基于不变特征获取均匀控制点遥感影像配准算法.pdfVIP

DOI ：CNKI:11-4415/P1813.012 网络出版时间：2010-11-19 18:13 网络出版地址：/kcms/detail/11.4415.p1813.012.html 基于不变特征获取均匀控制点的遥感影像配准算法 ①② ① ② ③ 倪希亮 ,丁 琳 ,江 涛 ,胡顺石 （①中国科学院遥感应用研究所，北京100101；②山东科技大学，山东青岛，266510 ；③湖南师范 大学，长沙 410006 ） 【摘 要】本文针对大幅影像配准时计算量大、配准点分布严重不均匀以及错配率高等问题进行了 研究，提出一种可保持配准点均匀分布的快速配准算法。首先对基准影像和待配准影像进行分块提 取 SIFT 特征算子，然后局部自适应地对影像块中的匹配点进行几何约束，从局部选取最优匹配点 以达到提高图像匹配准确率的目标。通过对大量不同类型的影像进行配准实验，实验结果表明，当 图像存在传感器差异甚至较高分辨率差异时，本文算法也能够快速准确地达到配准目的。 【关键词】SIFT；尺度空间；特征算子；图像匹配 【中图分类号】TP751 【文献标识码】A 【文章编号】1009-2307 （2011）02- - A remote sensing image registration algorithm by obtaining uniform control points based on invariant feature Abstract ：In this paper, a study aiming to the problems of large images’ high computing complexity 、 serious non-uniform distribution of matching points and highly wrong registration rate were carried out. Then a fast matching algorithm was proposed. The algorithm could keep matching points uniformly distributed. First, it obtained SIFT points feature descriptor from base image and warp image by cutting up them. Then, it selected the best block matching points adaptively using geometric constraint to improve the image matching accuracy. Numerous experiments were conducted for different type images ．The results showed that the proposed method was faster and more accurate. Besides, when the sensors and resolutions of the images are very different, the method could also obtain better result. Key words: Scale Invariant Feature Transform (SIFT); scale space; fea