A Text Conversion System Between Simplified and Complex Chinese Characters Based on OCR Approaches

An automatic conversion system capable of translating text between simplified and complex Chinese characters is presented in this paper. This OCR-based system demonstrates an efficient feature extraction algorithm to recognize either complex or simplified printed Chinese characters. A new postprocessing model is developed to facilitate meaningful conversion of words, as well as correction of character recognition errors. Experimental results show that the average recognition rates are about 99.2% and 95.3% for single-font and multi-font character recognition respectively. When tested with real documents printed in simplified Chinese characters, the recognition rate is 96.2% without using contextual information. Upon employing the proposed language model for postprocessing, the text conversion rate can be improved to 97.8%.