FR2976106A1 - SYSTEM AND METHOD FOR CONTROLLING THE ACCESS OF AN INDIVIDUAL TO A CONTROLLED ACCESS AREA - Google Patents

Abstract

The invention relates to a system for controlling the physical access of an individual to a controlled access area, the system comprising - at least one acquisition device adapted to capture at least one image or a stream of images an access corridor in which an individual wishing to access the controlled access zone moves before reaching said zone, - a computing platform, to which are sent the image or images captured by the acquisition device, the computing platform being adapted to implement on the one or more images identification processing of the individuals appearing on the one or more images and to allow access to the controlled access zone to an individual moving in the corridor, or to refuse him this access, depending on the result of said identification processing, the system being characterized in that it comprises at the end of the access corridor at least one mirror which reflects a view of the neck access port on the acquisition device, the acquisition device being oriented towards said mirror or adapted to be directed towards it, the captured image (s) of the access corridor in which an individual wishing access to the controlled access area moves before reaching it, being images reflected by the mirror (s).

Description

FIELD OF THE INVENTION

The object of the present invention is generally in the field of control and recognition of individuals, mainly by means of biometric information, in particular by means of face or iris recognition. The invention may in particular be implemented near controlled access areas in which only the identified and authorized persons can enter.

STATE OF THE ART

Access control systems in limited access areas must ensure quick and reliable identification of persons wishing to enter these areas, and therefore they use face identification systems on the fly most of the time, in which the detection and identification of people is automatic, so they do not have to stop to be identified. For this purpose, systems for identifying persons are known, in which an access zone is located at the end of an access corridor, whether or not there are walls, and acquisition devices such as cameras are positioned laterally on walls, walls or columns of the access corridor and are oriented towards the people they detect.

This type of solution, however, poses many problems. Indeed, to perform an identification by recognition of physical characteristics, such as for example the face, it is preferable to take frontal images of the individual to identify. However, if the cameras are positioned on the side and oriented towards the entry of the access corridor, from which the individuals to identify come, they have only a relatively short time to capture the images. In fact, when the individuals approach the cameras, progressing towards the door to be crossed, the angle of view of the camera increases and the physical characteristics are seen more and more side. For example, beyond a certain position, the cameras can no longer detect the two eyes of the individual, which is necessary to take a frontal image. Added to this problem is the fact that the capture of images on the fly causes problems of motion blur of people, and a variation of lighting as a function of the position in space.

To overcome these problems, a plurality of cameras are sometimes used, arranged on two walls, columns or side walls of the access corridor and used to carry out a three-dimensional reconstruction of the individual who wishes to access the restricted access area. but this reconstruction is long and difficult to implement. It does not allow the speed required to regulate the flow of people.

An alternative to this solution has been proposed, in which the cameras are provided with a high-zoom optics, to be able to capture the face of an individual to identify as soon as it enters the access corridor. However, such a system poses many problems. In the first place it is difficult to deploy on a site. In addition, an individual to be identified using the zoom may be obscured by another individual who is closer to the cameras. In addition, the illumination of individuals at this distance is very variable and difficult to control.

PRESENTATION OF THE INVENTION

Accordingly, it is an object of the present invention to provide a system for on-the-fly control of persons preparing to pass through a control gate without this control hindering their progress.

Another object of the invention is to solve the problems of variations in image brightness and pose, depending on the position of people in the access corridor upstream of the limited access area.

In this regard, the invention proposes a system for controlling the physical access of an individual to a controlled access zone, the system comprising at least one acquisition device adapted to capture at least one image or a data stream. images of an access corridor in which an individual wishing to access the controlled access zone moves before reaching said zone, a computing platform, to which are sent the image or images captured by the acquisition device the computing platform being adapted to implement on the one or more images identification processing of the individuals appearing on the one or more images and to authorize access to the controlled access zone to an individual moving in the corridor, or deny this access, depending on the result of the identification process. According to the invention, the system comprises, at the end of the access corridor, at least one mirror which reflects a view of the access corridor on the acquisition device, the acquisition device being oriented towards said mirror or adapted to be directed towards it, and the captured image (s) of the access corridor in which an individual wishing to access the controlled access area moves before reaching it, being images reflected by the mirror (s).

The control system proposed by the invention may further comprise at least one of the following features: it comprises a plurality of mirrors arranged in different orientations in order to reflect different views of the access corridor on the device or devices. acquisition, or comprises a single mirror, continuously deformed, it comprises a plurality of acquisition devices, each being oriented towards mirrors or mirror parts of different orientations, the acquisition device or devices are arranged laterally in the access corridor, access control of the controlled access zone is achieved by means of a door positioned at the end of the access corridor comprising the mirror or mirrors, and the opening of which is conditioned the results of the identification processing of the individuals to control the mirror or mirrors are arranged on the door, or are arranged at the end of the access corridor, the door being arranged laterally with respect thereto, at least one mirror is disposed at least one mark for calibrating at least one of the acquisition devices. The invention also relates to a method of controlling individuals used in a control system according to the invention.

DESCRIPTION OF THE FIGURES Other characteristics, objects and advantages of the present invention will appear on reading the detailed description which follows, with reference to the appended figures, given by way of non-limiting examples and in which: FIGS. and 1b represent embodiments of the individual control system provided by the invention; FIG. 2 schematically illustrates the elements used to perform the identification of an individual in the control system proposed by the invention. FIG. 3 schematically represents the steps of the method of controlling individuals proposed by the invention.

DETAILED DESCRIPTION OF THE INVENTION

Referring to Figures 1a and 1b, there is shown a control system of individuals. It is preferably located at the entrance to a limited access zone Z, in which only authorized individuals can enter. Most often the regulation of access is provided by a control gate 130 whose opening is conditioned to the identification of individuals wishing to enter the zone Z.

Upstream of the control gate 130 may be an access corridor 100, which may be embodied by two side walls 110, or walls, ropes, or columns. Access corridor is generally the area in which the individual I must evolve before reaching the zone Z limited access, without this area necessarily being bounded by walls or walls. In Figure la, the door 130 is disposed at the bottom of the corridor 100, and faces the entrance thereof. Alternatively, as illustrated in FIG. 1b, at the bottom of the corridor 100 is a rear wall 120 and the door 130 is located laterally, arranged in one of the walls 110.

In order to proceed to the identification of the individual I, an image or a video stream of the individual 1 is acquired by at least one acquisition system, and sent to a computing platform 220, when it is in front of the gate 130. The computing platform 220 analyzes said image or said video stream in order to identify the individual I, and depending on this identification, controls the opening of the door 130, or on the contrary refuses this opening.

To do this, on the rear wall 120, or on the door 130 if it faces the entrance of the corridor 100 is at least one mirror 140 which reflects a view of the access corridor 100. The mirror 140 can cover all or part of the rear wall 120, or all or part of the door 130.

Furthermore, in the access corridor 100 is installed at least one acquisition system 200 oriented towards at least one mirror 140, and adapted to capture an image or a video stream of the image of the corridor 100 reflected by the mirror 140. acquisition device is typically a camera that can acquire images in the visible spectrum, but also possibly in the infrared spectrum. It can still be a digital camera or a thermal camera. The mirror 140 and the acquisition device 200 are preferably oriented so that, in the image of the corridor reflected by the mirror, the individual wishing to access the access zone Z appears. Thus, the acquisition device or devices 200 can proceed to the acquisition of one or more physical characteristics of the individual I that make it possible to proceed to its identification. The physical characteristics can be any visual or thermal information relating to the individual I.

For example, the access corridor 100 may be horizontal, and the mirror is disposed at the bottom of the corridor 100, vertically and oriented towards the entrance of the corridor 100, so that the individual I is always facing the mirror when approach to the limited access zone Z. The acquisition device or devices 200 can be arranged at eye level of the individual I, and oriented towards the mirror with an angle between the acquisition device and the mirror normal. less than 90 °. This case is shown schematically in FIG. 1. Optionally, there may be at least one other acquisition device 201 pointed directly to a point of the access corridor 100 through which the individual 1 passes in order to directly capture an image or an image. video stream of the individual. Optionally, a plurality of mirrors 140 may be disposed on the wall 120 or on the door 130. By way of non-limiting example, there is shown in Figure lb four mirrors 140A, 140B, 140C and 140D, but the number is any. It is even possible to use a single deformed mirror continuously rather than a plurality of mirrors.

The mirrors 140 may be oriented differently from each other, to reflect views of the access corridor at different angles, and thus views of the individual I, when he crosses the corridor, at different angles. Depending on the orientation of the mirrors with respect to their position on the wall 120 or the door 130, they can also make it possible to obtain a maximum of quasi-frontal views of the individual I. The control system may also include a plurality of acquisition devices, three of which are referenced by the numbers 200A, 200B, and 200C are shown by way of non-limiting example in FIG. 1b. Each acquisition device 200 can point to one or more mirrors 140 and acquire images of the corridor 100 reflected in each of the mirrors 140. If the control system comprises several acquisition devices 200, they may or may not be synchronized by a synchronization module 210, in order to trigger the taking of image or video stream at the same time for all. The acquisition devices 200, 201 are connected by any means known per se to the computing platform 220. For example, they can be connected by a wired connection or a wireless network.

The computing platform 220 processes the images or video streams acquired by the acquisition devices 200, 201 and proceeds to the identification of the individual I by comparison of said images or video streams with other images or video streams. a database 230, and whether or not the opening of the door 130 according to this identification. The database 230 can be stored by the computing platform 220, or alternatively it can be at a distance from it and be connected to the computing platform 220 by any known means. In the case where several acquisition devices 200 or 201 send one or more images or video streams to the computing platform 220, the latter can select a single image (or a single video stream), the selection being made on criteria such as resolution, or on biometric quality criteria or less facial deformity. Alternatively, if the control system comprises a synchronization module 210 several images acquired at the same time can be used to implement a three-dimensional reconstruction of the individual. Alternatively, a three-dimensional reconstruction can be performed from one or more views of an acquisition device, and possibly without synchronization, depending on the algorithms used. Finally, on each mirror 140 can be positioned at least one marker 141, preferably positioned so as not to hide the reflection of the individuals I, and allowing the acquisition devices to perform a continuous calibration. The calibration may be optical, to correct certain deformations of the images due to the mirrors, and / or geometric, to facilitate the reconstruction if necessary in three dimensions. We can refer to the following publications for more details on the calibration that can be implemented: - Z. Zhang. "A new flexible technique for camera calibration". IEEE Transactions on Pattern Analysis and Machine Intelligence, 22 (11): 1330-1334, 2000. - Z. Zhang. "Flexible Camera Calibration By Viewing a Plane From Unknown Orientations". International Conference on Computer Vision (ICCV'99), Corfu, Greece, pages 666-673, September 1999.

We will now describe a control method of an individual implemented by the system described above, illustrated in Figure 3. The interactions between the various elements of the control system are also shown in Figure 2.

Acquiring an image or a video stream.

When an individual I enters the access corridor 100, in order to access the limited access zone Z and if necessary to pass through the door 130, its image is reflected by the mirror (s) 140 , and at least one acquisition device 200 can capture an image, preferably in digital format, or a video stream of the individual during an acquisition step 310.

Optionally, a preliminary step 300 of preparing for the acquisition can take place. This step 300 may contain a sub-step 301, during which each acquisition device 200 can detect the individual I, for example by detecting the eyes of the individual I on the mirror 140, and follow its movement, c that is to say its approximation of the mirror 140 and therefore its magnification of the mirror 140, in order to start the acquisition 310 of an image or a video stream at a time when the reflection of the individual presents specific characteristics in accordance with previously established criteria. The characteristics may relate to a size or resolution of the reflection in the image taken by the acquisition device, or a position of the individual I in the corridor 100, or it may be certain light conditions, the expression of the individual, the pose of the face, the reflections on the glasses if necessary, etc. Another sub-step, 302, of the acquisition preparation step 300 can take place if the acquisition system comprises several acquisition devices 200 and a synchronization module 210 for synchronizing these acquisition devices. During this step 302, called synchronization, this module 210 can send a synchronization signal SI to all the acquisition devices so that the acquisitions 310 of all the acquisition devices are done at the same time, as illustrated. in FIG. 2. This simultaneous acquisition makes it possible, if necessary, to reconstruct a three-dimensional image of the face of the individual I to proceed with his identification. Another acquisition preparation substep 303 can take place if, on at least one mirror 140, there are arranged pins 141 enabling the self-calibration of the acquisition devices 200. In this step 203, the acquisition devices can perform this self-calibration.

Transfer of the Image or Video Stream Then, each acquisition device 200 transfers to the computing platform 220 the video stream or the image S2 acquired by any known means during a step 320. Optionally, each device acquisition 200 can convert the video stream or the image S2 before transfer in a format allowing a faster and / or more secure transfer to the computing platform 220, during a step 311. The conversion may for example be a compression in an MPEG format (MPEG2, MPEG4, or MPEG4AVC) Identification of the individual from the image or the video stream When the computing platform 220 has received the image or the video stream S2, it uses to identify the individual I during a step 330 of identification. If the control system comprises a plurality of acquisition devices 200 which have each sent an image or a video stream S2 to the computing platform 220, the step 330 may comprise a first substep 331 for selecting a video source. image or a video stream used to proceed with the identification. For example, if the identification is based on face recognition, the selected images or video streams must represent the face of the individual I. In addition, the selection may be implemented by means of a particulate filter and / or or an algorithm for detecting the eyes of the individual. For more details on how the selection is implemented, reference can be made to the publication "Robust Real-Time Object Detection", Paul Viola, Michael Jones, IJCV 2001.

The selection can also relate to several images or several video streams in order to then proceed to a three-dimensional reconstruction of the individual I. This selection can be performed according to criteria such as the resolution, or on biometric quality criteria, of detection quality of the eyes, or the number of pixels between the eyes, etc. Once one or more images or video streams have been selected, the computing platform 220 performs a processing 332 and a coding 333 thereof. The image processing can be performed to detect a face or iris in an image to make biometric recognition from them, or more generally, to detect any type of visual biometric information, infrared, or thermal of the individual. In the case of a video stream, the treatment may even be for the analysis of an approach of the individual I, or to extract particular images of the video. Operations on the images can also be performed to correct certain deformations, related for example to the relative positions of the acquisition devices 200 and the mirrors 140. Such image recovery operations may be, by way of a non-limiting example, a affine transformation, or the inversion of an image by symmetry with respect to a vertical median axis, since the mirrors 140 invert the images. In the case where several images or video streams have been selected, the treatment of step 332 may also consist of a 3D reconstitution of the individual I, or more particularly of his face. The coding step 333, also performed by the computing platform 220, can take place before or after the processing step 332. It consists in coding the images or the video streams S2 before or after processing, in order to obtain a representative signal of these images, and in particular representative of the face of the individual I. It is these signals S3 which are used to carry out the identification of the individual, by comparison with other representative signals of faces. Examples of coding techniques that can be implemented here are described in the publication M. Turk and A. Pentland (1991). "Face recognition using eigenfaces". Proc. IEEE, Conference on Computer Vision and Pattern Recognition. pp. 586-591.

The computing platform then performs a mapping step 334 between the representative signals of the face of the individual S3 and other representative signals of faces contained in the database 230 and coded in the same way. Alternatively, the comparison of the signals S3 can be made with a scanned image in the corridor, at the entrance of the door and coded in the same manner as the images to be identified S3. This image may for example be on a badge or passport of the individual I (this option is not shown in the figures). In all cases, the computing platform evaluates the correspondence between signals representative of identically coded images, generating a similarity score between two images. In a manner known to those skilled in the art, if a pair of images obtains a similarity score greater than a previously established threshold, it is considered that the two images come from the same individual, with a potential error rate. less than an equally established level. The performance of face recognition algorithms is presented in the NIST report, Still-Image Face Recognition Algorithms (NIST Interagency Report 7709, Patrick J. Grother, George W. Quinn, P. Jonathon Philips, August 23, 2010) . Thus, the comparison of the signals S3 with a signal database or with a scanned and coded image makes it possible, in the first case, if the individual 1 has been identified in the database to identify him, and in the second case, case, it allows to determine if the individual I actually corresponds to the photo he holds. From this step 334 of matching the computing platform 220 determines during a deduction step 335 whether the access of the individual I is authorized or not, and therefore if it commands or refuses the opening of the door 130. For example, the database 230 may contain the images of the only persons authorized to cross the door 130. Alternatively, it may contain images of persons who are not allowed to enter the door.

Gate control The computing platform 220 then sends a command S4 to the gate 130 in a step 10 of instructions 340 so that it opens or remains blocked, depending on the identity of the gate. individual (I) thus determined.

This control system has many advantages. The presence of mirrors 140 on the wall 120 or the door 130 to which the individual I is heading allows the acquisition devices to have an almost frontal vision of the individual throughout his movement in the corridor. access 100. Therefore, the system can implement identification based on biometric data of the face recognition or iris type.

This further allows the acquisition devices 200 to track the individual I along the entire length of the lane 100, so that the acquisition devices 200 can easily focus on a particular point of the individual I. , for example the eyes, and can trigger the acquisition of images or video streams at an optimal time.

In addition, a mirror naturally attracts the gaze of people, and thus individuals I are more likely to look towards the mirror 140 than to an acquisition device that would be arranged on the side, so that the frontality of the images obtained is assured. Finally, the control method is non-invasive and does not disturb the flow of individuals crossing the corridor 100, since the acquisition devices 200 are positioned laterally with respect to the passage of the individual I. It is also compatible with three-dimensional recognition methods already known or with identification systems comprising acquisition devices directly pointed at the individual crossing the corridor.

Claims (10)

REVENDICATIONS1. A system for controlling the physical access of an individual to a controlled access area, the system comprising - at least one acquisition device (200, 201) adapted to capture at least one image or a stream of images ( S2) of an access corridor in which an individual (I) wishing to access the controlled access zone moves before reaching said zone, a computing platform (220), to which the one or more images are sent ( S2) captured by the acquisition device (200), the computing platform (220) being adapted to implement on the one or more images identification processing of the individuals (I) appearing on the one or more images and allowing the accessing or denying access to the controlled access area to an individual moving in the corridor, based on the result of said identification processing, the system being characterized by comprising at the end of the corridor access, at least one mirror (140) q ui reflects a view of the access corridor on the acquisition device, the acquisition device being oriented towards said mirror or adapted to be oriented towards it, and the captured image (s) of the access corridor wherein an individual (I) wishing to access the controlled access area moves before reaching it, being images reflected by the mirror (s). The control system of claim 1, comprising a plurality of mirrors (140A, 140B, 140C, 140D) arranged in different orientations to reflect different views of the access corridor (100) on the at least one device. acquisition (200), or comprising a single mirror (140) continuously deformed. 3. Control system according to the preceding claim, comprising a plurality of acquisition devices (200A, 200B, 2000), each oriented towards mirrors (140) or mirror parts of different orientations. 4. Control system according to one of the preceding claims, wherein the one or more acquisition devices (200, 201) are arranged laterally in the access corridor. 10 5. Control system according to one of the preceding claims, wherein the access control of the controlled access zone is achieved by means of a door (130) positioned at the end of the access corridor (100). ) comprising the mirror or mirrors, and whose opening is conditioned to the results of the identification processing of the individuals (I) to be controlled. 15 6. Control system according to the preceding claim, wherein the one or more mirrors (140) are arranged on the door (130), or in which they are arranged at the bottom of the access corridor, the door (130) being arranged laterally. compared to these. 7. Control system according to one of the preceding claims, wherein, on at least one mirror (140), is disposed at least one marker (141) for calibrating at least one of the acquisition devices ( 200). 25 8. Method for controlling the physical access of an individual to a controlled access zone, in which: at least one acquisition device (200, 201) captures at least one image or a stream of images (S2) of an access corridor in which an individual (I) wishing to access the controlled access area moves before reaching said area, a computing platform (220), to which are sent the the images (S2) captured by the acquisition device (200) implements, on the one or more said images, an identification processing of the individuals appearing on the one or more said images and, - the computing platform (220) authorizes the access to the controlled access area to an individual moving in the corridor, or denies it access, according to the result of said identification processing, the method being characterized in that it is implemented in a system according to the one of the preceding claims. 9. Control method according to claim 8, implemented in a system according to claim 7, further comprising a step of calibrating (303) acquisition devices (200) with at least one marker (141) arranged on the mirror or mirrors (140), before the acquisition of the images of the individual (I). 10. Control method according to one of claims 8 or 9, wherein the treatment (332) comprises a deformation of the image in order to correct deformations related to the relative positions between the acquisition devices (200) and the mirrors (140).