基于改进概率神经网络的纹理图像识别

引入差异演化(DE)算法来弥补基本概率神经网络的不足,从而提出一种基于改进概率神经网络(MPNN)的纹理图像识别方法。首先用树形结构小波包变换提取纹理图像的能量特征,用基于统计的纹理特征方法提取统计均值、平均能量、标准差和平均残余特征,得到纹理图像的特征矢量;然后用改进的概率神经网络训练纹理图像的特征矢量,从而实现纹理图像的识别。实验结果表明:采用基于改进概率神经网络的纹理图像识别方法较BP神经网络、RBF神经网络和基本的PNN有更高的识别正确率,且收敛更快。

Texture image recognition based on modified probabilistic neural network

The differential evolution method is introduced in this paper to make up the shortage of basic probabilistic neural network.Consequently,a new texture image recognition method based on Modified Probabilistic Neural Network(MPNN) is proposed.At first,it extracts the energy character with the shape of tree structure wavelet packet transform and extracts the statistical mean value,average energy,standard deviation,mean residual characteristics with the statistic method,The feature vector is obtained by the above characteristics.Then the feature vector of the texture image is trained by the MPNN.Thus the texture classification is identified.The experiment result indicates:compared with the BP neural network,RBF neural network and the basic probabilistic neural network,the modified probabilistic neural network has the higher accuracy and faster convergence speed.