增量式尺度估计下的相机位置解算  

Incremental scale estimation-based camera location recovery

作　　者：李梦晗 高翔 解则晓[1] 申抒含[2] Li Menghan;Gao Xiang;Xie Zexiao;Shen Shuhan(College of Engineering,Ocean University of China,Qingdao 266100,China;Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中国海洋大学工程学院,青岛266100 [2]中国科学院自动化研究所,北京100190

出　　处：《中国图象图形学报》2024年第10期2992-3007,共16页Journal of Image and Graphics

基　　金：国家自然科学基金项目(62373349,U22B2055,42076192)。

摘　　要：目的全局式从运动恢复结构(structure from motion,SfM)通过运动平均一次性恢复所有相机的绝对位姿,效率相对较高。运动平均中的平移平均主要负责解算相机在世界坐标系下的绝对位置,其求解过程因尺度歧义性、估计敏感性和求解不确定性的影响而较为困难。本文提出了一种基于增量尺度估计的平移平均方法,在消除尺度歧义性的同时提升了求解鲁棒性与准确性。方法本文将平移平均问题解耦为3个子问题:1)局部绝对尺度的增量式估计;2)全局绝对尺度的增量式估计;3)基于L1优化的尺度已知的绝对位置估计。结果在1DSfM数据集上进行对比实验,基线解算精度明显提升,解算相机百分比的均值达到96%。当引入两种不同的绝对旋转进行计算时,其绝对位置中值误差仅略差于BATA(bilinear angle-based translation averaging)与CReTA(correspondence reweighted translation averaging),排名第3,均值误差改善更为明显,分别排名第1和第2。相较于原始方法,本文方法在相机解算数量与位置解算精度上均有较大提升。结论本文方法综合了尺度分离思想与增量式参数估计思想,既消除了尺度歧义性,又保证了鲁棒性与高效性,求解所得的相机绝对位置稳定可靠。Objective The structure from motion(SfM)technique serves as the fundamental step in the sparse reconstruction process,finding extensive applications in remote sensing mapping,indoor modeling,augmented reality,and ancient architecture preservation.SfM technology retrieves camera poses from images,encompassing two main categories:incremental and global approaches.The global SfM,in contrast to the iterative nature of incremental SfM,simultaneously estimates the absolute poses of all cameras through motion averaging,resulting in relatively high efficiency.However,it still encounters challenges regarding robustness and accuracy.Rotation averaging and translation averaging constitute crucial components within the motion averaging.Compared with rotation averaging,translation averaging is more difficult due to the following three reasons:1)Only relative translation directions could be recovered by essential matrix estimation and decomposition,i.e.,the produced relative translations are scale ambiguous.2)Only cameras in the same parallel rigid component could their absolute locations be uniquely determined by translation averaging,while for rotation averaging,the requirement simply degenerates to the connected component.3)Compared with relative rotation,the estimation accuracy of relative translation is more vulnerable to the feature point mismatches and more likely to be outlier contaminated.In traditional approaches,the translation averaging method based on scale separation(L1SE-L1TA)calculates the relative baseline length between cameras before estimating the absolute locations,eliminates the scale ambiguity,and the solving range is no longer constrained by the camera triplet,but its robustness and accuracy still need to be improved.Incremental translation averaging(ITA)introduces the idea of incremental parameter estimation into the translation averaging process for the first time,which has good robustness and high accuracy.However,its solving process depends on camera triplets and may suffer from degeneracy during co

关 键 词：全局式从运动恢复结构 平移平均 尺度分离 基线长度求解 增量式参数估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]