Presentation 2005-12-15
A Weighted Voting Method for Combining Multiple Character Recognition Engines
Shoichiro KANEKO, Hideaki GOTO, Hiroaki KOBAYASHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) It is known that combining multiple character recognition engines by majority logic is useful for improving the accuracy of character recognition. In the previous work, however, the suitability of each recognition engine for the input characters was not taken into account. On the other hand, several methods for automatic selection of recognition engines using the suitability of the engines have been proposed. We combine the above two approaches and propose a new method for combining multiple recognition engines. Experimental results show that the recognition accuracy improves from 97.8% to 98.2% using 14 Japanese character sets with 3195 characters each.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) character recognition / majority logic / multiple character recognition engines / confidence weight
Paper # PRMU2005-125
Date of Issue

Conference Information
Committee PRMU
Conference Date 2005/12/8(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 Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Weighted Voting Method for Combining Multiple Character Recognition Engines
Sub Title (in English)
Keyword(1) character recognition
Keyword(2) majority logic
Keyword(3) multiple character recognition engines
Keyword(4) confidence weight
1st Author's Name Shoichiro KANEKO
1st Author's Affiliation Graduate School of Information Sciences, Tohoku University()
2nd Author's Name Hideaki GOTO
2nd Author's Affiliation Information Synergy Center, Tohoku University
3rd Author's Name Hiroaki KOBAYASHI
3rd Author's Affiliation Information Synergy Center, Tohoku University
Date 2005-12-15
Paper # PRMU2005-125
Volume (vol) vol.105
Number (no) 477
Page pp.pp.-
#Pages 6
Date of Issue