Moving Objects Detection And Tracking Based On Revised Optical Flow

Moving objects recognition and tracking is a quite hot topic in computer vision, and has a wide variety of application, such as video conference, navigation of robotics, virtual reality etc. There are two basic analysis approaches: the feature-based approach and the optical flow based approach. We adopt optical flow approach in this paper.Firstly, in order to eliminate image noise, separate the object from background and extract edge of the object, a series of pre-processing are carried out to the images, such as image smoothing, threshold segmenting and edge tracking. This makes preparations for the computation of optical flow field.Secondly, the gradient-based approach and the match-based approach are discussed in detect the moving object. The revised approach of optical flow is adopted, and simulated the Horn&Schunck algorithm, Lucas&Kanade algorithm and feature points matching algorithm.The experiment shows that Lucas & Kanade algorithm is better than the Horn&Schunck algorithm, and it also shows that revised approach of optical flow with accuracy.Thirdly, the tracking of mass center and Kalman filter algorithm is carried out in the tracking of moving object.The experiment shows that the tracking of mass center is simple, but not so accuracy. The revised function of Kalman filter algorithm can improve the accuracy.