作者: Koichiro Yamaguchi , David McAllester , Raquel Urtasun
关键词:
摘要: We consider the problem of computing optical flow in monocular video taken from a moving vehicle. In this setting, vast majority image is due to vehicle's ego-motion. propose take advantage fact and estimate along epipolar lines egomotion. Towards goal, we derive slanted-plane MRF model which explicitly reasons about ordering planes their physical validity at junctions. Furthermore, present bottom-up grouping algorithm produces over-segmentations that respect boundaries. demonstrate effectiveness our approach challenging KITTI benchmark [11] achieving half error best competing general one third algorithm.