The Moving-object Detection And Tracking Algorithm For The Complex Scene

Moving-object detection and tracking is the cutting-edge research in the field of computer vision. For a variety of factors that affect the algorithm results in the complex environment, this paper uses Gaussian mixture algorithms, particle filter algorithm as the main line, studies the target two aspects of the detection and tracking and combines with the image processing technology, subspace updating algorithm and mean shift algorithm. On the basis of the study of the traditional algorithms, we propose some new algorithms.The main contents of this paper are:(1) This paper presents an improved algorithm of Gaussian mixture model based on inter-frame differencing blocking model and adaptive learning rate for the problem of too large calculation, poor ability to adapt to the complex scenes and other issues. Introduction of the blocking model, effective integration of information of pixel ai rspace based on the inter-frame difference results to determine the suspicious for eground region and background region to improve the detection sensitivity; complex models used for suspicious areas to ensure the accuracy of the moving-object detection and simple model used to reduce the amount of computation; pass through ada ptive learning rate to accelerate the formation and regression of the background. E xperimental results show the algorithm can take into account the detection accuracy and computational cost.(2) In order to improve the robustness of visual tracking algorithm when the target appearance is rapidly changing, this paper presents a particle filter tracking algorithm based on mixture appearance models. On the particle filter framework, the histogram is utilized to roughly localize the object by a mean shift procedure. Then, a more precise eigenspace appearance model was invoked to infer the final state of the object. In this way, it will not only learn the changing trends of the target appearance rapidly, while avoiding the tracking drift by using the orthogonal subspaces. The e xperimental results show that the algorithm can keep strong robust in the light change, posture change, occlusion.(3) In order to combine the features the of the mean shift algorithm and particle filter algorithm, this paper presents a particle filter tracking algorithm based on mixture appearance models. On the particle filter framework, the histogram is utilized to roughly localize the object by a mean shift procedure. Then, a more precise eigenspace appearance model was invoked to infer the final state of th e object. The ex-perimental results show that it does not only avoid the tracking drift by using the orthogonal subspaces, but also learns the changing trends of the target appearance rapidly.