The Research Of Moving Object Detection And Tracking Algorithm In Serial Images

Motion target detection and tracking is one of the most important subjects in computer vision, which combines advanced technologies in image processing, pattern recognition, automatic control, artificial intelligence, computer and other relative fields. It has broadly applied in military visual missile guidance, video surveillance, medical image analysis, intelligent transportation and other fields. So this subject research has important theoretical significance and practical value.The research object of motion target detection and tracking is image sequence. Target detection in videos involves verifying the presence of an object in image sequence and possibly locating it precisely. Target tracking is to monitor an object's spatial and temporal changes during a video sequence, including its position, size, shape, etc. These two processes are closely related because tracking usually starts with detecting objects. Due to changes in illumination of the scene, background perturbations, precise detection and tracking of moving objects are still a challenge field of research.This essay analyzed the common methods on motion checking. Through combining the advantages and disadvantages, a new motion detection method is developed based on the background differencing and coterminous frame differencing. And this method developed the veracity of motion checking. For the background wound change with time and circumstance. a background model must be set up and refreshed. After roughly extracting the motion target, the treatments such as disposing the shadow, sticking the tab, analyzing the connected domain and so on were carefully researched.During the phase of the motion tracking, the essay estimates the motion location of next frame according to the feature that the changes of the target motion and the scene are rare between adjacent frames. To reduce the amount of calculation, the search can be arranged in the neighborhood of optimal matching. So this essay use kalman to track the center of motion target and refresh the result with estimated value every frame. To enrich the tracking means and drop the dependence of tracking, this essay researched the meanshift algorithm based on the color information. And from different sides this essay combined the meanshift and kalman. To adapt most reality, this essay analyzed the particle filter which applied to nolinear situation and the contrary of guass and improved the resampling algorithm. The experiment drew the essence of meanshift and integrate the information of color and location, whose weights are allotted. And excellent results have been obtained. Finally, the processing scheme of many motion targets is studied. Also, the experiment put up the parameter to judge the appearance of new targets and the disappearance and ancient targets.