Figure 2
From: Depth-image-based rendering with spatial and temporal texture synthesis for 3DTV
Flow diagram of the proposed virtual view synthesis method. First, the sprite of stationary scene is initialized with the first image frame of the input data. Next, each frame is compared with the sprite of stationary scene to extract more structure information. Then both the current frame and sprite of stationary scene are rendered to the same virtual view and merged together. Finally, the remaining disoccluded areas are filled by proposed oriented exemplar-based inpainting.