Non-iterative rigid 2D/3D point-set registration using semidefinite programming

摘要: We describe a convex programming framework for pose estimation in 2D/3D point-set registration with unknown point correspondences. give two mixed-integer nonlinear program (MINP) formulations of the problem when there are multiple 2D images, and propose relaxations both MINPs to semidefinite programs (SDP) that can be solved efficiently by interior methods. Our approach is non-iterative nature as we jointly solve correspondence. Furthermore, these readily incorporate feature descriptors points enhance results. prove exactly recover solution original nonconvex under noiseless condition. apply 3D models coronary vessels their projections obtained from intra-operative fluoroscopic images. For this application, experimentally corroborate exact recovery property absence noise further demonstrate robustness presence noise.