Skip to Main Content
In order to allow users to use handheld devices accessing video information, such as downloading and playing video files, there is a need to downscale the compressed video into lower spatial resolution and lower transmission bit rate. In this work, transcoding the compressed H.263 video into low spatial-resolution is discussed and realized. To reduce the computation cost, motion vectors from the incoming video stream are resampled and reused. We propose a novel approach to refine motion vectors adaptively according to the motion of every frame or every macroblock in a frame. The proposed approach can improve the video quality and reduce predictive residues of every frame, hence reduce the transmission bit rate. Implementation results suggest that the proposed approach produces better image quality and lower transmission bit rate than a number of previous approaches.