论文部分内容阅读
SAR图像配准是实现多时相图像监测的前提,但是由于SAR图像的斑点噪声、成像机理的特殊性,使得采用常规的特征匹配方法很难实现SAR图像的自动配准。本文在介绍旋转尺度不变特征(SIFT)提取特征点原理的基础上,利用SIFT方法对两个时相星载SAR图像存在不同的旋转角和分辨率存在差异进行了特征点提取和自动匹配试验,图像特征点自动匹配的有效率达到70%以上,结果表明提取SIFT特征点进行多时相SAR图像自动匹配是基本可行的。
SAR image registration is a prerequisite for multi-temporal image monitoring. However, due to the speckle noise of SAR images and the particularity of the imaging mechanism, it is difficult to realize the automatic registration of SAR images by conventional feature matching methods. In this paper, based on the introduction of SIFT feature extraction principle, the SIFT method is used to extract feature points and automatically match the difference of rotation angle and resolution between two time-borne SAR images. , The efficiency of automatic matching of image feature points reaches more than 70%. The results show that it is feasible to extract SIFT feature points to automatically match multi-temporal SAR images.