【作者】 李振宇； 庄春华； 张亚庭； 王普；

【Author】 Li Zhenyu Zhuang Chunhua Zhang Yating Wang Pu (Dept of Electronic Information & Control Engineering Beijing University of Technology Beijing 100022 China)

【机构】 北京工业大学电子信息与控制工程学院；

【摘要】 针对三维重建对图像匹配的要求和传统立体匹配算法得到的边缘上匹配点数目少的问题,提出了一种图像拐角检测算法,在此基础上提出了一种基于区域和特征相结合的由粗到细的二次匹配策略。在第一次匹配时仅对显著性高的拐角进行匹配,并创新的引入拐角边缘线方向约束,以减少匹配点对的数目为代价来降低误匹配的概率,从而得到较精确的基础矩阵;在第二次匹配时重新调整拐角检测参数,并引入Harries角点检测,利用第一次匹配时已计算出的基础矩阵引导得到更多、更精确的匹配点对。实验结果表明,所提出的算法比传统算法能够得到更多的边缘上特征点之间的对应,鲁棒性好、精度和效率高。更多还原

【Abstract】 In order to gratify the need of 3D reconstruction system and to solve the problem of poorly matching points on edges,this paper proposes an image matching method based on the new algorithm of junction detection. This method adopts a strategy from rough matching to delicate matching.At first,the junctions whose strength are high enough are matched,the direction angles of the junctions’ branches are used as a matching constraint.The probability of mismatching is decreased on the cost of reducing the matching number.Then the fundamental matrix is estimated by the RANSAC and normalized 8-point algorithm.At last,the parameters of the junction detection are modified and the Harries corner detection is added to get more feature points.More precise matching points could be obtained through the guide of the fundamental matrix.The experiment results show that this algorithm has good performance on speed,accuracy and could get more edge matching points.更多还原