Intelligent Video Surveillance Technology Based On Gaussian Mixture Model And Camshift Tracking

In recent years, with the advancement of computer and image processing technology, the intelligent video surveillance technology is developing rapidly and playing a significant role in public safety. As the core of intelligent video surveillance technology, detection and tracking of moving targets have been hot topics in the field of computer vision. In this paper, we did a expand research on moving target detection and tracking technology, designend and implemented an intelligent video surveillance system.First, this paper described the relevant background and status of the video surveillance system. Then it studied the moving target detection and tracking techniques in detail, and proposed effective improvements. Finally, the overall system design framework and modules are described in detail, and make a detailed analysis of the system operation results.The traditional Gaussian mixture algorithm using scene is limited, inadequate and poor real-time, so we use block matching and skipping treatment strategies to reduce the complexity of the algorithm, use dual modle which based on Gaussian mixture model and frame difference method to improve the reliability of motion detection. The camshift algorithm has serious interference of background point, select a target manually, multiple iterations, easy tracking failures and other issues.So we proposed a camshift tracking method based on linear prediction. We reduced human intervention and the noise by extract the movement through frame-difference operation and use the motion detection results as the target of automatic target tracking. We improved the intelligence and accuracy of the algorithm.