基于双概率神经网络的纹理图像识别

为提高纹理识别速度，在文献1纹理图像识别正确率较高的基础上，提出一种基于双概率神经网络（DPNN）的纹理图像识别方法。首先构造两个概率神经网络A和B，如果纹理特征明显，以较少的纹理特征能量特征作为网络A的输入参数即可识别，否则再加入统计特征和能量特征一起作为概率神经网络B的输入参数以达到较高的识别率。实验结果表明：采用双概率神经网络的纹理图像识别较文献1有更快的识别速度。

Texture Image Recognition Based on Double Probabilistic Neural Network

A new texture recognition method which is based on double probabilistic neural network is proposed in this paper for improving the recognition speed.It is on the basis of the literature 1 which has the higher recognition accuracy already.Construct two probabilistic neural network A and B at first,if the texture feature is obviously,only with the energy feature as the characteristic parameter of probabilistic neural network A can achieve the recognition,or add the statistical feature to the energy feature,ta...