A new frame-recompression algorithm and its hardware design for MPEG-2 video decoders

An algorithm and its implementation to recompress video frame data, which is to be stored into external memory, are proposed, thereby enabling reduction of memory requirements and bandwidth. To support random access capability, frames are organized as 1 * 8-pixel arrays, which are then compressed into 32-bit segments. The compression ratio is fixed at 50%. For both compression efficiency and low complexity, modified Hadamard transform and Golomb-Rice (GR) coding are utilized. Objective and subjective video quality assessments are performed. Efficient hardware designs including GR encoding/decoding and packing/unpacking are also proposed to reduce the total compression/decompression latency cycles. The proposed hardware designs are suitable for the integration with MPEG-2 decoders thanks to fast operation and simple interface logic.