Object Tracking under Occlusions by Hierarchical ierarchical Part art Matching atching
To tackle occlusions, a hierarchical part matching method based on a layer layered appearance ed model for object tracking is pr presented esented in this paper, which integrates global and partial region matching together to search the target object in a coarse to fine manner. In order t to reduce the ambiguity of o object localization localization, only the discr, discriminative parts are selected iminative for similarity computing with respect to their cornerness measure. The similarity between parts is computed he in a layer ayer-wise wised manner, based on which the state of occlusion occlusions can be inferred correct correctly ly. When occluded partially, . the object can be localize localized accurately d accurately, when occluded completely, the historical information hen of motion is applied to predict its position by a Kalman filter filter. The proposed tracking method is tested . on practical video sequences, and the experimental results sho show it can consistently provides accurate positions of the object for stable tracking, even under severe occlusions. w
obje object track ct tracking ing hierarchical part matching matching layered appearance model occlusion analysis
Li Jin Weiguo Sun Sha He Hong Pan
School of Automation, Southeast University, Nanjing, China, 210096
国际会议
桂林
英文
1-7
2011-11-01（万方平台首次上网日期，不代表论文的发表时间）