A fast VLSI chip for computing the two-dimensional discrete cosine transform

A VLSI chip which performs an 8-point discrete cosine transform (DCT) using distributed arithmetic and Li's (1991) algorithm has been successfully designed and tested. Its total throughput is as high as 250 Mb/s, which is of the order required for real-time applications such as the compression of video. Both the size and performance of the chip can be substantially improved through changes in its technology and design. Work on a 2-D DCT processor based on this unit is in progress.<
>