Hash-based face recognition system
First Claim
1. A method of identifying, within an input test image, a specific item of an item class, said method comprising:
- (a) identifying a sample of said item class within said input test image, and aligning the identified sample to a canonical sample of said item class to create a fitted sample;
(b) applying a first patch pattern onto said fitted sample;
(c) applying a second patch pattern onto said fitted sample, said second patch pattern overlapping said first patch pattern;
(d) computing binary features for select overlapping patches of said first and second patch patterns, each computed binary feature being based on a choice of pixel patterns;
(e) creating a different hash key for each computed binary feature, said hash key being based on patch location in which the binary feature was computed and the choice of pixel patterns used in its computation and the binary value of the computed binary feature;
(f) using the created hash keys to access their corresponding entries in a hash table, each entry including identification (ID) information identifying an identity of a previously registered specific sample of said item class and the corresponding log probability of that identity generating a binary feature similar to the computed binary feature of the hash key at the specific patch location;
(g) summing the log probabilities of each identity found in the hash table using the created hash keys, sorting the found identifies by cumulative log probabilities to find the highest probability match;
(h) deeming the identity having the highest probability match to most closely match said specific item.
2 Assignments
0 Petitions
Accused Products
Abstract
In a face recognition system, overlapping patches are defined on a canonical face. Random clusters of pixel pairs are defined within each patch, and binary features are determined for each pixel pair by comparing their respective feature values. An inverted index hash table is constructed of the binary features. Similar binary features are then determined on a library of registrable samples of identified faces. A log probability of each registrable sample generating a binary feature from a corresponding cluster of pixel pairs at each specific patch location is determined and stored in the hash table. In a search phase, similar binary features are determined, and a hash key is determined for each binary feature. The log probabilities for each identity found in the hash table are summed for all clusters of pixel pairs and locations and sorted to find the high probability match.
-
Citations
18 Claims
-
1. A method of identifying, within an input test image, a specific item of an item class, said method comprising:
-
(a) identifying a sample of said item class within said input test image, and aligning the identified sample to a canonical sample of said item class to create a fitted sample; (b) applying a first patch pattern onto said fitted sample; (c) applying a second patch pattern onto said fitted sample, said second patch pattern overlapping said first patch pattern; (d) computing binary features for select overlapping patches of said first and second patch patterns, each computed binary feature being based on a choice of pixel patterns; (e) creating a different hash key for each computed binary feature, said hash key being based on patch location in which the binary feature was computed and the choice of pixel patterns used in its computation and the binary value of the computed binary feature; (f) using the created hash keys to access their corresponding entries in a hash table, each entry including identification (ID) information identifying an identity of a previously registered specific sample of said item class and the corresponding log probability of that identity generating a binary feature similar to the computed binary feature of the hash key at the specific patch location; (g) summing the log probabilities of each identity found in the hash table using the created hash keys, sorting the found identifies by cumulative log probabilities to find the highest probability match; (h) deeming the identity having the highest probability match to most closely match said specific item. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
Specification