Fingerprint verification system utilizing a facial image-based heuristic search method

• US 6,963,659 B2
• Filed: 09/12/2001
• Issued: 11/08/2005
• Est. Priority Date: 09/15/2000
• Status: Expired due to Fees

• Current Assignee: FaceKey Corporation
• Original Assignee: FaceKey Corporation
• Inventors: Xu, Tianning, Tumey, David M., Arndt, Craig
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): CARTER, AARON W
• Application Number:
  US09/952,096
• Publication Number:
  US 20020034319A1
• Time in Patent Office: 1,518 Days
• Field of Search: 382115-127, 340/5.1, 340/5.2, 340/5.52, 340/5.53, 902/3, 285 68- 69, 713/186
• US Class Current: 382/116
• CPC Class Codes: G06V 30/2504 Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches
• CPC Class Codes: G06V 40/1365 Matching; Classification

Abstract

A biometric verification system for controlling access is provided that does not rely on a non-biometric discriminator, such as a PIN or magnetic card, to convert a one-to-many verification task to a one-to-one verification task. The system enrolls authorized users by obtaining digitized fingerprint templates from them and storing them in a database. Video cameras and fingerprint sensors are provided for use in authenticating persons seeking access. Software compares a digital representation of a captured human facial image with stored facial images in a database of facial images, generating a match confidence therefrom and rank-ordering the database from highest to lowest match confidence. The software then compares captured human fingerprints with stored fingerprint templates associated with the rank-ordered database to verify the identity of the person and provide an output signal indicative of recognition.

Claims

1. A one-to-many biometric verification system utilizing facial image comparisons to enhance the efficiency of a fingerprint verification analysis, the verification system comprising:

a storage device containing a database of facial image templates and fingerprint templates acquired from a plurality of enrolled individuals, said database associating each facial image template with at least one corresponding fingerprint template acquired from the same enrolled individual;

a camera for obtaining a newly acquired digital representation of a human facial image;

a fingerprint sensor for obtaining a newly acquired digital representation of a human fingerprint;

a processor associated with said storage device, camera, and fingerprint sensor; and

software resident on said processor for comparing the newly acquired digital representation of a human facial image with the stored facial image templates, identifying a subset of more than one, but less than all, of said stored facial image templates that most closely match the newly acquired digital representation of a human facial image, then comparing the newly acquired digital representation of a human fingerprint with the fingerprint templates associated with said subset, and producing an output signal indicative of whether an enrolled individual is recognized;

wherein the software is adapted to produce an output signal indicative of whether an enrolled individual is recognized without relying on a non-biometric discriminator to convert a one-to-many verification task to a one-to-one verification task;

wherein the software is also adapted to recognize an enrolled individual solely from fingerprint comparisons if no facial image templates are stored in the database.

2. The verification system of claim 1, wherein the software, if it has recognized an enrolled individual solely from fingerprint comparisons but not from facial image comparisons, is adapted to enter a learning mode in which it generates a new facial image template from said newly acquired digital representation of a human facial image and add it to the database.

3. The verification system of claim 2, further comprising an administrative control center and a plurality of remote client terminals, each remote client terminal comprising a storage device, camera, fingerprint sensor, and processor as set forth in claim 1, wherein:

the administrative control center is adapted to enroll a plurality of individuals by generating templates of their fingerprints and storing them in a master biometric database, the master biometric database being distributed to each of the remote client terminals for storage on their respective storage devices; and

each remote client terminal generating facial image templates from their respective cameras for inclusion in their respective databases;

whereby the verification system compensates for lighting, camera setup, and other environmental variations at the remote client terminals that variably affect quality of image data acquisition.

4. A one-to-many biometric verification system utilizing facial image comparisons to enhance the efficiency of a fingerprint verification analysis, the verification system comprising:

a storage device containing a database of facial image templates and fingerprint templates acquired from a plurality of enrolled individuals, said database associating each facial image template with at least one corresponding fingerprint template acquired from the same enrolled individual;

a camera for obtaining a newly acquired digital representation of a human facial image;

a fingerprint sensor for obtaining a newly acquired digital representation of a human fingerprint;

a processor associated with said storage device, camera, and fingerprint sensor; and

software resident on said processor for comparing the newly acquired digital representation of a human facial image with the stored facial image templates, identifying a subset of more than one, but less than all, of said stored facial image templates that most closely match the newly acquired digital representation of a human facial image, then comparing the newly acquired digital representation of a human fingerprint with the fingerprint templates associated with said subset, and producing an output signal indicative of whether an enrolled individual is recognized;

wherein the software is adapted to produce an output signal indicative of whether an enrolled individual is recognized without relying on a non-biometric discriminator to convert a one-to-many verification task to a one-to-one verification task;

wherein the subset comprises no more than ten facial image templates that most closely match the newly acquired digital representation of a human facial image.

5. A one-to-many biometric verification system utilizing facial image comparisons to enhance the efficiency of a fingerprint verification analysis, the verification system comprising:

a storage device containing a database of facial image templates and fingerprint templates acquired from a plurality of enrolled individuals, said database associating each facial image template with at least one corresponding fingerprint template acquired from the same enrolled individual;

a camera for obtaining a newly acquired digital representation of a human facial image;

a fingerprint sensor for obtaining a newly acquired digital representation of a human fingerprint;

a processor associated with said storage device, camera, and fingerprint sensor;

software resident on said processor for comparing the newly acquired digital representation of a human facial image with the stored facial image templates, identifying a subset of more than one, but less than all, of said stored facial image templates that most closely match the newly acquired digital representation of a human facial image, then comparing the newly acquired digital representation of a human fingerprint with the fingerprint templates associated with said subset, and producing an output signal indicative of whether an enrolled individual is recognized;

wherein the software is adapted to produce an output signal indicative of whether an enrolled individual is recognized without relying on a non-biometric discriminator to convert a one-to-many verification task to a one-to-one verification task;

wherein the camera is a video camera having a field of view and being adapted to acquire a stream of images, and wherein the software is adapted to sense motion in the field of view to trigger an image preprocessing routine in which the software identifies whether the motion is caused by a human face, and if the motion is determined to be caused by a human face, to identify at least one highest quality facial image from said stream of images for comparison with said stored facial image templates.