Research On Moving Target Detection And Tracking Oriented To Intelligent Video Surveillance System

The detection and tracking of the moving targets in video sequences are the twocore technologies in intelligent video surveillance systems, They are the key tointelligence and real-time applications for video surveillance technology, also are theresearch focus of computer vision direction.On the research of the moving objects detection and segmentation, we mainlyresearch the background substraction method and Gaussion Mixture Model. As thebackground subtraction method is fast and simple, it has been widly used in thedetection of the moving targets in the monitoring system. The key of this algorithm isthe accuracy of background extraction, so we need improved the update strategies.Against more complex background conditions, this paper uses Gaussian mixture model,effictively solves interference factors of complex scenarios, by improving andoptimizing the relative parameters of the Gaussian Mixture Mode.On the research of the moving objects tracking, With the results of the movingtarget detection, this paper makes the weighted bounding rectangle as the tracking area,constructs a cost function using the centroid weighted bounding rectangle and the colorhistogram, achieves match tracking of multi-sport goals through the Kalman filter. Instatistics of the target color histogram, With the motion detection results as the maskimage, the color distribution of the foreground is calculated using the idea of imagepyramid. The computing speed of color histogram was improved.The practical result of the system shows that the proposed algorithms meet theneeds of real-time and have a good detection and tracking results.