Influence of block size on motion vector estimation error in enhancement of video temporal resolution

摘要

Modern video recording devices record high resolution video material, and for efficient storage and transmission the video file size should be reduced by encoding. Depending on the content, decreasing the temporal or spatial resolution of a video signal before the encoding and increasing after the decoding can ensure higher sequence quality than compression of the material with original resolution. Before displaying the video material with the decreased resolution on a modern high resolution displays the resolution of the transmitted video signal should be enhanced in order to satisfy the users' perceived quality. Temporal resolution can be increased by interpolating of new frames between the existing frames in the downscaled sequence. The best quality of the interpolated frames can be achieved by using motion compensated frame interpolation (MCFI). The first step in MCFI is motion estimation (ME). In ME process the existing frames are divided into blocks, and motion vectors (MV) are calculated for each block. These MV are base for the estimation of the motion vectors between existing and frames that will be interpolated. The influence of the block size in the ME process on the estimated motion vector error is analyzed in this paper. The results show that in the interpolation process MV estimation error rises as the block dimensions decrease.