在基于SIFT算法的图像配准过程中有两个重要的环节:特征提取和特征匹配.针对算法在特征提取时存在的计算量大、复杂度高、速度慢等问题,该文提出了结合ORB算法的思想,对SIFT算法进行特征提取的优化,从而实现快速地提取图像的局部特征.在特征匹配阶段采用K最近邻的BBF搜索策略并结合RANSAC算法进行提纯,消除误配点.实验结果表明,改进后的算法降低了配准时间,提高了配准精度,适用于一些对实时性要求较高的场合.%In the process of image registration based on SIFT algorithm, there are two important steps: feature extraction and feature matching.In view of the problem of large computation, high complexity and slow speed in feature extraction,this paper puts forward the idea of combining ORB algorithm, the SIFT algorithm to optimize the feature extraction, so as to achieve the rapid extraction of local features of the image.In the feature matching stage, the BBF search strategy of K nearest neighbor is adopted and the RANSAC algorithm is used to purify and eliminate the error matching point.The experimental results show that the improved algorithm can reduce the registration time and improve the accuracy of registration, which is suitable for the occasions with high real-time requirements.
展开▼
