基于SIFT图像配准算法优化研究

Research on Image Registration Algorithm Optimization Based on SIFT

为了满足在图像配准过程中对于准确度和稳定性的要求,提出一种基于SIFT和改进的RANSAC图像配准算法。首先采用SIFT算法进行特征点提取和最近邻匹配方法进行粗匹配,然后对RANSAC算法进行改进,对两幅图像中关键点相似性的评价函数进行改进和优化,实现关键点对的精准匹配,最后进行图像配准。实验表明,算法能够提高图像匹配的准确度同时降低图像匹配时间,具有一定的适用性。

In order to meet the requirements of real-time precision and stability in image registration process,this paper proposes an algorithm based on SIFT and improved RANSAC image registration.Firstly,this work used the SIFT algorithm to extract the feature points and used nearest neighbor matching method for rough matching.Then the paper improved the RANSAC algorithm,so as to improve and optimize the evaluation function of similarity of key points in the two images,and to achieve precise matching of key points.Finally,image registration was performed.Experiments show that the proposed algorithm can improve the accuracy of image matching and reduce image matching time,and has certain applicability.