Presentation 1993/10/15
Study on a nonlinear normalization for pattern matching
Keisuke Hashimoto, Tadayoshi Shioyama,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) It is important in improving the recognition rate of handprinted Chinese characters to absorb the local variation in handprinting.A nonlinear normalization is considered as a method for absorbing the local variation.Projections of characters have informations of strokes parallel to direction of projection,which are useful to make a suitable normalized character for pattern matching.In this paper,we propose a nonlinear normalization method with the projections of an input character.For purpose of evaluating the proposed normalization method,we use the recognition algorithm called stroke density spectra method.In experimental results,we have obtained the recognition rate 96.4% for unlearned characters.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) local variation in handprinting / projections of an input character / stroke density spectrum
Paper # PRU93-69,HC93-43
Date of Issue

Conference Information
Committee PRU
Conference Date 1993/10/15(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) Study on a nonlinear normalization for pattern matching
Sub Title (in English)
Keyword(1) local variation in handprinting
Keyword(2) projections of an input character
Keyword(3) stroke density spectrum
1st Author's Name Keisuke Hashimoto
1st Author's Affiliation Department of Mechanical and System Engineering,Kyoto Institute of Technology()
2nd Author's Name Tadayoshi Shioyama
2nd Author's Affiliation Department of Mechanical and System Engineering,Kyoto Institute of Technology
Date 1993/10/15
Paper # PRU93-69,HC93-43
Volume (vol) vol.93
Number (no) 269
Page pp.pp.-
#Pages 7
Date of Issue