Preprocessing means for use in a pattern classification system
First Claim
1. A method using a computer, having a memory and a processor, for recognizing an object by processing reference feature vectors comprising the steps of:
- generating a collection of reference feature vectors, each of said reference feature vectors stored in said memory. and representing a reference pattern that belongs to one of a plurality of predefined classes stored in said memory;
associating with each class all reference feature vectors representing reference patterns belonging to that class; and
generating using said processor, for a selected class, an associated hierarchy of one or more sets of possibility regions, said sets stored in said memory wherein for a selected set in the hierarchy the number of possibility regions in said selected set is significantly less than the number of reference feature vectors belonging to said selected class, and such that each reference feature vector belonging to said selected class is contained in at least one possibility region of said selected set, and such that each possibility region of said selected set contains relatively few reference feature vectors not belonging to said selected class.
11 Assignments
0 Petitions
Accused Products
Abstract
A method for recognizing an object by processing reference feature vectors a collection of reference feature vectors is generated; each representing a reference pattern that belongs to one of a plurality of predefined classes. Each class has associated all reference feature vectors representing reference patterns belonging to that class. For a selected class, an associated hierarchy of one or more sets of possibility regions is generated, wherein for a selected set in the hierarchy the number of possibility regions in said selected set is significantly less than the number of reference feature vectors belonging to said selected class, and such that each reference feature vector belonging to said selected class is contained in at least one possibility region of said selected set, and such that each possibility region of said selected set contains relatively few reference feature vectors not belonging to said selected class.
-
Citations
55 Claims
-
1. A method using a computer, having a memory and a processor, for recognizing an object by processing reference feature vectors comprising the steps of:
-
generating a collection of reference feature vectors, each of said reference feature vectors stored in said memory. and representing a reference pattern that belongs to one of a plurality of predefined classes stored in said memory; associating with each class all reference feature vectors representing reference patterns belonging to that class; and generating using said processor, for a selected class, an associated hierarchy of one or more sets of possibility regions, said sets stored in said memory wherein for a selected set in the hierarchy the number of possibility regions in said selected set is significantly less than the number of reference feature vectors belonging to said selected class, and such that each reference feature vector belonging to said selected class is contained in at least one possibility region of said selected set, and such that each possibility region of said selected set contains relatively few reference feature vectors not belonging to said selected class. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
-
19. A method using a computer, having a memory and a processor, for obtaining classification information useful in classifying an unknown pattern as belonging to an associated pattern class comprising the steps of:
-
predefining and storing a plurality of pattern classes in the memory; providing to said processor a set of input data associated with a set of reference patterns, each reference pattern belonging to one of said pattern classes; manipulating said input data in said processor to obtain a collection of reference feature vectors, each reference feature vector representing one of said reference patterns; associating with each class in said plurality of pattern classes all reference feature vectors representing reference patterns belonging to that class; generating for a selected class in said plurality of pattern classes, an associated set of certainty regions, wherein; the number of certainty regions in said associated set is significantly less than the number of reference feature vectors belonging to said selected class; each certainty region contains a plurality of reference feature vectors belonging to said selected class and does not contain reference feature vectors that do not belong to said selected class; and substantially all reference feature vectors belonging to said selected class are contained in at least one certainty region; and wherein said generating step comprises the steps of; (a) determining approximately the largest region which can be formed to contain one or more of said reference feature vectors of said selected class and not contain reference feature vectors which are not of said selected class; (b) determining approximately the largest region which can be formed to contain one or more of said reference feature vectors of said selected class which are not contained in a previously formed region and not contain reference feature vectors which are not of said selected class; and (c) repeating step (b) until substantially all of said reference vectors of said selected class are contained in one or more of said regions; and storing, for each pattern class, a set of classification information representing said set of certainty regions formed for each pattern class; and generating, for each pattern class, an associated hierarchy of possibility sets of possibility regions wherein for a possibility set, the number of possibility regions in said possibility set is significantly less than the number of certainty regions in a certainty set of certainty regions associated with said selected class and such that each reference feature vector belonging to said selected class is contained in at least one possibility region of said possibility set, and such that each possibility region of said possibility set contains relatively few reference feature vectors not belonging to said selected class. - View Dependent Claims (20, 21, 22)
-
-
23. A method of recognizing an object using a computer, having a memory and a processor, by processing reference feature vectors comprising the steps of:
-
predefining and storing a plurality of predefined pattern classes in said memory; generating, using said processor, a collection of reference feature vectors, each reference feature vector stored in said memory and representing a reference pattern; associating with each predefined pattern class all reference feature vectors representing reference patterns belonging to that class; and for a selected class; (a) generating, using said processor, an associated certainty set of certainty regions, said set of certainty regions stored in said memory of said computer, wherein; the number of certainty regions in said associated certainty set is significantly less than the number of reference feature vectors belonging to said selected class; each certainty region of said associated certainty set contains a plurality of reference feature vectors belonging to said selected class and does not contain reference feature vectors not belonging to said selected class; and substantially all reference feature vectors belonging to said selected class are contained in at least one certainty region of said associated certainty set of certainty regions; and (b) generating an associated hierarchy of one or more possibility sets of possibility regions wherein for a selected possibility set in the hierarchy the number of possibility regions in said selected possibility set is significantly less than the number of reference feature vectors belonging to said selected class, and such that each reference feature vector belonging to said selected class is contained in at least one possibility region of said selected possibility set, and such that each possibility region of said selected possibility set contains relatively few reference feature vectors not belonging to said selected class. - View Dependent Claims (24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55)
-
Specification