Research on Moving Target Detection Method in Variable Background
Keqi Li, Sile Ma, Jiyang Chen, Yizhong Luan, Xiaojing Ma · 2019
It is the basis and research key of computer vision recognition to extract the moving target accurately and effectively from the video stream or image sequence. In this paper, a moving object detection method based on improved five-frame difference and edge detection is proposed to fit the motion target detection in the case of camera motion. By using the adaptive similarity threshold and the adaptive similarity preprocessing method based on fingerprint matching, the background overlap of the continuous frame image in the mobile background can meet the requirement of the frame background. The experimental results show that the proposed algorithm can improve the target recognition rate to 89%, and control the false positive rate to about 11%. It can detect the moving target accurately and meet the real-time requirement in the mobile background.