System and method for automatically verifying identity of a subject
First Claim
1. A method for comparing a first image of a predefined image resolution comprising a first plurality of pixels and a second image of the predefined image resolution comprising a second plurality of pixels to determine if the images are substantially similar, said method being implemented in a data processing system having a memory, comprising the steps of:
- (a) acquiring the first image;
(b) defining a plurality of adjacent reference templates positioned over a substantial portion of the first image, wherein each of said reference templates encloses a different portion of the first image;
(c) designating one reference template of said plurality of reference templates as a primary reference template, and all other reference templates as secondary reference templates;
(d) defining around each of said reference templates, a corresponding individual first selection area centered on each reference template, each of said first selection areas being larger than its respective reference template;
(e) determining a best position for each reference template within said corresponding first selection area, wherein said best position of each reference template is representative of a most unique portion of the first image within the respective first selection(g) storing, in the memory as a reference file, the most unique portion of each respective first selection area, wherein the most unique portion corresponding to said primary reference template is designated as a primary reference segment and wherein each most unique portion corresponding to a particular one of said plurality of said secondary reference templates is designated as a plurality of said secondary reference segments;
(h) storing a set of coordinate positions of each of secondary reference segments relative to said primary reference segment in said reference file in the memory;
(i) defining a reference area comprising the pixels of the first image within said primary reference segment and said plurality of secondary reference segments;
(j) acquiring the second image;
(k) defining a first search area over a substantial portion of the second image;
(l) retrieving, from said reference file in the memory, said primary reference segment and said plurality of secondary reference segments;
(m) comparing said primary reference segment with all primary reference segment-sized portions of the second image within said first search area to determine a location of a closest match between said primary reference segment and a like-sized portion of the second image, wherein said like-sized portion is designated as a closest match portion;
(n) positioning the primary reference segment at said at least one closest match location and said plurality of secondary reference segments over said second image in accordance with said set of coordinate positions of each said secondary reference segment relative to said primary reference segment;
(o) defining a plurality of second search area templates, each second search area template corresponding to each of said secondary reference segments, each second search area template being centered on of each said corresponding secondary reference segments;
(p) comparing each secondary reference segment with all secondary reference segment-sized portions of the second image within its corresponding second search area to determine a best match position for each secondary reference segment within its corresponding second search area to obtain a plurality of best match portions of the second image corresponding to said plurality of secondary reference segments;
(q) defining a verification area comprising pixels of the second image within said closest match portion and said plurality of best-match portions; and
(r) comparing said reference area to said verification area to determine a total response value, said total response value being indicative of the degree of similarity between said reference area and said verification area.
1 Assignment
0 Petitions
Accused Products
Abstract
A system and method for verifying identity of a user seeking access to a secured area or object is provided. The system employs a novel global image comparison approach by first acquiring a reference fingerprint image from a user during an enrollment stage, obtaining a set of multiple reference segments from the reference image representative of unique portions thereof, and later during a verification stage obtaining a verification fingerprint image from a user seeking access to the secured area or object, determining the most likely positions for the multiple reference segments over the verify image, and then comparing all pixels of the reference image enclosed by the multiple reference segments to all pixels in a portion of the verify image overlaid by the multiple reference segments in their current positions over the verify image in accordance with a dynamically determined threshold that is individually tailored for the user providing the reference image during the enrollment stage. When the threshold is exceeded, the user seeking access is granted access to the secured object or area; otherwise the user is rejected. The system of the present invention includes the capability of determining multiple most likely positions for the multiple reference segments over the verify image and then sequentially matching the reference segments to the verify image portion overlaid by them in their current position against the dynamic threshold until a match is found or until it is determined that the verify and reference images are not matched. To speed up the verification process, the verify image and one of the multiple reference segments may be temporarily coarsened during the process of determining multiple possible most likely positions for the multiple reference segments.
-
Citations
17 Claims
-
1. A method for comparing a first image of a predefined image resolution comprising a first plurality of pixels and a second image of the predefined image resolution comprising a second plurality of pixels to determine if the images are substantially similar, said method being implemented in a data processing system having a memory, comprising the steps of:
-
(a) acquiring the first image; (b) defining a plurality of adjacent reference templates positioned over a substantial portion of the first image, wherein each of said reference templates encloses a different portion of the first image; (c) designating one reference template of said plurality of reference templates as a primary reference template, and all other reference templates as secondary reference templates; (d) defining around each of said reference templates, a corresponding individual first selection area centered on each reference template, each of said first selection areas being larger than its respective reference template; (e) determining a best position for each reference template within said corresponding first selection area, wherein said best position of each reference template is representative of a most unique portion of the first image within the respective first selection (g) storing, in the memory as a reference file, the most unique portion of each respective first selection area, wherein the most unique portion corresponding to said primary reference template is designated as a primary reference segment and wherein each most unique portion corresponding to a particular one of said plurality of said secondary reference templates is designated as a plurality of said secondary reference segments; (h) storing a set of coordinate positions of each of secondary reference segments relative to said primary reference segment in said reference file in the memory; (i) defining a reference area comprising the pixels of the first image within said primary reference segment and said plurality of secondary reference segments; (j) acquiring the second image; (k) defining a first search area over a substantial portion of the second image; (l) retrieving, from said reference file in the memory, said primary reference segment and said plurality of secondary reference segments; (m) comparing said primary reference segment with all primary reference segment-sized portions of the second image within said first search area to determine a location of a closest match between said primary reference segment and a like-sized portion of the second image, wherein said like-sized portion is designated as a closest match portion; (n) positioning the primary reference segment at said at least one closest match location and said plurality of secondary reference segments over said second image in accordance with said set of coordinate positions of each said secondary reference segment relative to said primary reference segment; (o) defining a plurality of second search area templates, each second search area template corresponding to each of said secondary reference segments, each second search area template being centered on of each said corresponding secondary reference segments; (p) comparing each secondary reference segment with all secondary reference segment-sized portions of the second image within its corresponding second search area to determine a best match position for each secondary reference segment within its corresponding second search area to obtain a plurality of best match portions of the second image corresponding to said plurality of secondary reference segments; (q) defining a verification area comprising pixels of the second image within said closest match portion and said plurality of best-match portions; and (r) comparing said reference area to said verification area to determine a total response value, said total response value being indicative of the degree of similarity between said reference area and said verification area. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
-
-
17. A data processing system for comparing a first image comprising a first plurality of pixels and a second image comprising a second plurality of pixels to determine if the images are substantially similar, said system comprising:
-
scanning means for acquiring the first and the second images; memory means for storing data; and control means, connected to said scanning means and said memory means, operable for; causing said scanning means to acquire the first image; defining a plurality of adjacent reference templates positioned over a substantial portion of the first image, wherein each of said reference templates encloses a different portion of the first image; designating one reference template of said plurality of reference templates as a primary reference template, and all other reference templates as secondary reference templates; defining around each of said reference templates, a corresponding individual first selection area centered on each reference template, each of said first selection areas being larger than its respective reference template; determining a best position for each reference template within said corresponding first selection area, wherein said best position of each reference template is representative of a most unique portion of the first image within the respective first selection area, said most unique portions being of the same size as the respective reference templates; moving each reference template to its corresponding best position; storing, in said memory means as a reference file, the most unique portion of each respective first selection area, wherein the most unique portion corresponding to said primary reference template is designated as a primary reference segment and wherein each most unique portion corresponding to a particular one of said plurality of said secondary reference templates is designated as a plurality of said secondary reference segments; storing, in reference file in said memory means, a set of coordinate positions of each of secondary reference segments relative to said primary reference segment; defining a reference area comprising the pixels of the first image within said primary reference segment and said plurality of secondary reference segments; causing said scanning means to acquire the second image; defining a first search area over a substantial portion of the second image; retrieving, from said reference file in said memory means, said primary reference segment and said plurality of secondary reference segments; comparing said primary reference segment with all primary reference segment-sized portions of the second image within said first search area to determine a location of a closest match between said primary reference segment and a like-sized portion of the second image, wherein said like-sized portion is designated as a closest match portion; positioning the primary reference segment at said at least one closest match location and said plurality of secondary reference segments over said second image in accordance with said set of coordinate positions of each said secondary reference segment relative to said primary reference segment; defining a plurality of second search area templates, each second search area template corresponding to each of said secondary reference segments, each second search area template being centered on of each said corresponding secondary reference segments; comparing each secondary reference segment with all secondary reference segment-sized portions of the second image within its corresponding second search area to determine a best match position for each secondary reference segment within its corresponding second search area to obtain a plurality of best match portions of the second image corresponding to said plurality of secondary reference segments; defining a verification area comprising pixels of the second image within said closest match portion and said plurality of best-match portions; comparing said reference area to said verification area to determine a total response value, said total response value being indicative of the degree of similarity between said reference area and said verification area; and comparing said total response value to a predetermined total threshold, and when said total response value exceeds said predetermined total threshold, accepting the second image as substantially similar to the first image, and otherwise rejecting the second image as being not substantially similar to the first image.
-
Specification