An evolutive OCR system based on continuous learning

Author

Lebourgeois, Frank ; Henry, J.L.

Author_Institution

Eqiupe de Reconnaissance de Formes et Vision, Inst. Nat. des Sci. Appliquees, Toulouse

fYear

1996

fDate

2-4 Dec 1996

Firstpage

272

Lastpage

277

Abstract

The paper presents an evolutive OCR system based on a cooperation between the recognition stage and the contextual stage which makes possible continuous training. The authors use the contextual correction in order to modify the behavior of the recognition stage by adjusting the internal representation of character models. They also introduce a specific classifier suitable for continuous training. The proposed classifier is based on the k-nearest neighbor rule modified by the introduction of weights. During the continuous training, the system selects models of pattern which contribute actively to a correct recognition

Keywords

optical character recognition; pattern classification; classifier; contextual correction; contextual stage; continuous learning; continuous training; evolutive OCR system; internal character model representation; k-nearest neighbor rule; pattern model selection; recognition stage; weights; Character recognition; Context modeling; Dictionaries; Feature extraction; Feedback loop; Information analysis; Optical character recognition software; Pattern analysis; Pattern recognition; Reconnaissance;

fLanguage

English

Publisher

ieee

Conference_Titel

Applications of Computer Vision, 1996. WACV '96., Proceedings 3rd IEEE Workshop on

Conference_Location

Sarasota, FL

Print_ISBN

0-8186-7620-5

Type

conf

DOI

10.1109/ACV.1996.572073

Filename

572073