Japanese Kanji character recognition using cellular neural networks and modified self-organizing feature map

Cellular neural networks for extracting line segment features are proposed. The features include a middle point, length and angle of the line segment. Based on these features, appropriate standard patterns are selected. The feature distribution of the standard patterns is mapped onto that of the handwritten pattern. Feature mapping with structural constraints, which can provide flexible mapping and very fast convergence, is proposed. Feature mapping results based on the similarity between the distorted pattern and the mapped standard ones, convergence rate and deviation from the standard patterns are estimated. Computer simulation demonstrates distortion-free feature extraction and flexible feature mapping.<<ETX>>

[1]  Kenji Nakayama,et al.  Handwritten alphabet and digit character recognition using feature extracting neural network and modified self-organizing map , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[2]  Kim T. Blackwell,et al.  Classification of Japanese Kanji using principal component analysis as a preprocessor to an artificial neural network , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[3]  Y. Kimura.  Distorted handwritten Kanji character pattern recognition by a learning algorithm minimizing output variation , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[4]  K Fukushima,et al.  Handwritten alphanumeric character recognition by the neocognitron , 1991, IEEE Trans. Neural Networks.