Recognition of handwritten similar Chinese characters by neural networks

This paper presents a multi-stage neural networks for the recognition of similar Chinese characters. In this research, the authors have developed a three stage recognition structure: 1) an overlapped c-means clustering algorithm to implement a coarse classifier; 2) a Bayesian decision based neural network as a fine classifier; and 3) a two-layered feedforward neural network for similar character recognition. A personal computer based prototype recognition system has been built. By using the CCL/NCCR1 database (5401 characters/spl times/200 samples) as a benchmark, the training and testing results show that the proposed prototype system achieves some improvement on the efficiency (recognition time of 0.885 second per character on a Pentium-90 based PC) and robustness (recognition rate of 90.12% without any rejection, and 94.11% with 6.7% of rejection, respectively).