Moving Object Detection and Trajectory Analysis
|School||Dalian University of Technology|
|Course||Signal and Information Processing|
|Keywords||Video sequence Motion Detection Tracking Track fitting Speed ??extract|
As networks and the rapid development of multimedia technology, the image information with its intuitive, vivid manifestation of much popular. Compared to still images, moving target video image sequence provides richer information. Therefore, based on analysis of the video image of the moving target to become a research hotspot in the video. As a basic link field of computer vision, moving target detection and tracking of moving objects will be people interested in the background from an image sequence separated and established between successive images of moving objects based on characteristics of the corresponding match, find moving targets appear. Currently, moving target detection and tracking techniques in computer vision, intelligent video surveillance, traffic management automatic monitoring, detecting and tracking the human body has a very popular application. In this paper, fixed cameras capture a video image sequence moving target detection and tracking research, the main work is as follows: First, the imaging device capture video sequences were preprocessing, including image color space conversion, image denoising and image enhancement . Through these operations, improve image quality, making the image much easier to identify, enhance moving target detection accuracy. Second, moving target detection, analysis and comparison of the traditional method of moving target detection. On this basis, a kind of the frame difference method and background subtraction Combination of moving target detection algorithm, and the results combined with mathematical morphology, to give a clear outline of the moving target. Third, the introduction of real-time forecasting model to the target tracking process, using a quadratic polynomial trajectory predicting target in the next frame position coordinates of the image, and then to predict the location of the center within a certain window to search for a matching goal, achieve target tracking. This method reduces the data to be processed matching process, improving the target tracking speed. Finally, the use of spline curve fitting position of the target point coordinate data obtained moving object in a video image sequence trajectory. Further analysis of the target velocity, presents a sequence of video images of the selected sampling interval rule extraction target speed to solve the problem of low accuracy. This algorithm for video image sequence rigid and non-rigid object detection and tracking have received a good experimental results demonstrate the effectiveness of the proposed method.