Moving target detection based on adaptive edge extraction algorithm

摘要

Traditional three frames difference target detection algorithm is easy to be a hole phenomenon, especially when the holes appear in the target edges, the final binary image would be edge absence heavily, so it is difficult to obtain the whole target through morphological processing. Therefore, this paper proposes an adaptive edge extraction algorithm to obtain the target edges, which would make an edge supplement for the detection result of the three frames difference algorithm and then obtains the complete moving target through the morphological processing. This algorithm acquires the model parameters adaptively according to the moving characteristics of target, which could improve the accuracy and integrity of edge extraction. An experimental study performed shows that the proposed approach could extract moving targets from video sequences quickly and effectively.