Research On Moving Object Detection And Tracing Related Algorithms

Moving target detection and tracking is one of the key technologies in this field and has been widely used in city intelligent security guard system,traffic navigation,intelligent weapons,air surveillance,missile warning and many other fields.So,moving target detection algorithm and tracking algorithm has great practical significance, From the static background,this paper made valuable exploration and improvement on moving target detection algorithm and tracking algorithm.According to moving object detection algorithm,classical algorithms such as finite difference method and the optical flow method are introduced,while their strengths and weaknesses are analyzed,inter-frame difference method with the principle of simple, low computational complexity,but not very accurate contour extraction,snake algorithm can provide a more accurate profile of the object,but there is need to pre-set the object contour,the problem of computing capacity.To improve performance,the thesis put forward by the inter-frame difference method(symmetric differential) marked the first goal of the general contour,and then snake through the iterative computing algorithm to calculate a more precise contour of the objectives for object tracking,in the snake algorithm iterative process,the introduction of Douglas-Peucker(DP) algorithm snake algorithm without lowering the accuracy of the calculation at the same time reduce the amount of experimental results show that the fusion algorithm can be used for real-time video detection.In the section of moving object tracking,mean-shift algorithm is used in a relatively large number of methods,convergence is faster,the characteristics of the object can be information and spatial information effectively combined to avoid the use of complex models to describe the target shape,appearance and its movement,but if the color of moving object is similar to the background,or objects will be a serious loss of the object block,the object size changes also may result in loss of tracking,paper presented an improved mean-shift algorithm,the mean-shift algorithm and together with Kalman filtering,in accordance with the objectives set by the ratio of block threshold,the original algorithm to improve the object motion in the process of change and as a result of light obscured a serious goal of factors such as loss caused by background clutter and interference.The experimental results show that the improved algorithm to improve the moving object tracking algorithm robustness.At the last section of this paper,a brief introduction OpenCV-based object detection and tracking of the prototype system was presented,which is based on papers that the proposed detection and tracking algorithm developed in the main description of the prototype system has the function modules,and then summed up the full text of the work,and further specified on the moving target detection and tracking of research direction and focus of the next step.