Title :
Document image decoding approach to character template estimation
Author :
Kopec, Gary E. ; Lomelin, Mauricio
Author_Institution :
Xerox Palo Alto Res. Center, CA, USA
Abstract :
An approach to supervised training of document-specific character templates from sample page images and unaligned transcriptions is presented. The template estimation problem is formulated as one of constrained maximum likelihood parameter estimation within the document image decoding (DID) framework. This leads to a two-phase iterative training algorithm consisting of transcription alignment and aligned template estimation (ATE) steps. The ATE step is the heart of the algorithm and involves assigning template pixel colors to maximize likelihood while satisfying a template disjointness constraint. In one large-scale experiment, use of document-specific templates resulted in a character error rate that was about an order of magnitude less than that of a commercial omni-font OCR program
Keywords :
document image processing; error statistics; iterative methods; learning (artificial intelligence); maximum likelihood decoding; maximum likelihood estimation; optical character recognition; DID; aligned template estimation; character error rate; character template estimation; constrained maximum likelihood parameter estimation; document image decoding; document-specific character templates; document-specific templates; sample page images; supervised training; template disjointness; template pixel colors; transcription alignment; two-phase iterative training algorithm; unaligned transcriptions; Error analysis; Heart; Image recognition; Image segmentation; Iterative algorithms; Iterative decoding; Labeling; Maximum likelihood decoding; Maximum likelihood estimation; Parameter estimation;
Conference_Titel :
Image Processing, 1996. Proceedings., International Conference on
Conference_Location :
Lausanne
Print_ISBN :
0-7803-3259-8
DOI :
10.1109/ICIP.1996.560730
