Research Of Motion Object Detection Technology Based On MPEG-4

Moving object detection and tracking is one of active branches of imaging processing and computer vision, which study how to extract moving object region, track the object, and estimate the concerned moving parameter in serial video image. This technology has been applied widely in military application and video detection. This thesis does some research in moving object detection as follow:Focusing on the problem of moving object detection and tracking on MPEG-4 compressed streams captured by a stationary camera, an algorithm to detect the moving object using DC coefficients and motion vectors of the motion information in MPEG is presented. Much computation of DC coefficients is avoided through the method which reduces computational complexity. Testing results show that minimal decoding on compressed bitstream is performed by the method and the tracking speed can be dramatically improved; the computing time decreases much.When moving object detection under moving camera,GME(Global Motion Estimation)is an important tool widely used in moving object detection, and other fields. In this thesis, we propose a new method for the estimation of global motion from compressed image sequences. With regard to global motion models, we adopt self-adaptive with scalability global motion model. In order to improve accuracy and computational efficiency of global motion estimation, we adopt a non-iterative algorithm for extracting model parameters. Then, global motion model which match the camera best associated with the result are selected to estimate motion model parameters. Lastly, according to the estimated error, some sample pairs may be rejected as outliers and the parameters may be refined by estimating from the remaining data. Furthermore, when using this method, there's no need to decompress the compressed MPEG video sequence. So the speed of processing is very high, and it meets real-time applications.