An efficient motion re-estimation algorithm for frame-skipping video transcoding

Abstract

In video transcoding, the compressed video bit stream is often converted to the reduced frame rate video bit stream in order to reduce the bit rate. To speed up this operation, a video transcoder usually reuses the decoded motion vectors from the incoming bit stream. Previously, the bilinear interpolation method and forward dominant vector selection method were proposed to compose the missing motion vectors for reuse. In this paper, we propose a new algorithm, called the efficient-forward dominant vector selection, which utilizes the difference between dominant macroblock and motion vector from the incoming ones. We propose a motion vector refinement algorithm called the direction oriented search. The performance of experimental results can be improved while maintaining low computational complexity for a reduced frame rate video transcoder.