基于同态滤波的遥感薄云去除算法

目前常用的遥感薄云去除法是同态滤波法，但由于所采用的滤波器的局限性，传统的同态滤波法只能去除处于低频区域的云，而对处于高频区域的云则没有处理效果。因此在研究传统同态滤波算法，选取性能最佳的频域滤波器对图像进行处理的基础上，引入了空域3×3中值滤波，用像素点3×3邻域内9个像素点的中值来代替该点的像素值，旨在对前者无法去除的高频突变云区进行处理。而后结合主观视觉和客观统计分析两方面对仿真结果进行评价，经分析表明，用该方法处理过的影像在信息未受损失的情况下，高频突变云有所减少，达到了预期目的。

Algorithm for removing thin cloud from remote sensing digital images based on homomorphic filtering

The traditional homomorphic filtering frequently used in cloud-removing has effect on the clouds in low-frequency region, but is not effective for those in high-frequency region, because of the limitation of high-pass filters it used. Based on the study of the traditional homomorphic algorithm, the most effective high-pass filter in frequency domain is adopted. And then, in order to remove the cloud in high-frequency region, a popular nonlinear spatial filter, 3×3 median filter, is introduced, which replaces the value of the central pixel by the median of the gray levels in the 3×3 neighborhood of that pixel. At last, analysis of the experiment results in vision and statistics is carried out, which shows that the improved method is effective in removing the cloud in high-frequency region