基于多通道Gabor滤波器的高鲁棒灰度图像目标识别新方法

提出了一种针对低质量的灰度图像的基于多通道Gabor小波滤波器的高鲁棒目标识别新方法。主要是利用Gabor小波设计了滤波器,滤波器的中心频率是一个从低到高的范围。滤波器采用不同方向、不同尺度,从而组成多通道滤波器。对灰度图像直接进行小波变换,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灰度图像目标的特征,并对获得的小波特征归一化后输入到改进的BP神经网络分类器中进行分类识别。对四种不同的飞机灰度图像目标进行了分类识别仿真实验。结果表明,这种特征提取方法能有效地提取灰度图像目标纹理特征,并且对噪音和形状的变化具有强鲁棒性。在应用灰度图像对目标进行识别时,神经网络的训练时间减少到10min,识别率达到94%。

Robust gray image target recognition base on multi-channel Gabor wavelet filters

The method of low quality gray image target recognition is based on multi channel Gabor wavelets feature. Multi channels wavelet filters are designed by mostly Gabor wavelet, its center frequency is the range from low frequency to high frequency, its orientation and scale are different. Gary image is directly transformed by these wavelet filters, the feature of extracting gray image target is denoted by the coefficients of Gabor wavelet transform and its standard variance, the wavelet feature is normalized and input improved BP neural networks to classify. Finally, imitate experimentation is made by using 4 types plane gray image. The results indicate this method can effectively extract texture feature of gray image target, and has robust to noise and change of target shape.Not only neural networks training time is reduced to 10 minutes ,and recognition rate is 94% in applying to gray image target recognition by this method.