Method and apparatus for implementation of neural networks for face recognition
First Claim
1. A real-time object recognition system comprising:
- image detector means for capturing an input image of an object to be recognized;
means for storing a plurality of reference images, each of said reference images being indicative of a cluster of images of an object on record; and
a two-layer neural network comprising,a first layer for nonlinear joint correlation of said input image with each corresponding said reference image to generate a corresponding correlation signal, anda second layer having nonlinearity for combining said correlation signals to provide a combined correlation signal, wherein said combined correlation signal above a threshold is indicative that the object to be recognized is the object on record.
1 Assignment
0 Petitions
Accused Products
Abstract
A method and apparatus for implementation of neural networks for face recognition is presented. A nonlinear filter or a nonlinear joint transform correlator (JTC) employs a supervised perceptron learning algorithm in a two-layer neural network for real-time face recognition. The nonlinear filter is generally implemented electronically, while the nonlinear joint transform correlator is generally implemented optically. The system implements perception learning to train with a sequence of facial images and then classifies a distorted input image in real-time. Computer simulations and optical experimental results show that the system can identify the input with the probability of error less than 3%. By using time multiplexing of the input image under investigation, that is, using more than one input image, the probability of error for classification can be reduced to zero.
-
Citations
62 Claims
-
1. A real-time object recognition system comprising:
-
image detector means for capturing an input image of an object to be recognized; means for storing a plurality of reference images, each of said reference images being indicative of a cluster of images of an object on record; and a two-layer neural network comprising, a first layer for nonlinear joint correlation of said input image with each corresponding said reference image to generate a corresponding correlation signal, and a second layer having nonlinearity for combining said correlation signals to provide a combined correlation signal, wherein said combined correlation signal above a threshold is indicative that the object to be recognized is the object on record. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 59)
-
-
32. A method for real-time object recognition comprising the steps of:
-
capturing an input image of an object to be recognized; storing a plurality of reference images, each of said reference images being indicative of a cluster of images of an object on record; and nonlinearly jointly correlating said input image with each corresponding said reference image at a first layer of a two-layer neural network to generate a corresponding correlation signal, and combining said correlation signals at a second layer having nonlinearity of said two-layer neural network to provide a combined correlation signal, wherein said combined correlation signal above at least one threshold is indicative that the object to be recognized is the object on record. - View Dependent Claims (33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 60)
-
-
47. A real-time object recognition system comprising:
-
image detector means for capturing an input image of an object to be recognized; a plurality of reference images, each of said reference images being indicative of a cluster of images of an object on record, said reference images stored on an optical film deposited on a card; and a two-layer neural network comprising, (1) a first layer comprising, (a) a light source for illuminating each of said reference images to project corresponding reference images, (b) a spatial light modulator receptive to said input image, said spatial light modulator in response to illumination by said light source projecting said input image, and (c) detector means for detecting combined corresponding said reference images and said input image to provide a corresponding nonlinear joint correlation signal, each of said joint correlation signals being indicative of a joint correlation between said input image and each corresponding said reference image, and (2) a second layer having nonlinearity for combining said correlation signals to provide a combined correlation signal. - View Dependent Claims (48, 49, 52, 61)
-
-
50. The system of 47 wherein said spatial light modulator comprises an optically addressed spatial light modulator.
-
51. The system of 47 wherein said spatial light modulator comprises a liquid crystal light value.
-
53. A method for real-time object recognition comprising the steps of:
-
capturing an input image of an object to be recognized; storing a plurality of reference images each of said reference images being indicative of a cluster of images of an object on record, storing said reference images on an optical film deposited on a card; at a first layer of a two-layer neural network, (a) illuminating each of said reference images to project corresponding reference images, (b) illuminating a spatial light modulator receptive to said input image to project said input image, (c) detecting combined corresponding said reference images and said input image to provide a corresponding nonlinear joint correlation signal, each of said joint correlation signals being indicative of a joint correlation between said input image and each corresponding said reference image; and combining at a second layer having nonlinearity of said two-layer neural network said correlation signals to provide a combined correlation signal, wherein said combined correlation signal above at least one threshold is indicative that the object to be recognized is the object on record. - View Dependent Claims (54, 55, 58, 62)
-
-
56. The method of 53 wherein said spatial light modulator comprises an optically addressed spatial light modulator.
-
57. The method of 53 wherein said spatial light modulator comprises a liquid crystal light value.
Specification