Follow-up Study, Based On Temporal Correlation Of Moving Target Detection And Differential Contour Positioning

This paper concentrates on the method of moving objects detectionbased on time-space correlation and of moving objects location andtracking that used the differential contour, so we can detect and localizethe moving objects at a stationary scene.Firstly, the paper discuss the traditional three methods includingbackground difference, consecutive frames difference and optical flowthat are used to detect moving objects. And after pointing out theirapplications and their advantages and disadvantages separately, we giveour detection algorithm that based on time-space relativity. Ourdetection algorithm can avoid some disadvantages that exist in thetraditional methods, and improve the detection results.The traditional method on objects location and tracking is thecorrelation algorithm that based on template matching. But in the courseof practice, template matching will spend so much time that we more adoptthe improved correlation algorithm which try to raise efficiency bylooking for the better matching character and by minimizing the searchscope. And in fact, these efforts distinctly result in the occurrence ofsome excellent tracking algorithms. But because of the characteristicsof itself, the correlation algorithm is inefficient and fail to trace thenonrigid objects well. In order to overcome these disadvantages, in thispaper we give an algorithm based on the improved active contour model.In this algorithm, we try to get the rough contour from the consecutiveframes differential image firstly, and then by using snake algorithm,rough contour will converge to the object' s correct boundary which isthe result of moving objects location and tracking.