Multi-frame Motion Segmentation by Combining Two-Frame Results

Abstract In this paper we consider the motion segmentation problem on sparse and unstructured datasets involving rigid motions, motivated by multibody structure from motion. particular, assume only two-frame correspondences as input without prior knowledge about trajectories. Inspired success of synchronization methods, address introducing a two-stage approach: first, is addressed image pairs independently; then, results are combined in robust way to compute final multi-frame segmentation. Our synthetic real experiments demonstrate that proposed approach very effective reducing errors among it can cope with large amount mismatches. Moreover, our method be profitably used build pipeline.