本文阐述了Boxcar滤波器和极化Refined Lee滤波器的滤波原理,比较了两种滤波器的滤波效果。通过滤波后的极化图像分类精度的比较发现,滤波能显著提高监督分类的精度,滤波是提高分类精度必不可少的技术环节,Boxcar滤波在提高分类精度方面要优于Refined Lee滤波,此外滤波窗口的大小对分类精度也有重要的影响。 This paper describes the filtering principle of Boxcar filter and polarized Refined Lee filter, and compares the filtering effects of the two filters. By comparing the classification accuracy of filtered polarimetric images, it is found that filtering can significantly improve the accuracy of supervised classification. Filtering is an indispensable technical step to improve the classification accuracy. Boxcar filtering is better than Refined Lee filtering in improving classification accuracy. In addition, The size of the window also has an important effect on the classification accuracy.