Research On Moving Target Detection And Tracking Algorithm Based On Video

In recent years,moving target detection and tracking have been widely used in the fields of behavior analysis,assistant driving systems,and robot vision.Different from the detection and tracking of other moving objects,because of the change of pedestrian posture,the similarity of clothing color and background or other interferences,and the ground material occlusion of other uncertain factors,pedestrian detection and tracking have become research difficulties in the field of computer vision.To achieving accurate pedestrian detection and tracking in complex monitoring scenarios,the moving target detection and tracking technology are researched deeply.And pedestrian detection and tracking algorithms with better anti-interference and accuracy are designed in the paper.The main research contents are as follows:1.Aiming at the difficulties faced by moving target detection and tracking,the principles of commonly used moving target detection algorithms and the advantages and disadvantages of each algorithm are analyzed and compared.Based on the original algorithm,an improved moving target detection algorithm combining background modeling and three-frame difference method is studied.The experimental results show that the improved moving target detection algorithm performs better than the traditional moving target detection algorithm in target integrity and accuracy.The principle,advantages and disadvantages of common pedestrian tracking methods are analyzed.2.With the comparison and analysis of two kinds of moving target detection algorithms under the condition of camera motion,the background motion compensation difference method is used to realize the moving target detection under dynamic background.The feature matching are used to complete the motion estimation of the background based on comparison of four algorithms which in the background motion estimation.Aiming at accurate estimation of background motion and the detection effect of moving target,which are influenced by the high error rate of scale-invariant feature transform(SIFT)matching algorithm,so an improved matching algorithm is adopted.Experiments show that the background motion compensation difference method based on feature matching eliminates the influence of background motion caused by camera motion on detection accuracy.At the same time,the effectiveness of the proposed method for moving target detection in dynamic background is verified.3.In order to solve the problem of camera movement,target color similarity with other interferences,and target occlusion,etc.,when pedestrian tracking,the advantages and disadvantages of particle filter tracking algorithm are analyzed,and algorithm of a multi-feature fusion particle filter is studied.in which the weights of each feature can be adjusted adaptively.Experiments show that when the target is close to the background color,the target is interfered by other objects and partially occluded,when the background is more complex,the improved tracking algorithm has higher tracking accuracy and stability.4.By constructing the experimental platform,the pedestrian detection and target tracking algorithms studied in the paper are tested on the open test video set and the self-photographed video set.The results show,when background motion,similar object interference,partial occlusion and target deformation have appeared in the sequence of the video,the motion target detection and tracking algorithms has higher accuracy and better anti-interference ability.