An Efficient Point-Matching Method Based on Multiple Geometrical Hypotheses

Point matching in multiple images is an open problem computer vision because of the numerous geometric transformations and photometric conditions that a pixel or point might exhibit set images. Over last two decades, different techniques have been proposed to address this problem. The most relevant are those explore analysis invariant features. Nonetheless, their main limitation all alone cannot reduce false alarms. This paper introduces efficient point-matching method for three views, based on combined use techniques: (1) correspondence extracted from similarity features (2) integration partial solutions obtained 2D 3D geometry. strength novelty determination point-to-point through intersection geometrical hypotheses weighted by maximum likelihood estimation sample consensus (MLESAC) algorithm. proposal not only extends methods descriptors but also generalizes perspective projection model views. developed has evaluated types image sequences: outdoor, indoor, industrial. Our strategy discards wrong matches achieves remarkable F-scores 97%, 87%, 97% industrial sequences, respectively.