WO2007103833A1

WO2007103833A1 - A combined face and iris recognition system

Info

Publication number: WO2007103833A1
Application number: PCT/US2007/063242
Authority: WO
Inventors: Gary E. Determan; Vincent C. Jacobson; Jan Jelinek; Thomas Phinney; Rida M. Hamza; Terry Ahrens; George A. Kilgore; Rand P. Whillock; Saad Bedros
Original assignee: Honeywell International, Inc.
Priority date: 2006-03-03
Filing date: 2007-03-02
Publication date: 2007-09-13
Also published as: GB0815735D0; AU2007223336B2; GB2450022B; GB2450022A; IL193751A; AU2007223336A1; IL193751A0

Abstract

A system using face and iris image capture for recognition of people. The system may have wide field-of-view, medium field-of-view and narrow field-of-view cameras to capture images of a scene of people, faces and irises for processing and recognition. Matching of the face and iris images with images of a database may be a basis for recognition and identification of a subject person.

Description

A COMBINED FACE AND IRIS RECOGNITION SYSTEM

This application claims the benefit of U.S. Provisional Application No. 60 /778 ,770 , filed March 3, 2006.

This application claims the benefit of U.S. Provisional Application No. 60/807,046, filed July 11, 2006.

This application is a continuation-in-part of U.S. Patent Application No. 11/275,703, filed January 25, 2006, which claims the benefit of U.S. Provisional Application No. 60/647,270, filed January 26, 2005.

This application is a continuation-in-part of U.S. Patent Application No. 11/043,366, filed January 26, 2005.

This application is a continuation-in-part of U.S. Patent Application No. 11/372,854, filed March 10, 2006.

This application is a continuation-in-part of U.S. Patent Application No. 10/979,129, filed November 3, 2004, which is a continuation-in-part of U.S. Patent Application 10/655,124, filed September 5, 2003.

This application is a continuation-in-part of U.S. Patent Application 11/382,373, filed May 9, 2006.

This application is a continuation-in-part of U.S. Patent Application No. 11/672,108, filed February 7, 2007.

This application is a continuation-in-part of U.S. Patent Application No. 11/675,424, filed February 15, 2007. This application is a continuation-in-part of U.S.

Patent Application No. 11/681,614, filed March 2, 2007.

This application is a continuation-in-part of U.S.

Patent Application No. 11/681,662, filed March 2, 2007.

This application is a continuation-in-part of U.S. Patent Application No. 11/681,751, filed March 2, 2007,

Attorney Docket. This application is a continuation-in-part of U.S. Patent Application No. 11/681,470, filed March 2, 2007.

This application is a continuation-in-part of U.S. Patent Application No. 11/681,505, filed March 2, 2007. This application is a continuation-in-part of U.S. Patent Application No. 11/681,251, filed March 2, 2007, which claims the benefit of U.S. Provisional Application 60/807,046, filed July 11, 2006.

The government may have rights in the present invention.

Background

The present invention pertains to recognition systems and particularly to biometric recognition systems. More particularly, the invention pertains to combination face and iris recognition systems.

U.S. Provisional Application No. 60/778,770, filed March 3, 2006, is hereby incorporated by reference.

U.S. Provisional Application No. 60/807,046, filed July 11, 2006, is hereby incorporated by reference. U.S. Patent Application No. 11/275,703, filed January 25, 2006, is hereby incorporated by reference.

U.S. Provisional Application No. 60/647,270, filed January 26, 2005, is hereby incorporated by reference. U.S. Patent Application No. 10/979,129, filed

November 3, 2004, is hereby incorporated by reference.

U.S. Patent Application 10/655,124, filed September 5, 2003, is hereby incorporated by reference.

U.S. Patent Application 11/382,373, filed May 9, 2006, is hereby incorporated by reference.

U.S. Patent Application No. 11/043,366, filed January 26, 2005, is hereby incorporated by reference.

U.S. Patent Application No. 11/372,854, filed March 10, 2006, is hereby incorporated by reference. U.S. Patent Application No. 11/672,108, filed

February 7, 2007, is hereby incorporated by reference. U.S. Patent Application No. 11/675,424, filed February 15, 2007, is hereby incorporated by reference.

U.S. Patent Application No. 11/681,614, filed March 2, 2007, is hereby incorporated by reference. U.S. Patent Application No. 11/681,662, filed March 2, 2007, is hereby incorporated by reference.

U.S. Patent Application No. 11/681,751, filed March

2, 2007, Attorney Docket No. 1100.1449101, is hereby incorporated by reference. U.S. Patent Application No. 11/681,470, filed March

3, 2007, Attorney Docket No. 1100.1450101, is hereby incorporated by reference.

U.S. Patent Application No. 11/681,505, filed March 3, 2007, is hereby incorporated by reference. U.S. Patent Application No. 11/681,251, filed March 3, 2007, is hereby incorporated by reference.

Summary

The present invention is a combined face and iris recognition system.

Brief Description of the Drawing

Figure 1 is a diagram of a combined face and iris recognition system; Figure 2 is a diagram of an approach for the face and iris recognition system;

Figure 3 is a block diagram of the recognition system;

Figure 4 is an illustration of some of the hardware that might be used for the recognition system; and

Figures 5a, 5b and 5c illustrate the enclosure and display for the recognition system.

Description

The present system may relate to biometrics, face and iris recognition systems, image metrics, authentication, access control, monitoring, identification, and security and surveillance systems.

Some potential applications for the present system may include airport access control to secure areas, airport trusted traveler systems, border control--watch list surveillance, industrial access control, outdoor facility access control, military checkpoint surveillance, critical infrastructure .... Due to world wide increased security concerns there is a need to have the ability to accurately identify individuals at a distance. In some applications, it is essential to collect multimodal biometrics of multiple subjects standing or walking in an area. Among the various biometrics available nowadays for recognizing people, iris-based biometrics is the most accurate method for identifying people. There may be a present multimodal biometrics system relying on face and iris for guarantying high recognition rate in most challenging cases. The use of face and iris may improve the overall accuracy of the system. The human iris may provide a robust biometrics signature; however, collecting iris information from a distance appears challenging for several reasons. People may be in arbitrary places in the environment; people might not be looking straight into the device; people's head poses might be arbitrary; people's eyelids might occlude part of the iris signature and therefore provide a partial biometrics; and finally, people might be freely moving in the environment monitored by the present apparatus. Stand-off iris acquisition may also address challenging problems related to imaging the irises with sufficient resolution required for inferring a robust signature. Furthermore, a robust iris signature may be present at just a near infrared wavelength, requiring the use of a sensor capable of measuring such wavelengths, and the use of a near infrared illuminator to guarantee good imaging conditions .

Iris recognition may have high accuracy in identifying humans. Iris recognition systems may also have suitability, as an exceptionally accurate biometric can be derived from its extremely data-rich physical structure, stability over time, and non-contact acquisition. Related art iris recognition systems may require users to be within a few inches of the sensor and look directly into the sensor, or towards a specific area. The present system does not necessarily have such a requirement.

The present invention and system may use commercial off the shelf (COTS) face recognition technology combined with custom iris processing algorithms to accurately recognize subjects based on the face and iris at distances significantly greater than a few inches. The present combined face and iris recognition system (CFAIRS) may perform automatic illumination, detection, acquisition and recognition of faces and irises at ranges out to five meters (over 16 feet) . The system may also automatically recognize multiple subjects standing in a scene, and optionally enroll any subjects not previously seen in a database. Relative to the operation of the system, there may be a stationary wide field of view COTS stereo camera 11 for initial subject detection (Figure 1) . It may then use a pan-tilt-zoom (PTZ) COTS medium field of view camera 17 to zoom in and acquire the face of each subject within the field of view of camera 11. The acquired faces may be sent to a COTS face processing software package for recognition, such as with processors 18 and 29. In parallel, a modified PTZ COTS narrow field of view camera 22 may acquire images of the irises for each of the subjects. The system may use modified COTS flash flood illuminators 26 with a controller 27 to control the amount of illumination on a subject.

The acquired iris images may be processed for inferring a signature allowing recognition of a person. Various processing might be required for inferring such signature from an image such as pupil detection, iris segmentation, feature extraction and signature matching software. These processing steps might be optimized for irises acquired at a distance where off-axis gaze and eyelid occlusions are common. The iris processing software may be specifically designed to deal with untrained or unaware users, dealing with issues such as off-axis gaze and partial occlusions of the iris due to half-way closed eyes. Figure 1 is a diagram of the present combined face and iris recognition system 10. The system may have a wide field of view (WFOV) camera 11 for surveillance of a scene having one or more subjects of interest, such as people. Camera 11 might be a wide angle stereo camera for providing distance to the targets or subjects of interest. Camera 11 may be connected to an overall system processing unit 12. A camera module 13 may be connected to unit 12. It may be used for obtaining images of faces of people. A high resolution narrow field of view camera module 14 may be connected to unit 12. Module 14 may be used for obtaining images of a feature of a face such as an iris. An illumination module 15 may be connected to module 14. Module 15 may be used for illuminating items for use with the high resolution module 14 to obtain good images of both irises. Module 15 may be able to illuminate with infrared light. A system input/output interface electronics (I/O electronics) module 16 may be connected to unit 12, module 14 and module 15. Module 13 may have a medium f ie ld- o f -view (MFOV) camera (e.g., security camera) 17 that is connected to a face "process" 18 in unit 12. The term "processor" may be used in lieu of "process" in that a process would include processing. Such processes or processors may be a part of a larger processor, such as a system processor. A pan-tilt-zoom (PTZ) control unit 19 may be connected to the MFOV camera 17. The PTZ unit 19 may be connected to a surveillance process or processor 21 in the overall system processing unit 12. Module 14 may have a high resolution narrow field-of-view (NFOV) camera 22, and a pan-tilt and zoom (PTZ) control unit 23 connected to camera 22 and the surveillance processor 21. A NFOV camera controller 24 may be connected to the high resolution camera 22, the system I/O electronics module 16 and the surveillance processor 21. The camera might be connected to the WFOV stereo camera 11. The camera 22 may also be connected to an iris process or processor 25. Module 15 may have an illuminator module consisting of a number of near infrared illuminators 26 and an illumination controller 27 connected to the illuminators 26. Controller 27 may also be connected to I/O electronics module 16 and the NFOV camera controller 24.

The WFOV camera 11 may be connected to a WFOV process or processor 28 of unit 12. WFOV processor 28 may be connected to surveillance processor 21. The face processor 18 may be connected to the surveillance processor 21 and to a face/iris recognition processor 29. Iris processor 25 may be connected to a surveillance processor 21 and the face/iris recognition processor 29. The face/iris recognition processor 29 may be connected to the surveillance processor 21. The face/iris recognition processor 29 may be connected to a face/iris database 31.

The system 1/0 electronics module 16 may be connected to a system processor 32. The surveillance processor 21 may be connected to the system processor 32. A user interface 33 may be connected to the surveillance processor 21 and the system processor 32.

Figure 2 shows a processing flow in the present system 10 (Figure 1) . In a first step 35, subjects may be detected in the wide field of view image. This step 35 may use an open source or other suitable algorithms for face detection. A wide FOV stereo camera 11 may also compute the range to each of the subjects in a scene. A next step 36 may include prioritization of the subjects in the scene monitored by the present system. Detected subjects might be prioritized for iris and face acquisition. The prioritization scheme might use quality of the imaged biometrics (i.e., face and iris) or the motion of people in the scene. Quality of the imaged biometrics might be measured by the quantity or quality of the face and iris images captured, or confidence measures obtained after matching the subject with the database. After prioritization, the face processing (branch 37) and iris processing (branch 38) may occur in parallel . The face processing branch 37 may use the subject location and range information detected and computed by the wide FOV stereo camera 11 to point the medium field of view camera 17 at a subject. A face image may be acquired at step 44 and then sent to the COTS face recognition software or processor 18. The face recognition software or processor 18 may compute features on the face image and compare them against features in a stored database 31, at step 45, with a recognition processor 29 for identification of the subject. If the face or features do not appear in the database, then they may be enrolled in the database.

The iris branch 38 may use the subject location and range information detected and computed by the wide FOV stereo camera 11 to point the modified near infrared (NIR) illuminators 26 to illuminate the face at step 39. The system 10 may do the iris image acquisition in near IR to highlight lower contrast iris features. A next step 41 uses the subject location and range information detected and computed by the wide FOV stereo camera 11 to point the NFOV high resolution camera 22 at the subject, and zoom at the right scale, which may then acquire the iris image. The high resolution camera 22 may be modified to remove an IR blocking filter, or camera 22 may use a NIR sensor. The system may control the focus and zoom of camera 22, using the camera controller 24. As an example, a two part algorithm may be used to focus the iris camera 22 on the subject's iris. This algorithm may first use the subject range to compute the right zoom factor required for imaging irises with sufficient resolution, auto-focus the camera 22 using visible light, and then it may use the subject range to compute a focus offset to get the image in focus in the NIR wavelength. Other kinds of suitable algorithms may be used.

After iris acquisition, the iris images may be processed with custom iris processing algorithms to extract unique features, at step 42, with an iris processor 25. The extracted features may be matched to features stored in a database 31, at step 43, with the recognition processor 29, to identify the subject. If the extracted features are not matched to features stored in the database, the subject may be enrolled in the system by adding the features and subject information to the database.

Another step 46 in the system 10 process may be a display of the matched results for both the face and iris. Matched results may be a corroboration of the identity of the subject. For specific applications, these results may be passed to a security system (e.g., access control) . The entire process may then be repeated with a return to step 35 for other subjects. A custom architecture may maintain the inter-component communications and control for the system. Figure 3 shows a general layout of the present system. There may be camera or sensor module 61 for obtaining images of one or more subjects in a scene. Images of objects and features of each subject may be obtained by the camera module. These images may be processed by an object/feature processor 62. The processed images may go to a recognition processor 63 which may interact with an object/feature database 64, to perform matching object images and feature images with like images in the database 64 purposes of seeking recognition and identity of the subject. For an example, a subject may be a person, an object may be a face and a feature may be an iris. The subject, object and feature could represent other entities. The camera module 61 may take WFOV images of a scene with one or more subjects. The module 61 may take MFOV images of an object of the one or more subjects. Also, module 61 may take NFOV images of a feature of the object. The module 61 may take pictures of the various fields-of-view with just one adjustable camera with adjustable resolution, or with more than one camera. A general processor 65 may coordinate various operations of the module 61 and processors 62 and 63. The processor 65 may provide user interfacing and also a connection with the world outside of the present system. It may even provide the recognition processor 63 with an external database located at the outside 66 via internet cable, wireless and other media.

The present recognition system 10 may have an enclosure including the following items. The items may include a commercial off the shelf (COTS) wide field of view (WFOV) stereo camera 11, a COTS medium field of view pan-tilt-zoom (PTZ) security or other kind of camera system 13 and a narrow field of view, such as a high resolution, iris camera system 14. The narrow field of view camera system 14 may include a modified COTS high resolution camera 22, motorized COTS zoom lens, a narrow field of view iris camera controller 24, and a customized pan tilt zoom PTZ controller unit 23. One of the cameras 17 and/or 22 may be a standard PTZ camera. The present recognition system may also include an illumination system 15, system I/O electronics 16 and a processing system 12. The illumination system 15 may include a number of modified flash illuminators 26 and an illuminator controller 27. The processing system 12 may include several computers, a system processor, a digital signal processor or a customized processing board. The system may provide a user interface 33 having a monitor, a keyboard with a build-in mouse, and a keyboard, video, and mouse (KVM) switch. It may be noted that the COTS WFOV stereo camera 11 may be used for locating the position of people in the scene. The COTS medium field of view PTZ security camera 17 may be used to capture images of the people's faces. The high resolution iris camera system 14 may be used to capture near infrared iris images of the people in the scene. The iris camera 22 may be connected to the customized PTZ controller unit 23 and can allow the camera 22 to be pointed to people in the scene. The iris camera controller 24 may control a motorized COTS zoom lens which allows the system to zoom in on the irises in the scene. It may also be connected to a modified lens extension along with the extension lens communication hardware that enables the system to override the auto- focus control. The illumination system 15 may have a number of modified flash illuminators 26 and associated illuminator electronics. The illuminator electronics 27 may allow the system to selectively select the flash units used for each picture taken. The system I/O electronics 16 may be used for initializing components by switching on/off power to the devices. It may also be used to monitor the camera to detect when the auto-focus is ready and it may also be used to select illuminators before they are fired.

The processing system 12 which includes computers may be used for the processing and storage of the images as well as for the user interface 33. There also may be a monitor and a keyboard with a build-in mouse which is used for the user interface. The KVM switch may be used to switch between computers for debugging purposes. Some of these components may be modified. The COTS high resolution camera 22 may have a blocking infrared mechanism. A PTZ unit may be customized to fit the camera 22 with mechanical stops installed to limit the travel . The COTS zoom lens may be engineered with gears and a motor to allow electronic control over the zoom and focus settings. The lens extension may be disassembled and the electronics be modified. The flash illuminators 26 may be modified so that they can be remotely triggered and have the "on" time of the flash reduced. A visible light filter may be added to allow the illumination to be limited to mostly infrared.

Some of the operation of the present recognition system 10 may be noted. As a person enters a scene, a WFOV camera 11 may detect and send images to a WFOV processor 12. The WFOV processor may locate the person's head position and range and send coordinates of the head to a surveillance processor. The surveillance processor 21 may prioritize which person in the scene needs to be captured with either the MFOV camera 17 or the iris camera 22. Camera 17 may be a security camera. The surveillance processor may send PTZ coordinates to the camera PTZ 19. The camera PTZ 19 may move to a correct location and send a command to the camera 17, which may then take a picture of the person and send the image to the face/iris recognition processor 29. The face/iris recognition processor may check the database 31 for a face match and may enroll the face if it does not have a match.

The surveillance processor 21 may send the PTZ coordinates to the corresponding PTZ 23 of the narrow field of view (NFOV) camera 22. The PZT 23 may move the camera to the position and zoom coordinates and then send a command to the NFOV or iris camera 22. The iris camera 22 may use its built-in auto focus electronics to focus the image in the visible wavelength. The system may wait for the auto-focus to complete, and then apply a focusing offset, based on the range of the subject, to get the image in focus in the near infrared wavelength. The infrared illuminators may then fire and the iris camera 22 may then take a picture of the person and send the image to the face/iris recognition processor 29 via processor 25. The face/iris recognition processor 29 may check a database 31 for a match of the irises and may enroll the irises if there is no match. The face/iris recognition processor 29 may also keep track of pairing face and irises from the same subject.

A feature of the present recognition system 10 is that it may use either face, iris or both a face and iris in the recognition process. The system may work at large distances (≥ 5 m., 16+ ft . ) . Also, the system may handle a large number of subjects.

Figure 4 shows an example structure 53 containing system 10. A high resolution camera 22, a range estimation camera 11 and a face capture camera 17 are shown. Custom electronics 51 may encompass items of unit 12 and other items of a control and electronic nature. Also shown are flashes 52 which may be a part of illuminator 26.

Figures 5a and 5c show cabinetry for encasing and/or enclosing the system 10. Structure 53 of Figure 4 may be situated on top of cabinet 54 when open for use as shown in Figure 5a. Figure 5c shows the cabinet 54 closed up with structure 53 inside. The cabinet 54 may be rolled on wheels 55 and pulled by a handle 56. Associated with structure 53 may be a display/interface 57. It may show a scene 58 of several subjects one or more of which have a face image captured by a camera. An inset 59 may alternatively display a face, or a close-up of an eye or iris of the subject being imaged. Items 57, 58 and 59 are shown in Figure 5b. An alternative instantiation of this invention may combine the face and iris acquisition functions into a single camera. This may combine the narrow field of view camera and the medium field of view camera with a single camera that could image the full face and the irises at a high resolution.

An alternative instantiation of this invention may simplify the WFOV camera by discarding the stereo sensor and use anthropometric information computed from the wide field of view imagery of the WFOV camera to provide an estimate of the range of the subject in the scene.

Another instantiation of this invention may have the narrow field of view PTZ camera performing the functions of the wide field of view camera, the medium field of view camera, and the narrow field of view camera into a single camera. This camera may use a longer zoom setting to do the wide field of view subject detection, a medium field of view zoom setting to do the face acquisition and a narrow field of view zoom setting to do the iris acquisition . In the present specification, some of the matter may be of a hypothetical or prophetic nature although stated in another manner or tense.

Although the invention has been described with respect to at least one illustrative example, many variations and modifications will become apparent to those skilled in the art upon reading the present specification. It is therefore the intention that the appended claims be interpreted as broadly as possible in view of the prior art to include all such variations and modifications .

Claims

What is claimed is:

1. A subject recognition system comprising: a wide field-of-view camera; a medium field of view camera; a narrow field of view camera; and a processing system connected to the wide field-of-view, medium field of view, and narrow field of view cameras.

2. The system of claim 1, further comprising a controller for driving a zoom of the narrow field of view camera.

3. The system of claim 2, wherein: the wide field-of-view camera is for locating positions and ranges of subjects in an area; the medium field of view camera is for capturing images of the subjects; and the narrow field of view camera is for capturing images of iris features of the subjects .

4. The system of claim 3, wherein: the subjects are people; and the features are irises.

5. The system of claim 4, wherein the medium field of view camera is for capturing images of faces of the people.

6. The system of claim 3, wherein the wide field of view camera is for estimating a distance between the cameras and the subjects.

7. The system of claim 3, wherein the distance to the subject may be inferred by stereo or from a calibrated camera and anthropometric knowledge.

8. The system of claim 3, wherein a correct zoom factor of the narrow field of view is estimated for capturing iris images.

9. The system of claim 5, wherein the processing system comprises a mechanism for storage and retrieval of images.

10. The system of claim 9 further comprising: a mechanism for face and iris matching and/or enrolling connected to the processing system; and an interface module connected to the mechanism for face and iris matching and/or enrolling and to the wide field of view camera, the medium field of view camera and the narrow field of view camera.

11. The system of claim 10, wherein the interface module comprises an iris processor connected to the narrow field of view camera and to the mechanism for face and iris matching and/or enrolling.

12. The system of claim 10, further comprising an illumination system connected to the interface module .

13. The system of claim 12, wherein the illuminator is an infrared iris illuminator.

14. A method of detecting and recognizing subject, comprising : detecting one or more subjects in a scene; acquiring face images of the one or more subjects; acquiring iris images of the one or more subjects; matching face images to stored faces in a database; and matching iris images to iris images in a database .

15. The method of claim 14, wherein a level of recognition of a subject comprises a matching of a face image of the subject and a matching of an iris image of the subject with corresponding images, respectively, in a database.

16. The method of claim 14, further comprising illuminating faces of the one or more subjects with infrared light for acquiring iris images.

17. The method of claim 14, further comprising prioritizing the subjects for recognition after detecting one or more subjects in a scene.

18. The method of claim 14, wherein: the detecting the one or more subjects is in a wide field-of-view; the acquiring the face images is in a medium field-of-view; the acquiring the iris images is in a narrow field-of-view; the detecting the one or more subjects in a wide field-of-view, acquiring face images in a medium field-of-view and acquiring iris images in a narrow field-of-view are achievable with two or less image sensors.

19. A combined biometrics recognition system comprising: a camera; a camera controller connected to the camera; and a feature processor connected to the camera; and wherein the camera has an adjustable field-of- view for acquiring one or more subjects in a scene and their features.

20. The system of claim 19, wherein: the field-of-view of the camera is adjustable to a wide field-of-view for acquiring an image of the one or more subjects; and the field-of-view of the camera is adjustable to a narrow field-of-view for acquiring an image of the at least one feature of a subject .

21. The system of claim 20, wherein: the one or more subjects are persons; and the at least one feature of a subject is a face and/or an iris.

22. The system of claim 21, further comprising a feature recognition processor connected to the camera controller.

23. The system of claim 22, further comprising: a storage mechanism connected to the feature recognition processor; and wherein the feature recognition processor is for matching images of features from the camera with images of features from the storage mechanism.

24. The system of claim 23, further comprising an infrared illumination system for illuminating features of images to be acquired by the camera.