Research On Target Detection Algorithm Based On Global Motion Estimation Compensation

Moving target detection is an important research topic in the field of Computer Vision(CV).It is the basis of tracking,recognition and behavior analysis in video sequence.Target detection technology is widely used in military warfare,intelligent vehicles,video surveillance,industry and biomedical science.This paper mainly focuses on moving object detection algorithms in dynamic background,and improves and innovates the existing global motion estimation compensation algorithms.The main research contents and achievements are as follows:(1)In this paper,through the experimental analysis of several common feature matching algorithms,the robust SURF algorithm is selected to extract the interframe motion vector and is used to estimate the parameters of the camera model.In view of the low contrast of foreground and background brightness in video sequence images,it is first proposed that histogram equalization is used to stretch the gray level of the image,then the SURF feature matching between frames is matched,and RANSAC algorithm is added to eliminate mismatch points.The experimental analysis shows that the improved SURF algorithm can increase the number of interframe feature points,while filtering the error matching points,it is more robust than the traditional SURF algorithm.(2)In this paper,on the basis of the global motion estimation compensation algorithm,a global motion estimation compensation algorithm based on the background motion vector extraction is proposed for target detection.First,SURF+RANSAC algorithm is proposed to extract motion vectors from the whole image.Secondly,the threshold is set according to the displacement of the foreground and background motion vectors,and the foreground?background motion vectors are distinguished.The background motion vectors are selected to estimate the parameters.Finally,the calculated parameters are used to compensate the background of the current frame,and the difference between the current frame and the compensation frame is directly used to detect the moving target.Compared with the direct frame difference and the traditional global motion estimation compensation algorithm,the algorithm can reduce the background motion interference caused by the motion of the camera,at the same time,it improves the accuracy of the estimation of motion parameters and accurately detects the moving target.(3)In view of the small proportion of the whole image in the video sequence image,the global motion estimation compensation method based on the foreground and the background regions is proposed.First,the original sequence images are divided into three frames.The pixel values of the differential images are all judged to be the foreground and background blocks as the threshold,and to mark the block of the background region.Secondly,the motion vectors are extracted from the labeled background blocks and the parameters are estimated.Finally,the computed parameters are compensated for the current frame image,and the motion target is detected by differential detection with the compensation frame.Experimental analysis shows that the algorithm is more accurate and more effective than the direct frame difference algorithm and the traditional global motion estimation compensation algorithm.