论文部分内容阅读
提出了一种基于Gabor小波变换的多尺度、多方向的SAR图像去除斑点噪声及纹理分割算法．根据SAR图像的特点设计一组Gabor滤波器，对SAR图像进行二维Gabor变换，得到一组滤波后多分辨率、多方向的图像．通过对滤波后的图像分别进行非线性变换，再用非相干均值平滑滤出斑点噪声，并计算每个像素在选定窗口内的能量，以此检测出纹理特征，用均方误差聚类方法得到分割的图像．给出对SAR图像进行纹理分割的满意实验结果，对照试验表明，该方法优于空间灰度共现矩阵方法．
A multi-scale and multi-directional SAR image removal algorithm based on Gabor wavelet transform for speckle noise and texture segmentation is proposed. According to the characteristics of SAR images, a set of Gabor filters are designed and two-dimensional Gabor transform is performed on the SAR images to obtain a set of filtered multi-resolution and multi-directional images. Through the non-linear transformation of the filtered image, the non-coherent mean smoothed out the speckle noise and calculate the energy of each pixel in the selected window to detect the texture features, using mean square error clustering method Get the segmented image. Satisfactory experimental results on texture segmentation of SAR images are given. The experimental results show that this method is superior to the spatial gray co-occurrence matrix method.