Research On Moving Object Detection And Tracking Algorithm In Video Images

Video tracking is an important task in computer vision and refers to the process of continuously inferring the state of a target in the video sequence.Target trajectory which generated by positioning a target in each frame of the video and the complete target area that provided at each moment are the tasks of video tracking.With the application of tracking technology in the intelligent video surveillance,visual navigation,human-computer interaction,virtual reality,medical diagnosis and other fields,it is more significant to study target tracking.Not only a large amount of data on the video signal processing are necessary for the detection and tracking of moving targets in video sequences,but also corresponding judgments on the results need to made,such as the analysis of the behavior of moving targets,motion trajectories and so on,which required to be precise algorithms.Based on the OpenCV secondary development platform,video image three aspects including preprocessing,moving target extraction detection and moving target tracking are studied in this paper.The main research work has been achived as follows:1.The research on video image preprocessing.Denoise and morphological was processed of the image for the video frame sequence image are easily affected by interference noise and color similarity factors,which has resulted in the target feature information to be more obvious and conducived to the detection and extraction of the target image.2.The research on moving target extraction detection of the video image.The method of algorithm combining round LBP operator and background subtraction was used to detecte the moving objects in video images for the problem that the traditional detection algorithm does not have high accuracy when the moving target is detected due to light changes.The specific method as follows: At first,calculating the LBP texturefeature map of the background frame and the current frame.Then,xoring the background pixels and the corresponding pixels of the LBP texture graph of the current frame to generate a foreground probability graph.Secondly,adapting update the background according to the change of the foreground target probability graph.Finally,uing the background phase subtraction to get the detection target.The correctness and effectiveness of the algorithm is verified by the experiment results.3.The research on moving target tracking of the video images.The road video image sequence is studied in this paper.In this environment,the video image sequence is easily interferes by blocking,similar target colors and the environment other factors.Thus,the combination of Kalman filter and Camshift algorithm was used to track the moving target of the video images and complete the relevant target tracking experiment verification in this paper.4.Finally,an experimental platform was set up to conduct experimental verification research on the above algorithm.Faster target search speed of the video image motion target detection and extraction algorithm,better robustness of the video image moving target tracking of the combination of Kalman filter and Camshift algorithm and good tracking effect are showed in the experiment results.