Presentation 2000/2/22
A Method of Generating High Quality Online Character Recognition Dictionary Based on Training Samples
Katsuhiko Akiyama, Kazushi Ishigaki,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a method of generating a high quality dictionary from learning handwritten samples, for an online character recognizer which uses a time sequence of characteristic points as a dictionary template. Our method consists of two phases. In the first phase, we decide the optimal number of clusters by generating clusters using LBG, and generate the average template from each cluster by applying an elastic pattern matching algorithm. In the second phase, each average template is adjusted by a kind of LVQ to avoid miss-recognition between similar categories. In our experiment, the recognition accuracy improves about 2% while dictionary size increases little from 157KB to 166KB. Especially, the improvement is noticeable for small-stroke-categories.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) pen input / character recognition / online / recognition dictionary / generating pattern
Paper # PRMU99-235
Date of Issue

Conference Information
Committee PRMU
Conference Date 2000/2/22(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

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 Method of Generating High Quality Online Character Recognition Dictionary Based on Training Samples
Sub Title (in English)
Keyword(1) pen input
Keyword(2) character recognition
Keyword(3) online
Keyword(4) recognition dictionary
Keyword(5) generating pattern
1st Author's Name Katsuhiko Akiyama
1st Author's Affiliation Fujitsu Laboratories Ltd.()
2nd Author's Name Kazushi Ishigaki
2nd Author's Affiliation Fujitsu Laboratories Ltd.
Date 2000/2/22
Paper # PRMU99-235
Volume (vol) vol.99
Number (no) 649
Page pp.pp.-
#Pages 6
Date of Issue