会议专题

Video Synchronization Based on Projective-invariant Descriptor

  In this paper,we present a novel trajectory-based method to synchronize two videos of the same dynamic scene,which are recorded by stationary un-calibrated cameras from different viewpointsor over different time intervals.We achieve that by first constructing a cross ratio-based projective-invariant descriptor for each trajectory point in the two videos,and then finding the matched trajectory points or time pairs between the two videos by using their descriptors.More importantly,a new five-coplanar-points structure(or set)is proposed for each trajectory point to compute its cross ratio.The five-points set includes a trajectory point as the reference one and other four fixed points induced from the background scene.Their corresponding original points in the 3D real-world are co-planar even if the trajectory points of targets in the scene are non-planar.Owing to the constructed five-coplanar-points structure,the proposed projective-invariant descriptor and the corresponding video synchronization method work well for the scenes with general non-planar motions as well as planar motions.Experimental results demonstrate that the proposed method significantly outperforms some state-of-the-arts for those given videos captured under different conditions,some example including different frame rates,wide baseline,multiple moving objects,planar or non-planar motions.

Video synchronization Projective-invariant descriptor Cross ratio Five-coplanar-points structure Non-planar motion

SHAO Bei LI Yajun ZHANG Qiang LIU Yi

Center for Complex Systems,School of Mechano-electronic Engineering,Xidian University,Xi”an,China

国内会议

2017中国智能物联系统会议

厦门

英文

134-140

2017-11-17(万方平台首次上网日期,不代表论文的发表时间)