Motion Detecting And Tracking Based On Video

In recent years, visual analysis of moving object has become one of the most important research directions in the area of computer vision. It aims to detect, recognize and track the moving object from video sequence ,then it makes a description and comprehension to these behaviors. In real life, with a great deal of meaningful visual information being included in the moving objects, detecting and tracking moving objects has become a hot-spot and an important topic to scholars nowadays.In this paper, the motion detection and tracking algorithm of video objects and the experiments are mainly discussed. First, the paper discusses the background and meaning of the task and makes a brief introduction on some basic knowledge and technology related to our topic.Then it proposes a novel motion detection algorithm based on the combination of improved interframe difference and level set, and a tracking algorithm based on optical flow,interframe difference and level set theory.In the proposed method,we make full use of the optical flow information to control the course of level set evolution successfully.At last, we validate the algorithm through experiments and analyze its performance.The proposed algorithm is emulated on the platform of Visual Studio C++6.0 which runs on the system of Windows Vista, and the experimental data comes from the DynTex database. The experimental results shows that our algorithm is feasible and it enhanced the accuracy of motion detection and improved the tracking speed. Besides, compared with the existed methods, our algorithm reduces the computation and improves the system in real time.