Research and design of real-time human body video acquisition system

The paper presents a Mixed Reality Motion Model (MRMM) based on the definition of Mixed Reality. Then a human body video acquisition system based on MRMM is carried out. Firstly, real-time human body video acquisition is realized by using VFW. The image processing unit processes the video stream frame-by-frame. It turns the current frame to binary image by Frames Difference; the parameters are extracted according to pixels´ information and the characteristics of human body. Secondly, in order to identify the target region directly and improve the extraction quality, mathematical morphology is introduced in dealing with binary image. So the binary information is corrected to be more accurate. In the basis of the binary image, body image could be combined with the virtual picture frame by frame to form a new virtual video, including real person image and virtual background. Finally, median filter method is used to smooth the output video frame. The main function is to reduce the lays of frames. The experiment results are given to illustrate the algorithm effective and real-time.