论文部分内容阅读
给出了一种结合图像分割的合成孔径雷达(SAR)图像去噪算法,利用水平集图像分割方法将SAR图像分割得到多个连通区域,并利用基于结构相似性指数的非局部均值滤波(NLM-SSIM)去噪算法对每个连通区域进行去噪。对每个连通域分别去噪利于维持连通区域边缘的原有数值特征,同时也能够保证图像平滑区域的滤波效果,提高了去噪算法的性能。实验部分使用了合成孔径雷达图像中的道路、农田、沟壑和建筑图像块进行测试,将本文算法与非局部均值滤波(NLM)和NLM-SSIM算法进行了去噪效果比较,并通过等效视数(ENL)和边缘平均梯度比(EGR)评价指标验证了文中算法的有效性。
A synthetic aperture radar (SAR) image de-noising algorithm based on image segmentation is presented. The SAR image is segmented to obtain a plurality of connected regions by the level set image segmentation method, and the non-local mean filter based on the structural similarity index (NLM -SSIM) denoising algorithm for each connected area denoising. Denoising for each connected domain is advantageous for maintaining the original numerical features at the edge of the connected region, as well as for ensuring the filtering effect in the smooth region of the image and improving the performance of the denoising algorithm. The experimental part uses the road, farmland, ravines and building blocks in the synthetic aperture radar image for testing. Compared with the NLM and NLM-SSIM algorithms, the denoising effect of this algorithm is compared, (ENL) and edge-average gradient ratio (EGR) evaluation indicators verify the effectiveness of the proposed algorithm.