Dynamic bit-rate allocation with fuzzy measures for MPEG transcoding

Abstract

In this paper, we propose a dynamic bit-rate allocation algorithm for MPEG transcoding. The method consists of a bit-rate allocation with fuzzy measures and a least-distortion bit-rate reduction. Fuzzy measures are calculated by the code length, the discontinuity bluntness, and the neighborhood momentum in each DCT block. These measures are summed with weights and form a reduction fuzziness to indicate the degree of plausible reduction. Using the reduction fuzziness, each DCT block is filtered by the least-distortion reduction method to adjust the bit-rate for a target bandwidth. In the experiment, we show the results that the transcoded video quality by the method is better and the bandwidth is more regular than those of existing methods in both visually and quantitatively.