结合Harris和改进SIFT算法的遥感图像配准方法  

Research on Remote Sensing Image Registration Method Combining Harris and Improved SIFT

作　　者：王亚丽 WANG Yali(School of Computer Science and Technology,Kashi University,Kashi Xinjiang 844000,China)

机构地区：[1]喀什大学计算机科学与技术学院,新疆喀什844000

出　　处：《新疆大学学报（自然科学版）（中英文）》2021年第6期699-704,共6页Journal of Xinjiang University(Natural Science Edition in Chinese and English)

基　　金：喀什大学校内一般课题(2020)2738.

摘　　要：针对不同传感器或者不同波带的遥感图像配准中存在灰度差异性较大的问题,提出一种遥感图像配准算法.该算法通过Sobel算子与原图像卷积求取遥感图像间的一阶梯度图像来降低图像之间灰度差异,在梯度幅值图像上构建高斯尺度空间,在高斯尺度空间图像上检测Harris角点.用最近邻与次近邻比值法实现图像粗匹配,利用随机抽样一致性算法消除误匹配,并求得仿射变换模型参数,完成遥感图像配准.实验结果表明:该算法能克服遥感图像之间灰度差异问题,鲁棒性增强,与其它几种经典算法对比,对有较大灰度差异的遥感图像来说,可以成功实现配准,且配准精度较高.Aiming at the problem of large gray difference in remote sensing image registration of different sensors or different wavebands,a remote sensing image registration algorithm is proposed.This algorithm reduces the problem of grayscale differences between images by convolving the Sobel operator with the original image to obtain a gradient image between remote sensing images,and constructing Gaussian scale space on the gradient amplitude image,and detecting Harris corner on the Gaussian scale space image.The method of the ratio of the nearest neighbor to the next nearest neighbor is used to realize the rough image matching,the random sampling consensus algorithm is used to eliminate the mismatches,and the parameter of affine transformation model is obtained to register remote sensing images.The experimental results show that the algorithm overcomes the problem of gray-scale differences between remote sensing images and has enhanced robustness.Compared with several other classic algorithms,it can successfully achieve registration for remote sensing images with large gray-scale differences,and the matching precision is higher.

关 键 词：图像配准 遥感图像 尺度不变特征变换 HARRIS算子 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]