Researching On Moving Object Detection And Tracking In Video Images Based On The Feature

With the development of computer vision,the technology of object detection and tracking in video sequences become one of the important research direction.It not only integrates the knowledge of random mathematics,linear algebra,image processing,mathematical analysis and other related content of discipline and other fields.It have been widely applications whether in the military,civil,commercial,or traffic management in the field of intelligent identification,safety protection,etc.It has initially achieved a computer instead of the human eye,intelligent analysis of the movement of the outside world,for people to make judgments on the collection of video information Today.This paper introduces the purpose and significance of the study on the moving object detection and tracking in video images,and the main current situation of this research at home and abroad.The background difference method,the inter-frame difference method and the optical flow method are introduced in the moving object detection.In order to detect and recognize targets better,the video image needs to be denoised first.For the commonly used background modeling methods,this paper makes a horizontal and vertical comparison analysis to find the best possible algorithm,and finally decides to use the hybrid Gauss background modeling method to achieve target detection.In the aspect of moving object tracking,this paper is based on the Meanshift algorithm to track the object.Usually,the target model is set up first,then the feature of the target is extracted and analyzed to select a representative feature set.In each frame image in the next frame,above the target position as the center,a range of values of similar points and the characteristics,calculation of the Meanshift vector,and the feature vector point position is the most similar position to the target area determined by feature location.So in the Meanshift algorithm,feature extraction and similarity degree is particularly important.In the traditional meanshift algorithm,for high-speed or high speed moving targets,tracking failures often occur,and the fixed tracking window size is not suitable for a large change in the size of the video image.In this paper,the adaptive adjustment is introduced into the algorithm,and the recording and prediction of the overall motion direction and velocity of the target are introduced.The accuracy of the traditional meanshift algorithm in tracking the moving target with high speed and high speed is optimized.At the same time,the kernel function of the meanshift algorithm is processed,and the size of the tracking window is changed in real-time by changing the width of the kernel function,so that it is more suitable for the size of the target in the current video image,and the accuracy of the algorithm is improved...