A new MMSE encoding algorithm for vector quantization

Abstract

An algebraic interpretation of characterization of a quantizer point in terms of its neighbors is presented, along with a new method for finding the neighbors. A novel MMSE (minimum mean squared error) encoding algorithm for vector quantization is proposed. The use of a prequantizer for providing a good starting point to the new algorithm is discussed. Simulation results show that the new algorithm, used in conjunction with Gersho's (1982) method for finding the neighbors, effects a substantial decrease in the total number of operations required for encoding when compared with the fast algorithm proposed by Soleymani and Morgera (1988). The drawback of the new algorithm is the increased storage requirement.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>