Study On Correlative Technology Of 3D Conversion From 2D Video Based On Depth Image

3D video display is nowadays often seen as one of the most important representation of future information. Experts and general public regards 3D video as the next major step in video technologies. Compared with traditional 2D video, 3D video could provide more impressive perception of multimedia contents to customers. Not only the higher qulity of video, but also having an immersed sense make 3D video more and more popular. With the advantages which 2D video does not possess, 3D video has a broad application prospect, such as video display, medicine image reconstruction, virtual reality, computer animation, video game, etc. Thus 3D video generation technology emerges as the times require. For example, several special cameras have been designed to generate 3D model directly, which is the major source of 3D video. However, producing such video sequence is high cost.Another way is to convert current and past 2D video data into 3D format. Obviously, the 2D to 3D video conversion method is a convenient and economical way, which is the point which we are focusing on. Thus, in this paper we make an intensive study of the conversion between 2D and 3D video and related techniques based on the theory of binocular stereo vision. At the beginning, the dissertation designs the structure of conversion system. In this system motion estimation is performed between the frames of 2D video, in which this paper introduces the parameter of mutual information, proposes an adaptive motion estimation alorgrithm, which is appropriate to recover video depth information. This algorothm could reduce the computing complexity and improve depth map estimation performace at the same time. Then depth information is extracted from the motion vector. Binocular disparity is calculated according to the relationship between depth information and binocular disparity. Later left and right eye's views could be derived. Presenting these two slightly different views of every scene to the individual eyes with an appropriate disparity, a correct 3D perception can be realized within the mind.In one word, conversion technology between 2D video and 3D video is a new and difficult research topic. In this paper, a conversion system has been proposed which could complement the transformation from 2D video to 3D video. At the same time, it has been implemented on software platform. Although the performance of the current system is not quite ideal. The system has already met the original request of designing, and the visual results show that the proposed system is effeetively in 3D video generation.After the software implementation has matured, it is planed to be transplanted to hardware platform to provide it real practical application value. By the way, its great potential value on 3D video and relative application fields could be explored.