Presentation | 1999/7/16 Recognition of handwritten characters: a comparison between human reader and OCR using neural networks Kazuki SARUTA, Yoichi WATANABE, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In recent years the recognition of handwritten characters by OCR has been reported to mark as high as 98 -99% of accuracy rate. These results were obtained for the characters written in specific boxes or frames carefully. Such a high accuracy rate cannot be expected, however, for broken characters that result from rough printing. In the present paper we compared human readers' recognition ability for poorly handwritten or similar characters with the ability of the character recognition system using neural network (ELNET-II),for two handwritten character databases (ETL9B and IPTP). The experiments showed that human recognition ability was better, in comparison with OCR, especially in recognition of those characters that were broken or written in a cursive style. We can consider that the results depend on the difference of the processing ability of the local fluctuation of character strokes. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | handwritten character recognition / neural networks / poorly handwritten character / similar character |
Paper # | HIP99-22 |
Date of Issue |
Conference Information | |
Committee | HIP |
---|---|
Conference Date | 1999/7/16(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 | Human Information Processing (HIP) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Recognition of handwritten characters: a comparison between human reader and OCR using neural networks |
Sub Title (in English) | |
Keyword(1) | handwritten character recognition |
Keyword(2) | neural networks |
Keyword(3) | poorly handwritten character |
Keyword(4) | similar character |
1st Author's Name | Kazuki SARUTA |
1st Author's Affiliation | Faculty of Systems Science and Technology, Akita Prefectural University() |
2nd Author's Name | Yoichi WATANABE |
2nd Author's Affiliation | Faculty of Literature and Social Sciences, Yamagata University |
Date | 1999/7/16 |
Paper # | HIP99-22 |
Volume (vol) | vol.99 |
Number (no) | 186 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |