Device for iris recognition when passing
Patent Information
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-04-25
- Publication Date
- 2026-03-04
AI Technical Summary
Existing iris recognition systems that allow for identification while a person is walking are complicated and often produce fewer sharp images due to the use of single powerful cameras, leading to lower frame rates and increased susceptibility to artifacts, which affects the robustness of the identification process.
A device comprising a scene camera with a wide field of view and two iris cameras with smaller fields of view and fixed focal planes, where the iris cameras are height-adjustable and synchronized with the scene camera to capture sharp images of the periocular facial area as a person moves through a detection corridor, eliminating the need for complex camera systems like plenoptic cameras.
This setup increases the frame rate and sharpness of images captured, enhancing the likelihood of successful iris recognition and providing a more robust and reliable identification process while the person is in motion.
Smart Images
Figure EP2024061433_31102024_PF_FP_ABST
Abstract
Description
[0001] Iris recognition device when passing by
[0002] Field of the invention
[0003] The present invention relates to a device for iris recognition of a person as they pass by.
[0004] Background of the invention
[0005] To verify access authorization or to otherwise identify a person, biometric features such as fingerprints, voice recognition or iris recognition, etc. can generally be used.
[0006] In general, iris recognition involves capturing a person's eyes as individual images or within an image of the entire periocular facial area (i.e., the person's eye area), and then comparing the person's iris characteristics with stored data. These images of the person to be identified can be captured, for example, in a stationary device with a corresponding camera. However, such identity verification in a stationary device is time-consuming and not particularly convenient, as the person to be identified cannot simply walk on but must first stop at the stationary device and look into the camera.
[0007] Another option for iris recognition is to monitor a detection corridor through which the person to be identified walks, using appropriate cameras, and to capture the periocular facial area of the person as they walk through the detection corridor. This type of iris recognition is obviously much faster and more convenient, as the person walks through the detection corridor without stopping and is identified. Human behavior such as blinking, blinking away from the cameras, or reflections in glasses make it difficult to capture suitable iris images for successful identification of a person in passing.
[0008] Therefore, such devices usually attempt to capture as many sharp iris images as possible over the largest possible detection area of the person to be identified in order to increase the probability of capturing at least one suitable image for iris recognition.
[0009] Such devices that enable iris recognition in passing typically use a single high-performance camera, such as a 3D camera (e.g., a light-field camera (plenoptic camera)), or a camera with an adjustable working distance / focal plane to enable the iris to be captured at different distances from the camera as the person moves through the detection area. However, such complex systems limit the possible frame rate and therefore produce fewer sharp images. Furthermore, the resulting images are often more susceptible to artifacts that impair image sharpness. High image sharpness is one of the key requirements for robust iris recognition.
[0010] Description
[0011] Accordingly, an object of the present invention is to provide a reliable system for iris recognition in passing that does not require complicated camera systems.
[0012] This object is achieved by the subject matter of the independent claim. Exemplary embodiments emerge from the dependent claims and the following description.
[0013] According to one aspect, a device for walk-by iris recognition is provided. The device comprises a first iris camera with a first field of view and a fixed first focal plane, a second iris camera with a second field of view and a fixed second focal plane, a scene camera with a third field of view, and a controller. The first iris camera and the second iris camera are height-adjustable. The first iris camera, the second iris camera, and the scene camera are connected to the controller.The controller is configured to capture images of a person to be identified that is outside a combined capture range of the first iris camera and the second iris camera by the scene camera, determine a height of a periocular facial region of a person to be identified based on the captured images of the scene camera, adjust a height of the first iris camera and the second iris camera in response to the determined height of the periocular facial region such that the first iris camera and the second iris camera can capture the periocular facial region, and capture the periocular facial region with the first iris camera and the second iris camera, and determine an identity of the person to be identified based on the captured periocular facial region.
[0014] The device according to the invention can, in particular, monitor a detection corridor (as described below) through which persons to be identified pass. The detection corridor can, for example, be a suitably guided walkway or path that guides the person to be identified past the first iris camera, the second iris camera, and the scene camera. In general, however, the detection corridor merely describes the functionally detectable area of the camera arrangement used and does not necessarily have to be physically separated (e.g., by appropriate barriers).
[0015] The scene camera, the first iris camera and the second iris camera are each cameras which have a fixed focal plane, i.e. a fixed working range / working distance from the camera.
[0016] The scene camera monitors in particular an area outside a combined detection range of the first iris camera and the second iris camera and thus in particular also the entire area at an entrance or around the entrance of a detection corridor, as described further below. This also means in particular the area outside the detection corridor and the area between the entrance to the detection corridor and the combined detection range of the first iris camera and the second iris camera. However, the scene camera can of course also cover the combined detection range of the first iris camera and the second iris camera. When an identification process is initiated (for example, when a person enters the surveillance corridor or when such an identification process is started manually), the scene camera takes images of the entering person and transmits these images to the control system.
[0017] The controller may be any suitable control device, such as a general-purpose computer with appropriate memory and processor components, an ASIC, an FPGA, or any other suitable control device or combination of such control devices.
[0018] The controller extracts the height of the periocular facial area of the person to be identified walking through the surveillance corridor from the images of the scene camera. In particular, the scene camera can have a focus area in the vicinity of an entrance to the surveillance corridor so that sharp images of the person to be identified can be recorded at this position. However, the scene camera can in principle also record images of the person to be monitored at other locations along the surveillance corridor. The image sharpness of the scene camera is not an overly limiting factor, since the scene camera only performs the task of enabling the controller to determine the height of the periocular facial area, i.e., to enable a rough alignment of the system (in particular of the first iris camera and the second iris camera).The images from the scene camera therefore only need to be of good enough quality to enable such a height determination. As is easily apparent, however, this does not require extremely sharp images.
[0019] The scene camera sends the captured images to the control system, which uses appropriate image evaluation algorithms to determine the periocular facial area of the entering person, i.e. its height, based on the recorded image material.
[0020] The first iris camera and the second iris camera, in turn, are designed to record the actual images used / necessary for iris recognition. For this purpose, the first iris camera and the second iris camera each have a corresponding field of view and a corresponding focal plane. Generally, the field of view of a camera describes the angular range captured by the corresponding camera. The focal plane, in turn, describes the distance in front of the respective camera at which the sharpest images are recorded. The focal plane is generally determined by the focal length of the respective camera.
[0021] The first iris camera and the second iris camera are each cameras with a fixed focal plane and preferably a small field of view, so that images can be captured without significant distortion. With a large field of view, a high pixel count of an image sensor of the camera is also necessary to enable the necessary resolution for extracting the corresponding features (e.g., iris features). Accordingly, iris cameras with a small field of view enable the use of image sensors with a lower pixel count. The first iris camera and the second iris camera are preferably cameras that have a high frame rate, in particular a frame rate of at least 10 frames per second (fps).
[0022] In addition, the first iris camera and the second iris camera are height-adjustable, allowing the recording height of the first iris camera and the second iris camera to be adjusted. This height adjustability can be achieved in any conceivable way. For example, the first iris camera and the second iris camera can be mounted on separate height-adjustable mounts or on a common height-adjustable mount, as described below, which can be controlled by the controller. However, it is also conceivable for the iris cameras themselves to be mounted in a fixed position, and only their detection range can be adjusted (for example, by a mechanism or an optical system).After determining the height of the periocular facial area of the person to be identified using the scene camera, the controller adjusts the height of the first iris camera and the second iris camera to the height of the periocular facial area of the person to be identified, previously determined using the images from the scene camera. This height adjustability enables the use of cameras with a small field of view, in particular, since the camera is moved or adjusted to the eye level (i.e., the height of the periocular facial area) of the person to be identified, thus capturing the periocular facial area even with a small field of view.
[0023] By using two iris cameras, each with a fixed focal plane, different distances in front of the camera arrangement (i.e., in front of the first iris camera and the second iris camera) can be recorded sharply, or at least sharp enough to extract the iris features. This enables iris recognition of the person to be identified using simple cameras with fixed focal planes, i.e., with fixed focal lengths or working distances in front of the camera, while the person to be identified walks along the detection corridor. In particular, this arrangement avoids the need for complicated camera systems such as plenoptic cameras or other refocusable cameras. This also allows the first iris camera and the second iris camera to be operated at a high frame rate, significantly increasing the probability of successful identification.Furthermore, by eliminating unnecessary complexity, a more robust overall system with greater reliability is achieved.
[0024] In summary, the disclosed device uses a scene camera with a wide field of view and two or more iris cameras, preferably each with a smaller field of view and with fixed focal planes. The scene camera performs a rough capture of the periocular facial area, or its height, of a person to be identified. The heights of the iris cameras (i.e., their capture heights) are then adjusted to the height of the periocular facial area, and images of the periocular facial area are captured while the person walks through a corresponding capture corridor. Image data, or characteristics of the iris of the user to be identified, can then be extracted from this data and used to identify the person. The disclosed device enables the use of simple cameras with fixed focal planes as iris cameras.The increased frame rate and the resulting sharper images increase the probability of successful identification. The use of multiple cameras enables identification while the person to be identified moves along the detection corridor.
[0025] The disclosed device enables, for example, identity verification for the purpose of access control or any other identity verification independent of this. For example, in access control, a door or gate can be automatically opened when an authorized person has been identified. However, the control system can also have a display that outputs a verified identity, for example, to a security officer. Examples of applications include identity verification in airports, train stations, courthouses, corporate buildings, etc. However, this list is only exemplary.
[0026] It should be noted that, although described herein with two iris cameras, the disclosed device may also have more than two iris cameras, for example, three, four, five, or any number of iris cameras.
[0027] According to one embodiment, the device for iris recognition during passing further comprises a height-adjustable mount that supports the first iris camera and the second iris camera and is connected to the controller. The controller is configured to adjust the height of the first iris camera and the second iris camera by appropriately controlling the height-adjustable mount.
[0028] The height-adjustable mount can be, for example, a linear actuator or other motor or similar device to which the first iris camera and the second iris camera are attached. For example, such a linear actuator can be represented by a motorized, height-adjustable mount on a corresponding column to which the iris cameras are attached. However, this is only an example. Any height-adjustable mount that allows a change in the height of the iris cameras can be provided.
[0029] Instead of attaching both iris cameras to a common height-adjustable mount, each iris camera can also be attached to its own height-adjustable mount. If more than two iris cameras are planned, all iris cameras can be attached to a common height-adjustable mount, or each iris camera can be attached to its own height-adjustable mount. In this case, only some of the iris cameras can be attached to a common mount.
[0030] According to another embodiment, the device for iris recognition during walk-by further comprises a strobe light operating in the near-infrared range. The controller is configured to control the strobe light during the detection of the periocular facial area with the first iris camera and the second iris camera such that the strobe light emits light pulses during the detection of the periocular facial area.
[0031] The light pulses facilitate the recognition of iris features when they are directed at the eyes of the person being identified at the same time as the iris cameras are capturing images. For safety reasons, a person's eyes may only be exposed to a certain amount of light energy per unit of time. The use of a strobe light allows the light intensity to be increased without exceeding the permissible average light energy applied to the eyes over time.
[0032] It is also conceivable that a continuous distance measurement takes place during operation of the strobe light (for example, using a laser distance measurement, infrared distance measurement, or any other suitable distance measurement method), and the light intensity is adjusted based on the distance so as not to exceed a maximum permissible light intensity in the eyes. The light intensity would then decrease as the distance decreases.
[0033] According to a further embodiment, the light pulses are synchronized with the image capture by the first iris camera and the second iris camera.
[0034] By synchronizing the light pulses with the image recording, it is ensured that the iris of the person to be identified is illuminated by the iris cameras during each image recording, allowing the iris features to be better recognized and extracted.
[0035] According to a further embodiment, the third field of view is larger than the first field of view and than the second field of view.
[0036] Since the scene camera (which is assigned the third field of view) is used to detect the periocular facial area and align the iris cameras to this level, it must capture a larger angular range. In principle, however, a smaller field of view leads to less image distortion, for example, due to chromatic aberrations, excessive barrel distortion, corner blurring, and vignetting. However, since the scene camera only serves to roughly align to the height of the periocular facial area, such distortions are less significant here. However, to identify iris features, a realistic recording of the periocular facial area and thus of the iris of the person to be identified is desirable. The iris cameras therefore have a smaller field of view, which enables a more detailed recording of the iris and the use of an image sensor with a lower number of pixels.
[0037] According to another embodiment, the controller is configured to capture the periocular facial area with the first iris camera and the second iris camera while the person to be identified moves through a capture corridor. According to another embodiment, the first iris camera has a first depth of field, and the second iris camera has a second depth of field.
[0038] Depth of field is a measure of the extent of the sharp area in the object space of an imaging optical system, such as a camera. Depth of field thus describes the size of the distance range of a camera within which an object is sufficiently sharply imaged. Since the sharpest image is present in the focal area or focal plane, a large depth of field means a large area around the focal plane that is sharply imaged.
[0039] The first iris camera and the second iris camera can both have the same depth of field or different depths of field. However, the resulting depth of field also depends on the position of the focal plane. Thus, the depth of field (described below as the capture volume) of the first iris camera and the second iris camera can differ, even if both iris cameras have the same depth of field. This enables the first and second iris cameras (and potentially additional iris cameras) to capture different areas along the capture corridor in focus, so that together, sharp images of the periocular facial area can be captured over a larger area than with an iris camera alone.
[0040] According to a further embodiment, the first focal plane and the first depth of field define a first capture volume, and the second focal plane and the second depth of field define a second capture volume. The first capture volume and the second capture volume each represent spatial volumes in front of the corresponding one of the first iris camera and the second iris camera, which are each sharply imaged by the corresponding one of the first iris camera and the second iris camera.
[0041] According to a further embodiment, the first detection volume and the second detection volume differ from each other at least in sections. In particular, the first detection volume and the second detection volume can be located one behind the other in the direction of the detection corridor, so that each of the first iris camera and the second iris camera can sharply image different areas along the detection corridor. This makes it possible to capture sharp images of the periocular facial area over a larger area than with an iris camera alone. The acquired data set and thus the probability of successful identification are thus increased.
[0042] Since the first iris camera and the second iris camera are aligned at the same height (the height of the periocular facial area) during an identification process, the capture volumes have essentially the same vertical extent, unless the first field of view of the first iris camera and the second field of view of the second iris camera differ from each other. However, it is also conceivable for the first capture volume and the second capture volume to differ in height to a limited extent, as long as it is ensured that the periocular facial area can be captured by both iris cameras.
[0043] The fact that the first detection volume and the second detection volume differ from each other at least in sections means, in particular, that an overlapping area can also exist between the detection volumes (in particular along the detection corridor, i.e., along a horizontal distance from the iris cameras), which is sharply imaged by both the first iris camera and the second iris camera. This enables continuous, sharp image recording along the first detection volume and along the second detection volume.
[0044] According to a further embodiment, the first detection volume and the second detection volume are located one behind the other in the detection corridor through which the person to be identified moves during iris recognition.
[0045] The detection corridor runs at least in the area of the first and second
[0046] detection volume towards the first iris camera and the second iris camera, so that the person to be identified moves at least partially towards the first and the second iris camera in the detection corridor.
[0047] According to a further embodiment, the first iris camera, the second iris camera and the scene camera are each non-plenoptic cameras.
[0048] As already indicated in the background of the invention, plenoptic cameras (also referred to as light-field cameras) are powerful cameras capable of achieving greater depth of field. However, since such camera systems reduce the frame rate due to the large amount of data and thus produce fewer sharp images, the disclosed device dispenses with such cameras and instead uses several simple cameras with fixed focal planes, which are preferably different from one another. This enables a high frame rate with a simultaneous large, sharp detection area.
[0049] According to a further embodiment, the device for iris recognition when passing by further comprises a recognition device which is connected to the controller and is configured to detect the entry of the person to be identified into the detection corridor and to initiate an identification process in response thereto.
[0050] An identification process can thus be triggered automatically by the control system when the recognition device detects that a person to be identified enters the detection corridor.
[0051] However, it should be noted that an identification process can also be started manually, for example by operating an appropriate input device.
[0052] According to a further embodiment, the recognition device comprises at least one of a light barrier, an NFC device, an infrared distance sensor, and a motion detector, which are mounted at an entrance to the detection corridor. With an infrared distance sensor, the identification process can be started automatically, for example, depending on the distance of the person to be identified.
[0053] For example, a light barrier or a motion detector can detect when a person enters or approaches the detection corridor, whereupon the control system can automatically start an identification process.
[0054] For example, an NFC device can detect the approach of a transponder (such as an employee transponder worn by an employee attempting to enter a specific area) and then initiate an identification process. Particularly when identification is used to verify access authorization, such a transponder, combined with an NFC device, can implement an additional layer of security, as the identification process is only initiated when a person in possession of such a transponder approaches.
[0055] For example, the identification can then start and if the identification was successful, a gate or door can be opened.
[0056] However, this list is only exemplary and other detection devices are also conceivable.
[0057] According to a further embodiment, the device for iris recognition upon passing further comprises a data storage device with a database that stores biometric features of the periocular facial area of authorized persons. The controller is configured to determine the identity of the person to be identified by comparing it with the database.
[0058] The data storage can, for example, be part of the control system or can be independent of the control system but in communication with it. It is also conceivable for the data storage to be located in a cloud, so that a central database can be accessed from different locations. The database can, for example, contain data on a person's iris characteristics or other biometric data, particularly from the periocular facial area. The database is preferably encrypted to prevent or at least impede unauthorized access.
[0059] Short description of the characters
[0060] Fig. 1 shows a device for iris recognition during passing with a user to be identified who is about to enter a detection corridor, wherein the set height of the first and second iris cameras is too high to detect the periocular facial area of the user.
[0061] Fig. 2 shows the device from Fig. 1 shortly after the person to be identified has entered the detection corridor and after the height of the iris cameras has been adjusted.
[0062] Fig. 3 shows the device from Figs. 1 and 2 after the height of the iris cameras has been adjusted to the height of the periocular facial area of the person to be identified and while the second iris camera takes an image of the periocular facial area.
[0063] Fig. 4 shows an exemplary image of a periocular facial area as recorded by the first and second iris cameras.
[0064] Detailed description of exemplary embodiments
[0065] The representations in the figures are schematic and not to scale. Where the same reference numerals are used in different figures in the following description, they denote identical or similar elements. Identical or similar elements may, however, also be denoted by different reference numerals. Fig. 1 shows an exemplary device 100 for iris recognition as people walk by. The device 100 serves to identify a person 80 to be identified by detecting a periocular facial area 81 (see Fig. 4) of the person 80 to be identified.
[0066] The device 100 comprises a scene camera 30, a first iris camera 10, a second iris camera 20, a strobe light 60, a controller 50 with a memory 51, and a detection device 90, which can be configured, for example, as a light barrier 91 or an NFC device 92 (or other suitable devices, such as motion detectors, infrared distance sensors, etc.). The first iris camera 10, the second iris camera 20, and the strobe light 60 are also mounted on a common height-adjustable mount 40. Optionally, the scene camera can also be mounted on the height-adjustable mount 40. The first iris camera 10, the second iris camera 20, the scene camera 30, the strobe light 60, the detection device 90, and the height-adjustable mount 40 are each in electronic communication with the controller 50.The controller 50 receives and / or sends signals to each of these devices 10, 20, 30, 40, 60, 90 and thus controls the operation of the device 100.
[0067] The first iris camera 10 is a camera with a first field of view 11 and a fixed first focal plane 12 and the second iris camera 20 is a camera with a second field of view 21 and a fixed second focal plane 22. The first iris camera 10 has a first depth of field 13 and the second iris camera 20 has a second depth of field 23.
[0068] A "fixed" focal plane generally describes a focal plane that cannot be moved. Thus, the first iris camera 10 and the second iris camera 20 are not refocusable cameras, and in particular, they are not plenoptic cameras (light field cameras) or cameras with liquid lenses, but rather cameras with a fixed focal plane. The field of view of a camera generally defines the angular range that can be captured by the respective camera. Accordingly, the first field of view 11 defines an angular range that the first iris camera 10 captures, and the second field of view 21 defines an angular range that the second iris camera 20 captures.
[0069] The focal planes 12, 22 of the first iris camera 10 and the second iris camera 20 define a distance in front of the respective iris camera 10, 20 that produces the sharpest image. Generally, the focal plane of a camera corresponds to its focal length.
[0070] The depths of field 13, 23 of the iris cameras 10, 20, in turn, describe (as described above generally with regard to cameras) the size of a distance range in front of the respective camera within which an object is imaged sufficiently sharply. In general, the areas defined by the depths of field 13, 23 each extend from a plane between the corresponding focal plane 12, 22 of the corresponding iris camera 10, 20 and the corresponding iris camera 10, 20 to a plane that is located further out (i.e., further away from the camera) than the corresponding focal plane 12, 22 of the corresponding iris camera 10, 20. The depths of field thus define an area around the focal planes 12, 22 of the iris cameras 10, 20 that is imaged sufficiently sharply to extract the necessary image information, here for iris recognition or recognition of other biometric features of the periocular facial area 81.
[0071] The first focal plane 12 and the first depth of field 13 thus define (naturally together with the first field of view 11) a first detection volume 14 of the first iris camera 10. The second focal plane 22 and the second depth of field 23 in turn define (together with the second field of view 21) a second detection volume 24 of the second iris camera 20. The first detection volume 14 and the second detection volume 24 thus each define a spatial volume in front of the corresponding iris camera 10, 20, in which a sufficiently sharp image recording for iris recognition is possible. The first detection volume 14 and the second detection volume 24 are located one behind the other in Fig. 1 (and also Figs. 2 and 3). This enables a sharp image recording along a larger area along a detection corridor 70 with iris cameras 10, 20 with fixed focal planes 12, 22 (i.e. focal lengths).No complex refocusable camera systems, such as plenoptic cameras, which produce high data rates, are used. Since no such refocusable camera systems are used, the frame rate of the iris cameras 10, 20 can be higher, enabling accurate iris recognition as the person walks through the detection corridor 70 with simple iris cameras 10, 20 over a large area along the detection corridor 70.
[0072] The first detection volume 14 and the second detection volume 24 are arranged one behind the other in Figs. 1 to 3 and are spaced apart. However, it should be noted that the detection volumes 14, 24 can also partially overlap to enable continuous, sharp image capture through the two detection volumes 14, 24. Furthermore, it should be noted that the arrangement with two iris cameras 10, 20 is merely exemplary, and more than two iris cameras 10, 20 can also be present to further enlarge the total detection volume and thereby increase the probability of successful identification.
[0073] The scene camera 30 has a third field of view 31, which is larger than the first field of view 11 of the first iris camera 10 and than the second field of view 21 of the second iris camera 20. The scene camera 30 thus captures a larger angular range in front of the camera than the iris cameras 10, 20 and is used in particular for the rough alignment of the device 100 to the periocular facial area 81 during an identification process, as described further below.
[0074] A smaller field of view (i.e., a smaller captured angular range in front of the camera) is generally associated with less image distortion (as described above), but results in a smaller image section. In addition, a small field of view has the advantage that image sensors with a lower pixel density can be used while still achieving the required pixel resolution. Since the first iris camera 10 and the second iris camera 20 serve to capture the periocular facial area 81, and the images of the periocular facial area 81 are in turn used for iris recognition and thus for identification, the smallest possible image distortion and high resolution are desirable in order to be able to extract the information required for identification from the image data. The first iris camera 10 and the second iris camera 20 therefore have smaller fields of view 11, 21 than the scene camera 30.
[0075] After an identification process has been started, the scene camera 30 takes a (preferably two-dimensional) “scene image” which shows the person to be identified
[0076] 80 and sends corresponding image data to the controller 50. The controller 50 can then, with appropriate image evaluation algorithms and using the known geometry of the camera system, determine a height 82 of the periocular facial area
[0077] 81 of the person 80. For this purpose, the controller 50 can, for example, recognize structures in the images typical for a periocular facial area 81. For example, the memory 51 of the controller 50 can contain corresponding comparison data, and the controller 50 can also use machine learning algorithms. However, all other conceivable and suitable image evaluation algorithms are also possible. The height determination can, for example, also be based on a calibration or a mathematical model stored in the data memory 51 and allowing a specific position on the scene image to be assigned to a specific height.
[0078] Fig. 1 shows the device 100 together with a person 80 to be identified, before an identification process of the person 80 has been initiated. The height of the first iris camera 10 and the second iris camera 20 (ie, in particular, the height of the detection volumes 14, 24) is too high to detect the periocular facial area 81 of the person 80.
[0079] An identification process can be carried out manually, for example, by pressing a corresponding switch or button, or, as shown in Figs. 1 to 3, automatically by a recognition device 90. If the recognition device 90 is, for example, a light barrier 91 or an NFC device 92, the identification process can start, for example, when the person 80 passes the light barrier 91 or, for example, carries a transponder that can be recognized by an NFC device 92 and when the person 80 walks past the NFC device with the transponder.
[0080] Since the iris cameras 10, 20 have a relatively small field of view 11 and 21, respectively, the detection heights of the iris cameras 10, 20 must be aligned to the height 82 of the periocular facial area 81 in order to be able to capture images of the periocular facial area 81. For this purpose, the scene camera 30 first records one or more images after the person 80 enters the detection corridor 70 or even before the person 80 enters the detection corridor 70, and the controller 50 determines the height 82 of the periocular facial area 81 based on this image data, as described above. The controller 50 then adjusts the height of the first iris camera 10 and the second iris camera 20 to the height 82 of the periocular facial area 81, so that the iris cameras 10, 20 can capture images of the periocular facial area 81.For example, the first iris camera 10 and the second iris camera 20 can be mounted on a height-adjustable holder 40 (not explicitly shown), which can be, for example, a corresponding linear actuator or another device that moves the iris cameras 10, 20 in height.
[0081] In Figs. 1 to 3, an optional stroboscopic light 60 is also provided, which is configured to illuminate the periocular facial area 81 during image acquisition by the iris cameras 10, 20 and to stimulate the pupil reaction so that the iris features are clearly visible and extractable in the images. In Figs. 1 to 3, the stroboscopic light 60 is also attached to the height-adjustable mount 40 and can be adjusted in height together with the iris cameras 10, 20. However, this is not absolutely necessary. The stroboscopic light 60 can, for example, also be fixed in position and illuminate the entire detection corridor 70. The scene camera 30 can be mounted at a fixed height, or can also be height-adjustable and, for example, move with the iris cameras 10, 20 when their height is adjusted.In such configurations, the controller 50 takes the respective height of the scene camera 30 into account when determining the height. The scene camera 30 does not necessarily have to be mounted in the immediate vicinity of the iris cameras 10, 20 (as shown in Figs. 1 to 3), but can also be provided at other positions. All that is important is that the scene camera 30 can capture the detection corridor 70 and that the viewing angle of the scene camera 30 enables a determination of the height of the periocular area 81.
[0082] It should also be noted that the detection corridor 70 does not necessarily have to be a physically separated area, as shown in Figs. 1 to 3, but generally describes the detection area of the device 100. Physical barriers are not necessarily required for this purpose. However, during identification, the person 80 must face the iris cameras 10, 20 at least in the area of the detection volumes 14, 24 so that the iris cameras 10, 20 can capture images of the periocular facial area 81.
[0083] In Figs. 2 and 3, some elements and reference numerals, such as the controller 50 with the memory 52 and the height-adjustable holder 40 from Fig. 1, have been omitted for the sake of clarity.
[0084] Fig. 2 shows the device 100 from Fig. 1 after the person 80 to be identified has entered the detection corridor 70, the identification process has been started, and the first iris camera 10 and the second iris camera 20 have been adjusted to the height 82 of the periocular visual area 81, as described with reference to Fig. 1. In Fig. 2, the person 80 to be identified is still located in front of the first detection volume 14 and the second detection volume 24, but is moving towards the detection volumes 14, 24. In this section, optionally no further action of the device 100 initially takes place (after completion of the height adjustment of the iris cameras 10, 20 and possibly the strobe light 60). Fig. 3 shows the device 100 after the person 80 to be identified has entered the second detection volume 24. At least from this point onwards, the controller 50 instructs the first iris camera 10 and the second iris camera 20 to begin image capture.However, the controller 50 can also continuously capture image data with the iris cameras 10, 20 from the beginning of the second section shown in Fig. 2 (i.e., after the person 80 enters the capture corridor 70). This eliminates the need for further determination of the person's position (e.g., by means of another light barrier or by other means) in order to determine when image capture should start. The controller can then, for example, discard images that do not have a recognizable periocular facial area 81. In addition, the controller 50 instructs the stroboscopic light 60 to illuminate the capture volumes 14, 24 with stroboscopic light pulses. The person 80 thus walks through the first capture volume 14 and the second capture volume 24, while the corresponding iris camera 10, 20 captures images of the periocular facial area 81 and sends them to the controller 50.The strobe light can optionally be synchronized with the image capture by the iris cameras 10, 20, so that a light pulse is emitted whenever an image is being captured by one of the iris cameras 10, 20. This ensures that each image is sufficiently illuminated and that the iris features can be extracted from the images of the periocular facial area 81.
[0085] The controller 50 then extracts the iris features from the images captured by the iris cameras 10, 20 and compares them with corresponding comparison data in a database in order to determine the identity of the person 80 to be identified. The database can be stored, for example, in the data storage 51. However, the comparison data can also be stored, for example, in a centralized cloud storage, so that use by devices 100 at different locations is possible. Regardless of the storage type, the comparison data is preferably stored in encrypted form in order to prevent misuse through unauthorized reading of the comparison data. The device 100 can also be integrated into a more comprehensive system. If the identity is determined, for example, for access control, the device 100 (orThe device (their control 50) communicates with a door mechanism and, for example, automatically opens a door, gate, or barrier if the detected identity corresponds to that of an authorized person. However, any other application for identity verification is also conceivable.
[0086] Fig. 4 shows a typical image of a periocular facial area 81. In such a recording, the eyes and in particular the iris are clearly visible, so that these biometric features can be used to identify the person 80, as described above.
[0087] In general, it should be noted that the identification can not only take place on the basis of the iris features, but can also take into account or at least take into account other biometric facial features of the periocular facial area 81, such as corresponding structures of the eye area.
[0088] List of reference symbols
[0089] 10 first iris camera
[0090] 11 first field of vision
[0091] 12 first focal plane
[0092] 13 first depth of field
[0093] 14 first recording volume
[0094] 20 second iris camera
[0095] 21 second field of view
[0096] 22 second focal plane
[0097] 23 second depth of field
[0098] 24 second recording volume
[0099] 30 scene camera
[0100] 31 third field of vision
[0101] 40 height-adjustable bracket
[0102] 50 Control
[0103] 51 data storage
[0104] 60 strobe lights
[0105] 70 detection corridor
[0106] 80 person to be identified
[0107] 81 periocular facial area
[0108] 82 Height of the periocular facial area
[0109] 90 Detection device
[0110] 91 light barrier
[0111] 92 NFC device
[0112] 100 Iris recognition device
Claims
Patent claims 1. A device (100) for iris recognition as people walk by, the device (100) comprising: a first iris camera (10) with a first field of view (11) and a fixed first focal plane (12); at least one second iris camera (20) with a second field of view (21) and a fixed second focal plane (22); a scene camera (30) with a third field of view (31); and a controller (50); wherein the first iris camera (10) and the second iris camera (20) are height-adjustable; wherein the first iris camera (10), the second iris camera (20), and the scene camera (30) are connected to the controller (50); wherein the controller (50) is configured to capture, via the scene camera (30), images of a person (80) to be identified who is located outside a combined detection range of the first iris camera (10) and the second iris camera (20);wherein the controller (50) is configured to determine a height (82) of a periocular facial area (81) of the person (80) to be identified based on the captured images of the scene camera (30); wherein the controller (50) is configured to adjust a height of the first iris camera (10) and the second iris camera (20) in response to the determined height (82) of the periocular facial area (81) such that the first iris camera (10) and the second iris camera (20) can capture the periocular facial area (81); and wherein the controller (50) is configured to capture the periocular facial area (81) with the first iris camera (10) and the second iris camera (20) and to determine an identity of the person (80) to be identified based on the captured periocular facial area (81).
2. The walk-by iris recognition device (100) according to claim 1, further comprising a height-adjustable mount (40) supporting the first iris camera (10) and the second iris camera (20) and communicating with the controller (50); wherein the controller (50) is configured to adjust the height of the first iris camera (10) and the second iris camera (20) by controlling the height-adjustable mount (40).
3. The walk-by iris recognition device (100) according to any one of the preceding claims, further comprising a strobe light (60) operating in a near-infrared range; wherein the controller (50) is configured to control the strobe light (60) during the detection of the periocular facial area (81) with the first iris camera (10) and the second iris camera (20) such that the strobe light (60) emits light pulses during the detection of the periocular facial area (81).
4. The walk-by iris recognition device (100) of claim 3, wherein the light pulses are synchronized with the image capture by the first iris camera (10) and the second iris camera (20).
5. A walk-by iris recognition device (100) according to any one of the preceding claims, wherein the third field of view (31) is larger than the first field of view (11) and the second field of view (21).
6. The walk-by iris recognition device (100) according to any one of the preceding claims, wherein the controller (50) is configured to capture the periocular facial area (81) with the first iris camera (10) and the second iris camera (20) as the person to be identified (80) moves through a capture corridor (70).
7. The walk-by iris recognition device (100) according to any one of the preceding claims, wherein the first iris camera (10) has a first depth of field (13); and wherein the second iris camera (20) has a second depth of field (23).
8. The walk-by iris recognition device (100) of claim 7, wherein the first focal plane (12) and the first depth of field (13) define a first detection volume (14); wherein the second focal plane (22) and the second depth of field (23) define a second detection volume (24); wherein the first detection volume (14) and the second detection volume (24) together form the combined detection area of the first iris camera (10) and the second iris camera (20) and each represent volumes of space in front of the corresponding one of the first iris camera (10) and the second iris camera (20), which volumes are each sharply imaged by the corresponding one of the first iris camera (10) and the second iris camera (20).
9. The device (100) for iris recognition during passing according to claim 8, wherein the first detection volume (14) and the second detection volume (24) are at least partially different from each other.
10. Device (100) for iris recognition during passing according to claim 8 or 9, wherein the first detection volume (14) and the second detection volume (15) are located one behind the other in the detection corridor (70) through which the person to be identified (80) moves during iris recognition.
11. A walk-by iris recognition device (100) according to any one of the preceding claims, wherein the first iris camera (10), the second iris camera (20) and the scene camera (30) are each non-plenoptic cameras.
12. The walk-by iris recognition device (100) of any preceding claim, further comprising a recognition device (90) in communication with the controller (50) and configured to detect the entry of the person (80) to be identified into the detection corridor (70) and to initiate an identification process in response thereto.
13. The iris recognition device (100) according to claim 12, wherein the recognition device (90) comprises at least one of a light barrier (91), an NFC device (92), an infrared distance sensor, and a motion detector mounted at an entrance of the detection corridor (70).
14. Device (100) for iris recognition when passing by according to one of the preceding claims, further comprising a data memory (51) with a database which stores biometric features of the periocular facial area (81) of persons authorized to access, wherein the controller (50) is configured to determine the identity of the person to be identified (80) by comparing it with the database.