Presentation 1995/9/28
Fast Handwritten Character Recognition with Large Scale Multi-templates
Shuji SENDA, Michihiko MINOH, Katsuo IKEDA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recently, it has been reported that quadratic discriminant functions are effective in handwritten character recognition. However, the quadratic discriminant functions have problems that they can not deal with the distribution other than the normal one and that their parameters are difficult to be esti-mated. In this paper, we apply a non-parametric discrimination, k nearest neighbor rule, to handwritten character recognition, using a large scale multi-template dictionary and a linear discriminant function. In order to accelerate the recognition process, we propose to use a rough classifier that decides character candidates by thresholding normalized distances. In the experiment of 3036-category classification with ETL9B, 9 nearest neighbor classifier achieved a high recognition rate of 97.48%.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) rough classification / multi-template / handwritten character recognition / ETL9B
Paper # PRU95-116
Date of Issue

Conference Information
Committee PRU
Conference Date 1995/9/28(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Understanding (PRU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Fast Handwritten Character Recognition with Large Scale Multi-templates
Sub Title (in English)
Keyword(1) rough classification
Keyword(2) multi-template
Keyword(3) handwritten character recognition
Keyword(4) ETL9B
1st Author's Name Shuji SENDA
1st Author's Affiliation Kyoto University()
2nd Author's Name Michihiko MINOH
2nd Author's Affiliation Kyoto University
3rd Author's Name Katsuo IKEDA
3rd Author's Affiliation Kyoto University
Date 1995/9/28
Paper # PRU95-116
Volume (vol) vol.95
Number (no) 278
Page pp.pp.-
#Pages 6
Date of Issue