基于子波分解的多通道神经网络纹理分割方法

描述了一种体现多通道滤波技术的神经网络纹理分割方法，决策神经网络（ＤＢＮ）可提高纹理分类的精度，同时纹理的子波变换降低了图像数据间的相关性，提高了网络的学习效率，实验表明本文提出孤方法分类误差较低，获得了令人满意的纹理分割效果。

A WAVELET TRANSFORMATION BASED MULTICHANNEL NEURAL NETWORK METHOD FOR TEXTURE SEGMENTATION

A neural network texture segmentation method, in which multichannel filtering is embodied, was proposed. Multichannel filtering technology is a very effective method for texture segmentation. Instead of using a general filter bank, the texture feature extraction and classification tasks were performed in this paper by the same unified neural network. Decision based neural network was adopted to improve the accuracy of classification. Wavelet transformation of texture was used to decrease the correlation of texture data and increase the efficiency of networks learning. Experiments show that the proposed method achieves lower error rates than other methods and a satisfactory result is obtained.