Frame Rate Up-conversion Based On Image Occlusion Analysis

With the rapid development of HDTV and advanced multimedia system, current video source can't satisfy human's visual requirements, so how to enhance the temporal resolution of video sequence has become a popular subject. Under this background, frame rate up-conversion (FRUC) came out. FRUC is also called temporal interpolation, mainly applied on frame rate conversion of advanced multimedia system, code rate recovery of low bit rate video system and lost frame reconstruction of wireless video communication system.Most of FRUC use motion compensation method. According to the source of motion vector (MV), FRUC is divided into two categories. The first category which MV comes from motion estimation between previous frame and current frame is used for raw video format, like YUV. The second category which MV comes from encoder or transmitter is used for coded video format, like H.264. Our research belongs to the first category.This thesis proposes a FRUC algorithm based on image occlusion analysis. The whole algorithm contains seven steps: bilateral motion estimation, boundary macro-block MV modification, adaptive MV smooth, occlusion analysis, best segmentation for occlusion macro-block, adaptive bilateral weighted interpolation and static pixel modification.Experimental results show that the proposed algorithm provides better image quality than the existed algorithm using similar routine both objectively and subjectively.