| Now computer technology and hardware equipment are large-scale used.Technology of video target tracking is becoming an indispensable key technology in area of computer vision.Many areas of advanced technologies and research achievements connect with video target tracking technology intimately,for example,pattern recognition,picture processing,automatic control,biology,physics,etc.At the same time,target tracking has been widely used in human-computer interaction,medical diagnosis and military monitoring,etc.But,video target tracking algorithms are facing lots of difficulties because of the complexity and diversity of video background and many of changes,for example,the changes of attitude and illumination and scale factors,etc.By summed up the field of target tracking classic algorithms and key technologies in recent years,this paper present a new algorithm that combined with super-pixel and block system and spatio-temporal context.Firstly,according to the spatial and temporal context learning algorithm,this paper introduces super-pixel division technology,which can use more image edge details information.Research is mainly focused on the spatial and temporal context learning algorithm.The concrete process of the algorithm is detailed introduced,which using bayesian framework model of target and its relative areas that combine with the relationship between time and space of local context area.Then calculate target posteriori probability for a maximum result,which is the optimal target coordinates.Secondly,because single target always losts in tracking problems,so block tracking strategy is raised in this paper.By using middle visual information of super-pixel segmentation,the new algorithm set up multiple trace point on target that can tracking by independent significance super-pixel blocks and get more accurate target center position.Finally,several experimental results are presented to verify accuracy of the algorithm for this paper. |
PDF Full Text Request