Moving Object Detection And Trajectory Analysis

With the rapid development of network technology and multimedia technology, image information endears itself to the people by its form of vividness and intuition. Compared with static image, more abundant information is provided with video sequences of moving object. Therefore, moving object analysis based on video sequences becomes a hotspot in video research. As a basic link in the field of computer vision, moving object detection and tracking are referred to separating the moving object which is interested from the background, and corresponding matching problems are established based on object's features in the continuous images in order to find the object's position. Nowadays, moving object detection and tracking are widely used in computer vision, intelligent video surveillance, automatic traffic management, human detection and so on.In this paper, we research on the moving object detection and tracking in video sequences captured by a fixed camera. The main tasks are as follows:Firstly, video sequences captured by imaging equipment are pre-processed, including conversions of image color spaces, image de-noising and image enhancing. Through these operations, the image quality is improved. So that the images are easily identified and the accuracy of moving object detection is improved.Secondly, in terms of moving object detection, the traditional methods of moving object detection are analyzed and compared. On this basis, a method which is a combination of frame difference and background subtraction is discussed. Mathematical morphology processing is supplied to the result and the moving object with a clear outline is obtained.Thirdly, a real-time forecasting model is introduced to the process of object tracking, the object's position in the next frame is estimated by quadratic polynomial. Then the object is searched within a certain window which makes the predict value as the center and the object tracking is accomplished. This method reduces the matching complexity and improved the tracking speed.Finally, the object's trajectory is obtained by fitting the object's positions using spline curve. Further, the object's speed is analyzed and a rule of sampling the video sequences is presented to solve the low precision of the moving object's speed.Experimental results show that this method can effectively detect and track rigid object and non-rigid object in video sequences.