Fast-moving target tracking based on mean shift and frame-difference methods

(整期优先)网络出版时间:2011-04-14
/ 1
Themeanshifttrackerhasdifficultyintrackingfastmovingtargetsandsuffersfromtrackingerroraccumulationproblem.Toovercomethelimitationsofthemeanshiftmethod,anewapproachisproposedbyintegratingthemeanshiftalgorithmandframe-differencemethods.Theroughpositionofthemovingtargetisfirstlocatedbythedirectframe-differencealgorithmandthree-frame-differencealgorithmfortheimmobilecamerascenesandmobilecamerascenes,respectively.Then,themeanshiftalgorithmisusedtoachieveprecisetrackingofthetarget.Severaltrackingexperimentsshowthattheproposedmethodcaneffectivelytrackfirstmovingtargetsandovercomethetrackingerroraccumulationproblem.