L. Ji and J.J. Zhang (USA)
Chinese handwriting recognition, neural network
Based on previous research in Chinese handwriting recognition [18], an improved system is described in this paper. At the heart of this recognition system lays sets of connected artificial neural networks. Input Chinese characters are partitioned into 12 categories according to the structure of each character. The first set of neural networks is trained to classify the 12 partitions, and the second set of neural networks is trained to recognize each individual partitions. The final recognition is based on the learned partition information. The improved system utilizes the shift-twist technique to pre-process input data, and some coding strategies.
Important Links:
Go Back
