An Occlusion Recovery Method Based on 3D Subspace

Shigang LIU,Yali PENG,Chongzhao HAN,Yinghua Wang
DOI: https://doi.org/10.3321/j.issn:0253-987X.2009.04.003
2009-01-01
Abstract:To recovery the occlusions during feature track, a method for occlusion recovery is presented based on 3D subspace under orthographic projection. The method mainly relies on the fact that the rows in the matrix including all the image points span the same linear subspace as the rows in the matrix including all the space points. A linear iterative algorithm is hence proposed. The advantages of the proposed method are that the positions of the occlusions are found from all the visible points, and all the images as well as all the image points are treated in the same way. The experiments with both simulated and real data sets show that the re-projection error of the method is approximately 15% higher than the image noise, and the method has good convergence property with small errors.
What problem does this paper attempt to address?