Similar image matching via global topology consensus

Abstract Recovering three-dimensional structure from images is one of the important researches in computer vision. The quality feature matching keys to obtaining more accurate results. However, as different objects or surfaces have similar with same elements and typography, camera pose estimation will be wrong task fail. This paper proposes a new mismatch elimination algorithm based on global topology consistency. We first formulate mathematical model constraints, then convert into grid matching, calculate confidence grids according changes angle displacement between correspondence vectors, remove mismatches low confidence. experiments demonstrated that our proposed method performs better than state-of-the-art methods accomplish outlier match rejection image could helped obtain correct reconstruct complete object models.