论文部分内容阅读
论述利用小波变换提取合成孔径雷达 ( SAR)图像的纹理信息 ,借助 Daubechies3正交小波 ,对图像进行小波分解 ,将小波变换各个频带输出的 L1范数作为纹理分类的特征 ,并采用神经网络分类法对图像分类。选取徐州市卧牛山矿开采沉陷区的 Radarsat卫星 SAR图像进行分类研究表明 ,该方法能有效地对 SAR图像进行分类。
Discusses the use of wavelet transform to extract the texture information of Synthetic Aperture Radar (SAR) images. By means of Daubechies3 orthogonal wavelet, the image is decomposed by wavelet. The L1 norm output from each band of wavelet transform is taken as the feature of texture classification. Neural network classification Sort images. The classification of Radarsat satellite SAR images in the subsidence area of the Beiyuoshan Mine in Xuzhou City shows that this method can effectively classify the SAR images.