含大面积平坦区域高分辨率遥感图像快速配准  

Fast registration method of high-resolution remote sensing image containing large flat area

作　　者：许夙晖 慕晓冬[2] 柯冰 陈晓 

机构地区：[1]第二炮兵工程大学研究生管理大队,西安710025 [2]第二炮兵工程大学信息工程系,西安710025 [3]第二炮兵工程大学通信办公室,西安710025

出　　处：《计算机应用》2013年第A01期182-185,共4页journal of Computer Applications

基　　金：国家863计划项目

摘　　要：针对含有大面积平坦区域的高分辨率遥感图像配准时由于需要遍历整个图像而导致的时效和精度不高的问题,提出了一种基于信息熵和加速鲁棒性特征(SURF)的快速配准算法,在提取特征点前剔除了图像中的平坦区域,提高了运算效率。在图像配准时,针对SURF算法容易提取大量错误点对的缺点,采用低时间复杂度的K-D树最近邻搜索法对SURF算法进行改进,并用基于相似三角形的错误匹配点剔除算法剔除错误点对,提高了匹配精度。实验结果表明,此算法在含有大面积平坦区域的高分辨率遥感图像配准中比SURF算法优越。Aiming at the low efficiency and matching accuracy of the registration on high-resolution image containing large flat area because of traversing the whole image, an image registration method based on Speed-Up Robust Feature （SURF） and entropy was proposed. And the fiat areas were eliminated before the feature point extraction to improve the operation efficiency according to the method. The SURF algorithm was improved based on high-time-efficient K-D tree nearest neighbor searching method. After that, similar triangle algorithm was used to remove the mismatched feature points and improve the matching accuracy. The experiment result shows that the proposed method is superior to SURF solving the registration of high-resolution image containing large flat area.

关 键 词：遥感 图像配准 平坦区域 信息熵 加速鲁棒性特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]