光学景象匹配基准图的滤波方法研究

对光学景象匹配制导的实时图与基准图的特点进行了对比，分析了前人提出的实时图常用的滤波方法，认为这些方法并不完全适用于基准图的波波处理。然后根据基准图的特点提出了和同态滤波及最小方差滤波的两步法对基闪图进行滤波，这一方法特别适用于因薄云覆盖而引入的低频噪声的处理，并同时兼顾了图像的增强与恢复。通过实验证明，该方法效果良好，对于光学景象匹配基准图的预处理具有一定的实用价值。

A FILTERING METHOD OF SCENE MATCHING REFERENCE IMAGE

Scene matching reference image is contrasted with real time image in their feature. Common filtering methods of real time image are analyzed. It is considered that these methods of filtering are not fully suitable for reference image. Then, a new two step filtering method has been extracted according to the feature of reference image. The first step one is homomorphic filtering and the second step is minimum variance filtering. This method is suitable for the filtering of bump; it also considers the recovery and intensify of image. It is proofed by a simple experiment, that this method is valuable to image preprocessing of scene matching reference