Study On 2D To 3D Video Conversion Technology Based On Motion Information

3D expression is an important form of the future information. With the rapid development of 3D video, in the stereoscopic display market, there are obvious contradictions between the rapid popularization of stereoscopic display devices and the lack of stereoscopic sources. To solve this contradiction, on the one hand,3D media resources can be obtained from the research of 3D capture equipment, such as stereo cameras. On the other hand, a large number of the existing 2D media resources can be converted into 3D media resources. If the solution of 2D to 3D video conversion is given, not only can solve the scarcity of 3D media resource, but also avoid the waste of existing 2D media resources.In the basis of looking up a large number of literatures home and abroad, this paper establishes the 2D to 3D video conversion scheme. The method of obtaining motion information like optical flow and block matching etc is introduced. The process of generating depth map by using of motion vector extracted from variable macro block matching motion estimation is explained.In the process of depth map generation, through the analysis of motion vector field, background motion vector is obtained. This paper proposes dynamical background adjusting method based on motion vector field analysis. This method eliminates the background region error depth information. The depth value of object becomes more accurate.After generation of macro block level depth value, through the analysis of relation of high-pass frame coefficient after motion compensation temporal filter(MCTF) and pixel matching, the depth adjusting algorithm based on interframe pixel matching is proposed. The algorithm corrects the macro block level depth map. The pixel level depth map is obtained through distinction between foreground and background pixels in object edge block. Then morphological dilation and erosion algorithm is adopted to eliminate the discrete noise points and hole. Final depth map is acquired. Through experiment, this method can effectively improve the depth map precision.In order to adapt to the 3D video signal transmission in heterogeneous network environment, the method of depth compensation temporal filter (DCTF) based on wavelet to achieve 3D scalable video coding is proposed. Aiming at the feature of 2D+depth schema 3D video, the implementation of 3D video with temporal, spatial and quality scalable is built up and validated by experiment.