A rotation invariant printed Chinese character recognition system

In this paper, we present an approach to construct an optical character recognition (OCR) system for recognizing rotated printed Chinese characters. Our system includes preprocessing, feature extraction, candidates selection, and target character recognition. The proposed system has a three-stage structure designed mainly to reduce the time complexity in the recognition process. Our goal is to recognize the complete set of frequently used 13 053 printed Chinese characters with arbitrary orientations. The overall recognition rate of our system reaches 97.4%. Ó 2001 Elsevier Science B.V. All rights reserved.