Pattern recognition system

A device for recognizing authorized persons, comprising:

• first means for receiving at least one image and corresponding image identification information from each of a plurality of authorized persons;
• second means for processing the images received by said first means into image information;
• third means for determining the most distinctive aspects of said image information, said third means including means for performing an In-Class to Out-of-Class study including:
• means for generating an In-Class Variation Matrix;
• means for generating an Out-Class Variation Matrix;
• means for normalizing said In-Class Variation Matrix;
• means for normalizing said Out-Class Variation Matrix;
• means for generating a feature Matrix;
• means for normalizing said feature Matrix into a normalized feature Matrix;
• means for partitioning said normalized feature Matrix into bricks;
• means for prioritizing said bricks, and
• means for creating a feature template vector whose elements correspond to a subset of said bricks;
• fourth means for forming feature vectors of the magnitudes of said most distinctive aspects of said image information;
• fifth means for storing said feature vectors, said fifth means including neural network processor means adapted to store said feature vectors;
• sixth means for receiving a query image of a potentially authorized person;
• means for applying said query image to said fourth means to form a query feature vector corresponding to said query image;
• seventh means for querying said fifth means to determine whether said query image is within a predetermined range of similarity to any of the images of said plurality of authorized persons;
• eighth means, responsive to said seventh means, for outputting a signal indicative of whether said potentially authorized person is one of said plurality of authorized persons.

A device for recognizing the simultaneous presence of a plurality of authorized persons, comprising:

• first means for receiving at least one image and corresponding image identification information from each of a plurality of authorized persons;
• second means for processing said at least one image received by said first means into image information;
• third means for determining the most distinctive aspects of said image information, said third means including means for performing an In-Class to Out-of-Class study including:
• means for generating an In-Class Variation Matrix;
• means for generating an Out-Class Variation Matrix;
• means for normalizing said In-Class Variation Matrix;
• means for normalizing said Out-Class Variation Matrix;
• means for generating a feature Matrix;
• means for normalizing said feature Matrix into a normalized feature Matrix;
• means for partitioning said normalized feature Matrix into bricks;
• means for prioritizing said bricks, and
• means for creating a feature template vector whose elements correspond to a subset of said bricks;
• fourth means for forming feature vectors of the magnitudes of said most distinctive aspects of said image information;
• fifth means for storing said feature vectors, said fifth means including neural network processor means adapted to store said feature vectors;
• sixth means for receiving a query image from each of a predetermined number of potentially authorized persons, said predetermined number being two or more;
• seventh means for applying said two or more query images received by said sixth means to said fourth means to form feature vectors corresponding to said two or more query images;
• eighth means for querying said fifth means to determine whether said two or more query images are within a predetermined range of similarity to any of the images of said plurality of authorized persons;
• ninth means, responsive to said eighth means, for outputting a signal indicative of whether all of said two or more potentially authorized persons are authorized persons.

A pattern recognition system, comprising:

• first means for receiving a plurality of target images and corresponding target image identification information;
• second means for processing said plurality of target images into image information;
• third means for determining the most distinctive aspects of said image information, said third means including means for performing an In-Class to Out-of-Class study including:
• means for generating an In-Class Variation Matrix;
• means for generating an Out-Class Variation Matrix;
• means for normalizing said In-Class Variation Matrix;
• means for normalizing said Out-Class Variation Matrix;
• means for generating a feature Matrix;
• means for normalizing said feature Matrix into a normalized feature Matrix;
• means for partitioning said normalized feature Matrix into bricks;
• means for prioritizing said bricks, and
• means for creating a feature template vector whose elements correspond to a subset of said bricks;
• fourth means for forming feature vectors of the magnitudes of said most distinctive aspects of said image information;
• fifth means for storing said feature vectors, said fifth means including neural network processor means adapted to store said feature vectors;
• sixth means for receiving a query image;
• seventh means for generating a query feature vector from said query image;
• eighth means for querying said fifth means to determine if any previously stored feature vectors are within a predetermined range of similarity to said query feature vector and for generating a signal indicative of such.