A fixed-rate quantizer using block-based entropy-constrained quantization and run-length coding

Abstract

A fast and efficient quantization technique is described. It is fixed-length, robust to bit errors, and compatible with most current compression standards. It is based on entropy-constrained quantization and uses the well-known and efficient Viterbi algorithm to force the coded sequence to be fixed-rate. Run-length coding techniques are used to improve the performance at low encoding rates. Simulation results show that it can achieve performance comparable to that of Huffman coded entropy-constrained scalar quantization with computational complexity increasing only linearly in block length.