Incremental scale estimation-based camera location recovery
Objective The structure from motion(SfM)technique serves as the fundamental step in the sparse reconstruc-tion 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:incre-mental and global approaches.The global SfM,in contrast to the iterative nature of incremental SfM,simultaneously esti-mates 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 tra-ditional approaches,the translation averaging method based on scale separation(L1SE-L1TA)calculates the relative base-line 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 trans-lation 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 collinear camera motion.To solve the above problems,this study proposes a translation aver-aging method based on incremental scale estimation(ISE-L1TA),which eliminates the scale ambiguity and enhances the method's robustness and result accuracy.Method Incremental SfM has been proven to be highly accurate and robust,mak-ing it a preferred choice for many applications.It has shown to be particularly effective in handling large datasets and over-coming the challenges posed by complex real-world scenarios.Recognizing its potential,researchers have sought to trans-fer the incremental parameter estimation ideology to other related tasks,such as incremental rotation averaging(IRA)and ITA.In particular,IRA is designed to estimate the camera absolute rotations incrementally and efficiently.Meanwhile,ITA is performed for the camera absolute locations,enabling it to handle outliers effectively and avoid error propagation.Overall,the adoption of incremental parameter estimation ideology for motion averaging tasks demonstrates the versatility and effectiveness of this approach.With its ability to handle complex datasets and overcome a range of challenges,the incremental parameter estimation ideology holds great promise for future research in the field of 3D reconstruction and beyond.In this study,ISE-L1TA is proposed by incorporating the scale separation strategy and incremental parameter esti-mation ideology.Specifically,the translation averaging problem is decomposed into three sub-ones and sequentially solved:1)incremental estimation of local absolute scale,2)incremental estimation of global absolute scale,3)scale-aware absolute location estimation based on L,optimization.The input of our proposed method is the pairwise scale invari-ant feature transform point matches,and its output is the absolute camera locations.First,the relative motion between cam-eras is obtained by estimating and decomposing the essential matrix.Next,the two-view triangulation is performed to calcu-late the relative depths in the local coordinate system.On the basis of depth ratios,incremental estimations are conducted for the local and global absolute scales.Subsequently,the relative baseline length between cameras is computed,and rota-tion averaging is performed for absolute rotation estimation,enabling the final scale-aware absolute location estimation.Result We performed experimental tests to evaluate the selection of scale distance function and scale distance threshold.The experimental results confirmed that the normalized perfect square deviation function effectively eliminates the impact of scaling effects.Furthermore,the incremental scale estimation method shows good robustness and insensitivity to scale dis-tance threshold and achieves remarkably higher baseline accuracy compared with L1SE.The experiments were conducted on the 1DSfM dataset.In comparison with various state-of-the-art methods including bilinear angle-based translation aver-aging(BATA),correspondence reweighted translation averaging(CReTA),ITA,and L1SE-L1TA,our proposed method exhibited the following performance:1)In terms of the number of cameras solved,the average percentage of successfully solved cameras using the proposed method is 96％.2)The median error of absolute location estimation is slightly worse than that of BATA and CReTA and ranks third overall under different absolute rotations.3)In terms of the mean error in absolute location estimation,the proposed method has remarkable advantages,ranking first and second respectively.Com-pared with the original L1SE-L1TA,the method in this study has a great improvement in the number of cameras solved and the accuracy of locations estimated.Conclusion The proposed method combines the concept of scale separation with incre-mental parameter estimation.By integrating these two ideas,our method effectively eliminates scale ambiguity while ensur-ing the effectiveness of outlier rejection and maintaining a concise solving process.As a result,the obtained absolute cam-era locations are stable and reliable.
global structure from motiontranslation averagingscale separationbaseline length computationincremen-tal parameter estimation
万方数据
维普
