Abstract

Structure-from-motion (SfM) is an important computer vi- sion problem and largely relies on the quality of feature tracking. In im- age sequences, if disjointed tracks caused by ob jects moving in and out of the view, occasional occlusion, or image noise, are not handled well, the corresponding SfM could be significantly affected. In this paper, we address the non-consecutive feature point tracking problem and propose an effective method to match interrupted tracks. Our framework con- sists of steps of solving the feature ‘dropout’ problem when indistinctive structures, noise or even large image distortion exist, and of rapidly rec- ognizing and joining common features located in different subsequences. Experimental results on several challenging and large-scale video sets show that our method notably improves SfM.