Biometric recognition

Patent metadata

Field	Value
Publication number	US-2016196469-A1
Application number	US-201615043305-A
Country	US
Kind code	A1
Filing date	Feb 12, 2016
Priority date	Apr 24, 2013
Publication date	Jul 7, 2016
Grant date	—

Abstract

A fused image of the person's hand is accessed, the fused image having been generated using a segmented graylevel image and a segmented color image. The hand in the fused image is identified. One or more finger tips and one or more finger valleys in the fused image are identified. One or more fingers of the hand are segmented, based on the identified finger tips and finger valleys. The one or more fingers of the hand are labeled. One or more features for each finger of the hand are determined.

First claim

1 . A biometric verification system comprising:
a computer storage medium; and at least one processor configured to interact with the computer storage medium to perform operations comprising:
accessing a color image of a hand associated with a person, the color image having been captured by an image capture device;
generating, using the color image, a graylevel image of the hand;
segmenting the graylevel image of the hand generated using the color image into foreground portions and background portions;
segmenting the color image of the hand into foreground portions and background portions; and
combining the segmented graylevel image with the segmented color image based on the foreground portions and background portions of the segmented graylevel image and the foreground portions and background portions of the segmented color image to form a fused image of the hand.

2 . The biometric verification system of claim 1 , wherein segmenting the graylevel image of the hand into foreground portions and background portions comprises:
applying a threshold-based binarization process to pixels of the graylevel image; determining a first group of pixels in the graylevel image that satisfies a threshold of the threshold-based binarization process, and designating the first group of pixels as a foreground portion of the graylevel image; and determining a second group of pixels in the graylevel image that does not satisfy the threshold of the threshold-based binarization process, and designating the second group of pixels as a background portion of the graylevel image.

3 . The biometric verification system of claim 1 , wherein segmenting the color image of the hand into foreground portions and background portions comprises:
applying a k-means clustering process to pixels of the color image of the hand to generate k clusters of pixels; determining, for each cluster among the k clusters of pixels, an average RGB value for a center of the cluster; determining a first group of clusters, among the k clusters, for which the average RGB value for the center of the cluster satisfies a threshold, and designating the first group of clusters as a foreground portion of the color image; and determining a second group of clusters, among the k clusters, for which the average RGB value for the center of the cluster does not satisfy the threshold, and designating the second group of clusters as a background portion of the color image.

4 . The biometric verification system of claim 1 , wherein combining the segmented graylevel image with the segmented color image based on the foreground portions and background portions of the segmented graylevel image and the foreground portions and background portions of the segmented color image to form the fused image of the hand comprises:
determining, for a first pixel in the color image and a second pixel in the graylevel image that corresponds to the first pixel, whether at least one of the first pixel or the second pixel is designated in the foreground portions of the color image or the graylevel image; and based on a determination that at least one of the first pixel or the second pixel is designated in the foreground portions of the color image or the graylevel image:
determining a third pixel in the fused image that corresponds to the first pixel and the second pixel; and
designating the third pixel in a foreground portion of the fused image.

5 . The biometric verification system of claim 4 , further comprising:
based on a determination that neither the first pixel nor the second pixel is designated in the foreground portions of the color image or the graylevel image: performing a first comparison between the first pixel in the color image and at least one pixel in the foreground portions of the color image; performing a second comparison between the second pixel in the graylevel image and at least one pixel in the foreground portions of the graylevel image; and based on the first comparison and the second comparison, designating the third pixel in a foreground portion of the fused image.

6 . The biometric verification system of claim 5 , wherein:
performing the first comparison comprises determining a Euclidean distance between the first pixel and a center of a cluster of pixels generated by a k-means clustering process; and performing the second comparison comprises determining a difference between a value of the second pixel in the graylevel image and a threshold in a threshold-based binarization process.

7 . The biometric verification system of claim 1 , wherein combining the segmented graylevel image with the segmented color image based on the foreground portions and background portions of the segmented graylevel image and the foreground portions and background portions of the segmented color image to form the fused image, comprises:
determining whether a portion of the fused image corresponds to at least one of a portion of the segmented graylevel image that is segmented into a foreground portion or corresponds to a portion of the segmented color image that is segmented into a foreground portion; and based on a determination that a portion of the fused image does not correspond to at least one of a portion of the segmented graylevel image that is segmented into a foreground portion or corresponds to a portion of the segmented color image that is segmented into a foreground portion, determining whether a value, of the portion of the segmented graylevel image that is not segmented into a foreground portion, that reflects a likelihood that a portion is a foreground portion satisfies a threshold value and whether a distance between the portion of the segmented color image that is not segmented into a foreground portion and another portion of the segmented color image that is segmented into a foreground portion satisfies a distance threshold; and based on a determination that the value, of the portion of the segmented graylevel image that is not segmented into a foreground portion, that reflects a likelihood that a portion is a foreground portion satisfies the threshold value and that the distance between the portion of the segmented color image that is not segmented into a foreground portion and the other portion of the segmented color image that is segmented into a foreground portion satisfies the distance threshold, designating the portion of the fused image as a foreground portion.

8 . A computer-implemented method comprising:
accessing a color image of a hand associated with a person, the color image having been captured by an image capture device; generating, using the color image, a graylevel image of the hand; segmenting the graylevel image of the hand generated using the color image into foreground portions and background portions; segmenting the color image of the hand into foreground portions and background portions; and combining the segmented graylevel image with the segmented color image based on the foreground portions and background portions of the segmented graylevel image and the foreground portions and background portions of the segmented color image to form a fused image of the hand.

9 . The method of claim 8 , wherein segmenting the graylevel image of the hand into foreground portions and background portions comprises:
applying a threshold-based binarization process to pixels of the graylevel image; determining a first group of pixels in the graylevel image that satisfies a threshold of the threshold-based binarization process, and designating the first group of pixels as a foreground portion of the graylevel image; and determining a second group of pixels in the graylevel image that does not satisfy the threshold of the threshold-based binarization process, a

Assignees

• Accenture Global Services Ltd

Inventors

• Wabgaonkar Harshawardhan Madhukar
• Paul Sanjoy

Classifications

• G06V10/803

  of input or preprocessed data · CPC title

• G06F18/251

  of input or preprocessed data · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06V40/11 Primary

  Hand-related biometrics; Hand pose recognition · CPC title

• G06T7/0081

  Physics · mapped topic