作者: Kai-Tat Fung , Yui-Lam Chan , Wan-Chi Siu
关键词:
摘要: Transcoding is a key technique for reducing the bit rate of previously compressed video signal. A high transcoding ratio may result in an unacceptable picture quality when full frame incoming bitstream used. Frame skipping often used as efficient scheme to allocate more bits representative frames, so that acceptable each can be maintained. However, skipped must decompressed completely, which might act reference nonskipped frames reconstruction. The newly quantized discrete cosine transform (DCT) coefficients prediction errors need re-computed with previous frame; this create undesirable complexity well introduce re-encoding errors. In paper, we propose new algorithms and novel architecture frame-rate reduction improve reduce complexity. proposed mainly performed on DCT domain achieve transcoder low With direct addition error compensation feedback loop, are reduced significantly. Furthermore, control dynamically adjust number according motion vectors due such decoded sequence have smooth better transcoded pictures. Experimental results show that, compared conventional transcoder, frame-skipping robust, produces fewer requantization errors, has computational