Detection And Tracking Of Moving Object In Video Sequence

Computer vision have been developed greatly the past twenty years, and moving object detection and tracking issue, as an important part of the field of computer vision, attract more and more scholars involved to carry out extensive research work and achieved fruitful research results. Moving target detection and tracking technology is now being widely used in medical image analysis, artificial intelligence, video surveillance systems, and military fields, etc. With the enlargement of the scope of its application, to improve the robustness of complex environmental factors and testing precision of the tracking algorithm has become a current research emphasis in the field.Firstly, conventional algorithms of moving object detection and tracking arithmetic including each advantage and defect are analyzed and researched, and then we propose a method that combine the VIBE arithmetic and point trajectories segmentation algorithms based on spectral clustering. By use the VIBE algorithm to get the target model information and combine the spatio-temporal information of the point trajectories, we can detect object more accurately. To some extent, the algorithm is a kind of joint of target detection and tracking algorithm, when detect the object as well as completed the target tracking, and the tracking results as a feedback to correct the detection output, that may improve the detection accuracy and reduce the leakage detection, error detection and so on Experiments show that the algorithm of this paper, both on the standard test set and in their own recorded video test set, compared with other algorithms, both have better tracking effect.Finally, experiments show that the algorithm of this paper, both on the standard test set and in their own recorded video test set, compared with other algorithms, both have better tracking effect...