ABSTRACT

The movement of a camera can cause different movement characteristics of the target and the background in images. Thus, it is difficult to effectively isolate moving targets from a dynamic background. Based on this concept, this paper proposes a target detection algorithm based on dynamic background compensation. First, the Harris corner detection algorithm is used to extract the feature points in two adjacent frames. To ensure the accuracy of camera motion model parameters, the algorithm combines with the outside point filtering algorithm to remove the false matching points and feature points on the moving target after preliminary feature point matching. Then, we use the camera motion model parameters for motion compensation after obtaining the relatively static background. Finally, we detect moving targets by using a common static background target detection algorithm.