( 11 ), and the calculation platform ( 13 ) being adapted to control the opening of the door, and wherein fraud detection and identification of the individual condition opening of the door ( 11 ).