Lossy Lempel-Ziv algorithm for large alphabet sources and applications to image compression

Abstract

Two practical lossy Lempel-Ziv schemes are proposed in this paper. The schemes are suitable for compression of sources with large alphabet cardinality. Computer simulation results obtained by compressing the output of a Gaussian source revealed a performance comparable to the performance of scalar quantization followed by entropy coding. Compression of DCT transformed and subband decomposed images showed a performance close to the performance of JPEG compressed images.