摘要 |
A method of reducing the memory required for decompression of a compressed frame by storing frames in a compressed format using DCT compression and decoders for implementing such a method are disclosed. The decoder is coupled to a memory where the frame can be stored. The decoder includes a decoder module having a parser, a block decoder module and a motion compensation engine. The decoder module is coupled to a DCT encoder module, which has an output coupled to the memory. The decoder also includes a stored DCT decoder module, which has an input coupled to the memory, a first output coupled to the motion compensation module and a second output that functions as an output of the decoder. In operation, any prediction frames needed for motion compensation decompression of the compressed frame are decompressed in the stored DCT decoder module. The compressed frame is decompressed in the decoder module to obtain a decompressed frame. The decompressed frame is compressed in the DCT encoder module to obtain a recompressed frame. The recompressed frame is then stored in memory. In a DCT based decoder, this is only performed for frames having interpicture prediction errors, such as P and B frames. Frames that do not have interpicture prediction errors, such as I frames can by-pass most of the decoder module, and all of the DCT encoder module. The compressed frame can be stored in the memory without having been decompressed and recompressed.
|