基于Gabor小波纹理特征的目标识别新方法

给出了一种基于Gabor小波纹理特征的目标识别新方法．主要是利用Gabor小波设计了一种多通道小波滤波器。对图像目标直接进行小波变换，用Gabor小波变换系数的模的平均值和其标准方差来表示抽取的图像目标的特征，把获得的小波特征归一化后输入到改进的BP神经网络分类器进行分类识别．最后。进行了一系列的仿真实验，结果表明，这种特征提取方法能有效提取图像目标纹理特征，并且对噪音和形状的变化具有鲁棒性．在应用于目标识别时，神经网络的训练时间减少到lOmin，识别率达到94％．

Novel method of target recognition based on Gabor wavelet texture feature

The novel method of target recognition based on Gabor wavelets feature is proposed. Mostly multi channels wavelet filters is designed using Gabor wavelet, image target is directly transformed by wavelet filters. The feature of extracting gray image target is denoted by the coefficients of Gabor wavelet transform and its standard variance. And the wavelet feature is normalized and input into improved BP neural networks to classify. Finally, a series of imitate experimentations are conducted. The results indicate that this method can effectively extract texture feature of gray image target which is robust to noise and change of target shape. Applying this method to gray image target recognition, neural networks training time is reduced to 10 minutes, recognition rate is 94%.