基于小波分解的星载SAR图像纹理信息提取

Extracting Textural Information of Satellite SAR Image Based on Wavelet Decomposition

论述利用小波变换提取合成孔径雷达 (SAR)图像的多尺度纹理信息 ,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范数作为纹理特征值。选取徐州市区局部的Radarsat卫星SAR图像 ,提取出纹理图像 ,实验证明 。

Texture extraction from SAR image is an important frontier field because SAR image has characteristics of high spatial resolution and abundant texture information. Wavelet transformation can provide an effective tool for multi scale texture analysis of SAR image. This paper describes a method to extract multi scale textural information from satellite SAR image using wavelet transformation. Firstly the SAR image is decomposed using Daubechies3 orthogonal wavelet and l 1 norms of every band image are computed as feature value of image pixels. According to the feature value on every channel, texture images which respectively denote texture on different scale of SAR image are produced. The method has been applied for the extraction the textural informtion from a Radarsat image of Xuzhou city. The experiment shows that the facial target textural information can be effectively extracted using the mehtod.