Method of selecting the best enroll image for personal identification

• US 5,978,494 A
• Filed: 03/04/1998
• Issued: 11/02/1999
• Est. Priority Date: 03/04/1998
• Status: Expired due to Term

• Current Assignee: Sensar, Inc. (Advent International Corporation)
• Original Assignee: Sensar, Inc. (Advent International Corporation)
• Inventors: Zhang, Guang Hua
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): Bali, Vikkram
• Application Number:
  US09/034,593
• Time in Patent Office: 608 Days
• Field of Search: 382/115, 382/117, 382/118, 382/209, 382/156, 382/255, 382/205, 382/225, 340/825.3, 340/825.4
• US Class Current: 382/117
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries

Abstract

A method is described for selecting an enrollment image or standard from multiple close-up images of a subject such as a human eye. The selected image is the one image that best represents the appearance of that subject to an image processing system which identifies subjects by comparison of later taken images of subject to the selected image.

Claims

1. A method of selecting a master enroll image against which other images will be compared using image matching techniques to identify a subject in the image, the master enroll image being selected from a plurality of images wherein each of said plurality of images is formed of pixels, the method comprising the steps of:

a) collecting a first set M of images of a subject, said first set M of images being collected at different focus settings;

b) computing a focus value F of each image in said first set M, and selecting a second set N of images having a focus value within a selected range;

c) comparing each image in the second set N of images with all of the other images in the second set N to determine if a match is found between that image and each other image in the second set N using the image matching techniques;

d) for each image in set N recording how many successful matches were made;

e) forming a third set S of images which is comprised of those images from set N having at least a predetermined number of successful matches; and

f) selecting the master enroll image from said third set S of images.

2. The method of claim 1 wherein step e) further comprises:

a) if said third set S of images has one member, selecting that image as the master enroll image; and

b) if said third set S of images has more than one member, selecting an enrollment image from the third set S based upon a selected feature of that image.

3. The method of claim 2 wherein the images are images of the human eye and a quality of the image is accessed using the Hamming distance associated with an iris code comparison.

4. The method of claim 3 wherein the selected feature is a median Hamming distance for each selected image in said third set S of images, and the master enroll image is selected as the image generating a comparison with the smallest median Hamming distance.

5. The method of claim 1 also comprising determining if any images have successfully matched at least 90% of the other images and if so selecting the enrollment image from such images and if not repeating steps a through f.

6. The method of claim 1 wherein each image in the first set M of images contains a person'"'"'s iris.

7. The method of claim 1 wherein the subject is at least a portion of an exterior of a package.

8. The method of claim 1 wherein each image contains a biometric feature of a human.

9. The method of claim 8 wherein the biometric feature is a fingerprint.

10. The method of claim 1 wherein at least some members of set M are created by producing subsequent images from a first image by digitally processing the first image to change focus of the first image thereby producing at least one subsequent image.

11. The method of claim 1 wherein the selected range for the focus value F is 0.30<

F<

0.60.

12. The method of claim 1 wherein the focus value F of step b is replaced by some other metric selected from the sets of metrics consisting of rotation of image and number of gray scales.

13. The method of claim 1 wherein the images are infrared images.

14. An apparatus for selecting a master enroll image against which other images will be compared using matching techniques to identify a subject in the image comprising:

a) at least one memory containing a first set M of images of a subject, said first set M of images having been collected at different focus settings, a second set of images N and a third set of images S;

b) at least one processor connected to the at least one memory the at least one processor being able to compute a focus value F of each image in said first set M, select a second set N of images having a focus value within a selected range, compare each image in the second set N of images with all of the other images in the second set N to determine if a match is found between that image and each other image in the second set N using the image matching techniques, record for each image in set N how many successful matches were made, form a third set S of images which is comprised of those images from set N having at least a predetermined number of successful matches, and select the master enroll image from said third set S of images.

15. The apparatus of claim 14 also comprising at least one imager which creates the first set of images and is connected to the at least one memory.

16. The apparatus of claim 15 wherein the imager is a camera.

17. The apparatus of claim 15 wherein the imager is a CMOS device.

18. The apparatus of claim 14 wherein the apparatus is contained within a computer.

19. The apparatus of claim 14 wherein the images are infrared images.