一种新的神经网络畸变不变模式识别方法

A New Method for Distortion-Invariant Pattern Recognition Using Neural Network

本文以Ｗｉｄｒｏｗ等人的方法为基础，提出一种新的神经网络畸变不变模式识别方法──最小距离畸变不变模式识别网络。数值模拟显示，新方法具有完全的畸变不变性、相当高的识别精度、实时的识别速度、强大的抗噪能力、极短的自适应时间以及显著降低的运算量。

Based upon the method of Widrow et al.,a new approach for distortion-invariant pattern recognition, minimum-distance distortion-invariant pattern recognition neural network, is proposed. Numerical simulation shows that, the new approach has complete distortion -invariance,fairly high recognition rate, real-time recognition speed, powerful noise resistance,shorter adaptive time and lower computation amount.