Research On Moving Target Detection And Tracking Of Image Guidance System

Precision-guided weapons,as the main combat weapon in high-tech and informationbased warfare,play a vital role in contemporary warfare.As an important guidance method for precision guided weapons,image guidance is mainly used to detect,track and accurately strike targets from complex backgrounds.In the detection stage,the target is small and the background is complex.While the target is moving,the seeker camera is moving at the same time.The coupling of the motion makes it difficult to separate the target from the background.In the tracking stage,there are many disturbances such as background change,scale change,occlusion,deformation and so on.Therefore,the research on automatic detection and adaptive tracking of moving targets has important research significance and engineering value for the realization of intelligent image guidance.Based on the problems of moving target detection and tracking in image guidance system,this paper makes a thorough study from three aspects: target detection,global tracking and local tracking based on feature point.The main points are as follows:Firstly,combining optical flow method with the motion compensation method,a moving target detection method based on Pyramid LK optical flow in dynamic background is proposed.At the same time,the improved OTSU algorithm is used to binarize the image,and the region growing method is used to fill the hole and remove the noise。Secondly,based on Background-Aware Correlation Filters,scale correlation filter is used to achieve scale adaptation.Meanwhile,sparse update strategy is used to improve the speed of the algorithm.The position filter response and color histogram response are fused in a fixed weight to compensate for the decrease in accuracy caused by speed optimization.Thirdly,in the local tracking stage,a tracking algorithm based on SURF features and optical flow computing was proposed.SURF features are extracted from the images and tracked by optical flow algorithm.Based on the consistency constraint,target size,rotation angle and center point position are calculated to achieve rapid local tracking of the targets.Finally,the algorithm is written based on VS2013 and OpenCV and the test is implemented on JN-IGH57 X platform.The results show that the algorithm proposed in this paper can detect moving targets in a dynamic background and track targets when their scales change rapidly.