Study On Moving Object Detection And Tracking Based On Image Sequence

The technology of moving object detection and tracking in image sequences is a key technology of Computer Vision System, It concerns many fields such as computer application, pattern recognition,image process, artificial intelligence and mathematics. Now, it has wide application in medical image analysis, robot navigation, intelligence transportation, video surveillance and other fields. So the research of moving object detection and tracking has important theoretical significance and practical value.The paper mainly studies the moving object detection and tracking in image sequences, which are captured by static single-camera. On the research of moving object detection:it mainly introduces the temporal difference detection method and background subtraction detection method. Based on temporal difference detection method, the paper implements temporal differential multiplied detection method, which has the anti-interference ability for small changes in the scene. And it also implements the Surendra estimation to update the background in the background subtraction. Concerned about the characteristics of temporal difference detection method and background subtraction detection method, the paper proposes a new method that combines temporal differential multiplied and Surendra background subtraction, which can detect the object better.Now the Mean shift is a common object tracking algorithm. In order to track the object rightly, the paper use Mean shift algorithm for tracking object. Because of the failure in the object occlusion with Mean shift algorithm, the paper designs the method that combines the Mean shift algorithm with Kalman filer for target tracking. When the object occlusion occurs, Kalman filter is introduced for object tracking. This method makes use of Kalman for parameter identification so as to make it has ability to estimate the coming state, which ensures the right tracking and high reliability when the object occlusion occurs.