论文部分内容阅读
本文针对SAR和光学遥感图像的异源遥感图像匹配的问题,采用6种基于图像灰度信息的相似性度量方法进行研究和实验。在求得图像间旋转和尺度参数的基础上,重点比较了如何提取平移量。采用了包括归一化互相关、相似率、结构相似性、交互方差、互信息和f散度的方法,并根据SAR图像特性进行了部分改进,得到了不同场景下的实验结果,最后对结果进行了比较和分析。结果表明针对不同场景中匹配方法各有优势,其中归一化互相关方法和f散度方法在场景适应性及精度上表现出了很好的适应性。
In this paper, six kinds of similarity measure methods based on image gray information are studied and experimented for the problem of heterogeneous remote sensing image matching between SAR and optical remote sensing images. On the basis of obtaining the rotation between the images and the scale parameters, the emphasis is placed on how to extract the translation. The methods including normalized cross-correlation, similarity, structural similarity, interaction variance, mutual information and f divergence are adopted, and some improvements are made according to SAR image characteristics. Experimental results under different scenarios are obtained. Finally, Have carried on the comparison and the analysis. The results show that the matching methods in different scenes have their own advantages, of which the normalized cross-correlation method and the f-divergence method show good adaptability in scene adaptability and accuracy.