A VLSI chip set for high-speed lossless data compression

A VLSI implementation of a lossless data compression algorithm is reported. This is the first implementation of an encoder/decoder chip set that uses the Rice (see JPL Publication 91-1, 1991) algorithm and provides an introduction to the algorithm and a description of the high-performance hardware. The algorithm is adaptive over a aide entropy range. Its performance on several 8-b test images exceeds other techniques employing differential pulse code modulation (DPCM) followed by arithmetic coding, adaptive Huffman coding, and a Lempel-Ziv-Welch (LZW) algorithm. A major feature of the algorithm is that it requires no look-up tables or external RAM. There are only 71000 transistors required to implement the encoder and decoder. Each chip was fabricated in a 1.0- mu m CMOS process and both are only 5 mm on a side. A comparison is made with other hardware realizations. Under laboratory conditions, the encoder compresses at a rate in excess of 50 Msamples/s and the decoder operates at 25 Msamples/s. The current implementation processes quantized data from 4 to 14 b/sample.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>