Recovering one Missing Element at a Time in SFM under Occlusion
When there is occlusion, the measurement matrix collecting trajectories of features points in SFM would be incomplete. In this paper, we have presented a method to recover missing elements in an incomplete measurement matrix one by one. We also discussed the concept of Relevance between a target missing element and its relevant known elements, and a measurement matrix transformation process, utilizing in determining the order of the recovery of missing elements. Experiments on both synthetic and real data showed that our method work efficiently.
J. Z. Wang "Recovering one Missing Element at a Time in SFM under Occlusion", Key Engineering Materials, Vols. 439-440, pp. 664-673, 2010