Presentation 1997/3/18
Character Extraction based on Multiscale Analysis
Hidekata HONTANI, Koichiro DEGUCHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a new character extraction method to extract characters from a gray level images based on multi-scale analysis of the images. This method treats a character as a tigure which looks blob-like globally, while line-like locally. In order to detect these both features of the global and local structures, our method blurs the image with a Gaussian filter, and observes the change of the shape of the image gray-level profile. As the scale of the Gaussian filter increases, the shape of the image gray-level profile changes. At smaller scale, ridge points are observed at the centers of the strokes of the character. At larger scale, all small structures are smoothed out, and an umbilic point may be observed at a center of the character. Based on this observation our method extracts character regions from a gray-level image.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) character extraction / multi-scale analysis / scale-space / characteristic points extraction
Paper # PRMU96-180
Date of Issue

Conference Information
Committee PRMU
Conference Date 1997/3/18(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) Character Extraction based on Multiscale Analysis
Sub Title (in English)
Keyword(1) character extraction
Keyword(2) multi-scale analysis
Keyword(3) scale-space
Keyword(4) characteristic points extraction
1st Author's Name Hidekata HONTANI
1st Author's Affiliation University of Tokyo()
2nd Author's Name Koichiro DEGUCHI
2nd Author's Affiliation University of Tokyo
Date 1997/3/18
Paper # PRMU96-180
Volume (vol) vol.96
Number (no) 598
Page pp.pp.-
#Pages 8
Date of Issue