The Community for Technology Leaders
Green Image
Issue No. 08 - Aug. (2012 vol. 34)
ISSN: 0162-8828
pp: 1605-1617
D. Steedly , Microsoft Corp., Redmond, WA, USA
D. Nister , Microsoft Corp., Redmond, WA, USA
Yekeun Jeong , Microsoft Corp., Redmond, WA, USA
R. Szeliski , Microsoft Corp., Redmond, WA, USA
In-So Kweon , Dept. of Electr. Eng., KAIST, Daejeon, South Korea
In this paper, we present results and experiments with several methods for bundle adjustment, producing the fastest bundle adjuster ever published in terms of computation and convergence. From a computational perspective, the fastest methods naturally handle the block-sparse pattern that arises in a reduced camera system. Adapting to the naturally arising block-sparsity allows the use of BLAS3, efficient memory handling, fast variable ordering, and customized sparse solving, all simultaneously. We present two methods; one uses exact minimum degree ordering and block-based LDL solving and the other uses block-based preconditioned conjugate gradients. Both methods are performed on the reduced camera system. We show experimentally that the adaptation to the natural block sparsity allows both of these methods to perform better than previous methods. Further improvements in convergence speed are achieved by the novel use of embedded point iterations. Embedded point iterations take place inside each camera update step, yielding a greater cost decrease from each camera update step and, consequently, a lower minimum. This is especially true for points projecting far out on the flatter region of the robustifier. Intensive analyses from various angles demonstrate the improved performance of the presented bundler.
iterative methods, cameras, conjugate gradient methods, image motion analysis, image reconstruction, computer vision, bundle adjustment, computational perspective, convergence persepctive, block-sparse pattern, reduced camera system, block-sparsity, BLAS3, memory handling, variable ordering, sparse solving, exact minimum degree ordering method, block-based LDL solving method, block-based preconditioned conjugate gradient method, embedded point iteration, camera update step, structure from motion, 3D reconstruction, Cameras, Convergence, Sparse matrices, Jacobian matrices, Linear systems, Memory management, Barium, point iterations., Computer vision, bundle adjustment, structure from motion, block-based, sparse linear solving

D. Steedly, D. Nister, Yekeun Jeong, R. Szeliski and In-So Kweon, "Pushing the Envelope of Modern Methods for Bundle Adjustment," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 34, no. , pp. 1605-1617, 2012.
173 ms
(Ver 3.3 (11022016))