基于单应矩阵的相对位姿改进算法  

Improved solution for relative pose based on homography matrix

作　　者：张培科[1,2] 武元新[2] 蔡奇[1,2] ZHANG Peike;WU Yuanxin;CAI Qi(School of Aeronautics and Astronautics, Central South University, Changsha 410083, China;School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China)

机构地区：[1]中南大学航空航天学院,长沙410083 [2]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《计算机工程与应用》2017年第15期25-30,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61422311);湖南省自然科学杰出青年基金(No.2015JJ1021)

摘　　要：从单应矩阵恢复相对位姿在视觉导航、视觉伺服应用中具有重要价值,证明了单应矩阵的新性质,并利用该性质改进了基于单应矩阵分解的相对位姿估计算法。与已有算法相比,该算法的候选解个数减半,并扩大了适用范围。理论分析、合成数据和真实数据测试均表明,改进方法使从候选解中筛选出唯一解的运算时间减少了50%,提高了位姿估计的总体运算效率。The relative pose solution from homography matrix is vital for the visual navigation and visual servo applications.This paper proves new properties of the homography matrix and utilizes them to improve relative pose solution bythe decomposition of homography matrix.Compared with previous methods,it only has a half of candidate solutions andenlarges the applicability range.Theory analysis,synthetic and real data tests show that the improved method saves50%time in choosing the right relative pose from the candidate solutions and improves the total computational efficiency.

关 键 词：单应矩阵 比例模糊性 相对位姿 奇异值分解 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]