Research On Moving Object Detection Based On Optical Flow Mechanism

Moving object detection is an important branch of computer vision. In recent years, moving object detection technology has been widely used in people's daily life and military domain, especially in the auxiliary driving field, for example intelligent traffic and automatic driving. Separating the moving object quickly and accurately from the background is a basic and key technology for further image analysis and processing. So, a series of moving object detection method are presented through making a thorough study continuously. However, there are still some problems to be resolved, for example the shake. In this paper, we study the moving targets detection algorithms under different conditions; analyze the advantages and disadvantages of every screening algorithm and improve the moving object detection algorithm under the dynamic scene.In this paper, the methods of moving object detection in static and dynamic scene are studied deeply, especially the dynamic scene. We summarize the definition, principle and calculation method in detail, especially we give the detailed expatiation and analysis of lump matching and search algorithm.In the process of video, there are always some shake of the carrier which makes the video output unsteady. How to eliminate the shake is the focus of this paper. In comparison with various moving object detection methods in dynamic scene, optical flow algorithm is the better one. Moreover, block matching algorithm is the better one in the computational methods of optical flow. So after analyzing the advantage and disadvantage of the various shake eliminating methods, the thesis adopts the shake eliminating method based on block matching. Combining the shake eliminating method and the optical flow method, we present a robust optical flow algorithm, which is adaptive to the shake. Differing from the earlier shake compensation, the optical flow image rather than the original image is compensated by the compensation amount. As well as, because the shake eliminating method and the optical flow computational method use the same block matching method, the algorithm can save the computing time efficiently. Experiment results show our approach can eliminate the dithering availably, and separate the moving object out of the image exactly.