Research On Body Tracking Based On Computer Vision

Object tracking is a popular problem in computer vision, it has a wide range of applications in visual surveillance, visual navigation of robots, human-computer interaction, and medical diagnose, etc. This article research the tracking technology when there is occlusion of similar object in the video based on visual surveillance.In this paper, following work has been done:On the research of moving object extraction, the paper researched main algorithms of object segmentation. By analyzing the advantages and drawbacks of these algorithms, a thresholding method for image segmentation is proposed, which adopts fixed threshold associated with adaptive threshold based on the gray value of head.When it comes to object tracking, an improved tracking algorithm is presented. This paper regards color distribution and texture as the observation model, and builds it by the combination of HSV color histograms and local binary pattern. In order to avoid interference during the tracking, the weight of each feature is introduced. On the other hand, these weights are and modified according to the similarity between target and local background to ensure the stronger distinguish feature always get more reliance.Finally, an object-chain is built for people counting based on the results of moving body tracking.On the basis of above research, experimental results show that compared with the original Mean Shift tracking algorithm, the algorithm proposed in this paper has a more stable robustness and self-recover capability when the object is sheltered by similar object.