Mixed Pattern Segmentation by Using Chaotic Neural Networks

IPSJ Web Site 1 Citations 14 References

Search this article

Description

Many pattern segmentation systems have been developed by artificial neural networks. Those systems including the perceptron and the associative memory are capable to map an input to one of the memorized patterns. However in order to extract plural patterns from a mixed figure at once we have to embed a specialized mechanism in their systems. The segmentation problem is still one of the difficult problems in image processing and speech recognition. This paper proposes a new pattern segmentation system using chaotic neural network. The proposed system can discriminate several patterns at once even when they are overlapped each other. Our system is also reconfigurable because it consists of two simple components: ﾒback-propagation'' and ﾒchaotic neuron model'' where the combination of two components plays a key role.

Journal

Citations (1)*help

See more

References(14)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top