A paper form processing system with an error correcting function for reading handwritten Kanji strings
- Central Research Laboratory, Hitachi, Ltd., Tokyo (Japan)
This paper presents a paper form processing system with an error correcting function for reading handwritten kanji strings. In the paper form processing system, names and addresses are important key data, and especially this paper takes up an error correcting method for name and address recognition. The method automatically corrects errors of the kanji OCR (Optical Character Reader) with the help of word dictionaries and other knowledge. Moreover, it allows names and addresses to be written in any style. The method consists of word matching {open_quotes}furigana{close_quotes} verification for name strings, and address approval for address strings. For word matching, kanji name candidates are extracted by automaton-type word matching. In {open_quotes}furigana{close_quotes} verification, kana candidate characters recognized by the kana OCR are compared with kana`s searched from the name dictionary based on kanji name candidates, given by the word matching. The correct name is selected from the results of word matching and furigana verification. Also, the address approval efficiently searches for the right address based on a bottom-up procedure which follows hierarchical relations from a lower placename to a upper one by using the positional condition among the placenames. We ascertained that the error correcting method substantially improves the recognition rate and processing speed in experiments on 5,032 forms.
- Research Organization:
- Nevada Univ., Las Vegas, NV (United States)
- OSTI ID:
- 68597
- Report Number(s):
- CONF-9404212--
- Country of Publication:
- United States
- Language:
- English
Similar Records
Lexicon-based word recognition without word segmentation
An evaluation of an automatic markup system