| With the high speed development of computer video and telecommunicationtechnology, Traditional2D TV just can’t meet people’s needs of high-quality videobecause of its disability of displaying the real scene, finally3D TV came into being.Meanwhile, it was approved as the “Twelfth Five-Year Plan” important project.3DTV Project consist of3D video source, delivery and3D video-display related tech,finally making3D video decide the destiny of3D TV’s future because of the maturityof telecommunication and3D display technology.Currently there are two method of making high quality3D video. Firstly3Dvideo can be made directly by two cameras with the same parameters, but the cost isvery high. Secondly superb2D video can be made into3D video via adapting properalgorithm, and this method, which has a significant contribution to3D TV because ofits low cost and abundance, is highly worthy of research.The dissertation has improved the already exist algorithms, and then realize theimproved algorithm with software coding. At the beginning, it timely tracks the objectin the video sequences based on the Camshift and Kalman filter algorithms, and it canaccurately locate the object in the video, so it’s very good for the segmentation andgood depth video can be produced. And then a prior hypothesis of depth gradient andblock-based grouping algorithm are used to assign the depth of depth video sequencesproduced by the Grabcut algorithm, and then the right2D video can be made with theoriginal left2D video and the related depth video based on the DIBR (Depth ImageBased Rendering) Algorithm, finally the left and right2D video can be fused into the3D video, and the dissertation applies the color distribution technology to display the3D video and show the experiment result. |
PDF Full Text Request