基于RANSAC算法的改进特征匹配方法研究  

Improved Feature Matching Method Based on Ransac Algorithm

作　　者：周艳青[1] 薛河儒[1] 马学磊[1] 白洁 姜新华[1] 杜雅娟 ZHOU Yanqing;XUE Heru;MA Xuelei;BAI Jie;JIANG Xinhua;DU Yajuan(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot 010011,China;School of Data Science and Application,Inner Mongolia University of Technology,Hohhot 010051,China)

机构地区：[1]内蒙古农业大学计算机与信息工程学院,呼和浩特010011 [2]内蒙古工业大学数据科学与应用学院,呼和浩特010051

出　　处：《内蒙古农业大学学报（自然科学版）》2021年第6期87-92,共6页Journal of Inner Mongolia Agricultural University(Natural Science Edition)

基　　金：国家自然科学基金项目(61461041,31960494);内蒙古自治区高等学校科学研究项目(NJZY19062);高层次人才引进科研项目(NDYB2018-50);内蒙古科技攻关项目(2020GG0169);内蒙古自治区自然科学基金(2020BS60003)。

摘　　要：为了提高图像特征点匹配的速度和准确性,提出基于RANSAC算法的极线约束特征匹配算法。首先,将特征点进行规范化处理,并拟选取8个特征点对,利用RANSAC方法计算出包含内点最多的基础矩阵,并优化基础矩阵;接着,求取极线方程,通过最小对极距离准则来寻找准确的匹配点,若匹配点和待匹配点到对应的极线距离小于阈值,且两距离之最小,该点保留,否则删除。最后,从匹配率、误匹配率两方面来分析提出的算法性能,并从计算准确性和适应性方面验证算法。实验结果表明,该方法的正确匹配率为98.18%,距离的平均相对误差为0.082 3%。In order to improve the matching velocity and accuracy,the feature matching method with epipolar line constraints based on the RANSAC algorithm was proposed in this paper. First,the feature points were normalized,and eight pairs of feature points were randomly selected,the fundamental matrix with the most interior points was computed and optimized by the RANSAC algorithm. Then,the epipolar line equation was obtained. Through the minimum epipolar distance criterion,the accurate matching points were found. If the distances from the matching points and the points to be matched to the corresponding epipolar line were less than the thresholds,respectively,and the sum of the distances was also less than the threshold,the points were retained,otherwise they were deleted. Finally,the performance of the proposed algorithm was analyzed based on the matching rate and mismatching rate,and the algorithm was verified by calculating the accuracy and adaptability. Experimental results showed that the correct matching rate was 98. 18%,and the average relative error of the distance was 0. 082 3%.

关 键 词：RANSAC算法 特征匹配 极线约束 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]