System and method for unconstrained on-line alpha-numerical handwriting recognition
First Claim
1. A method for recognizing a to-be-recognized character using a character recognition database comprising the step of constructing said character recognition database by performing the steps of:
- inputting a plurality of character specimens for each model character to be recognized,for each inputted model character, organizing said specimens into at least one class, wherein the character specimens of at least one model character are organized into plural classes,extracting values of a feature vector, containing plural features, of each of said inputted character specimens, to produce a feature value vector, containing plural feature values, for each of said character specimens,for each of said classes, forming a mean feature value vector, containing plural mean feature values, as an unweighted average of said feature value vectors of said character specimens of said respective classes,storing as said character recognition database said plural mean feature value vectors, including plural mean feature value vectors for said at least one of said model characters for recognizing a to-be-recognized character as a respective model character, andfor at least one model character, adjusting said classes of said character specimens to stabilize said classes for character recognition by reassigning one or more character specimens to different classes of said respective model character and repeating said step of forming said mean feature value vectors of said adjusted classes.
1 Assignment
0 Petitions
Accused Products
Abstract
A character recognition system includes a character input device, such as a stylus and tablet or optical scanner, for receiving inputted characters, and a processor. The processor determines which of a number of model characters best matches the inputted character. To that end, the processor compares each inputted character to each of a plurality of classes into which the model characters are organized. Specifically, the processor extracts a feature value vector from the inputted character, and compares it to the mean feature value vector of each class. The processor recognizes the inputted character as the model character corresponding to the mean feature value vector which is closest to the feature value vector of the inputted character. The processor also constructs the database from multiple specimens of each model character. The processor organizes the specimens of each model character into multiple classes. The processor then determines the mean feature value vector of each class.
-
Citations
27 Claims
-
1. A method for recognizing a to-be-recognized character using a character recognition database comprising the step of constructing said character recognition database by performing the steps of:
-
inputting a plurality of character specimens for each model character to be recognized, for each inputted model character, organizing said specimens into at least one class, wherein the character specimens of at least one model character are organized into plural classes, extracting values of a feature vector, containing plural features, of each of said inputted character specimens, to produce a feature value vector, containing plural feature values, for each of said character specimens, for each of said classes, forming a mean feature value vector, containing plural mean feature values, as an unweighted average of said feature value vectors of said character specimens of said respective classes, storing as said character recognition database said plural mean feature value vectors, including plural mean feature value vectors for said at least one of said model characters for recognizing a to-be-recognized character as a respective model character, and for at least one model character, adjusting said classes of said character specimens to stabilize said classes for character recognition by reassigning one or more character specimens to different classes of said respective model character and repeating said step of forming said mean feature value vectors of said adjusted classes. - 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. A system for recognizing a to-be-recognized character using a character recognition database comprising:
-
a character input device for electronically receiving a plurality of inputted character specimens for each model character to be recognized in constructing a database and a processor connected to said character input device for electronically constructing a database by electronically organizing said c specimens into a plurality of classes wherein the character specimens of at least one model character are organized into plural classes said processor also for electronically extracting values of a feature vector, containing plural features, of each of said inputted character specimens, to produce a feature value vector, containing plural feature values, for each of said character specimens, for each of said classes, forming a mean feature value vector, containing plural mean feature values, as an unweighted average of said feature value vectors of said character specimens of said respective classes, and storing as said character recognition database said plural mean feature value vectors, including plural mean feature value vectors for said at least one of said model characters for recognizing a to-be-recognized character as a respective model character, and for at least one model character, adjusting said classes of said character specimens to stabilize said classes for character recognition by reassigning one or more character specimens to different classes of said respective model character and repeating said step of forming said mean feature value vectors of said adjusted classes. - View Dependent Claims (24, 25, 26, 27)
-
Specification