METHOD AND APPARATUS FOR VECTOR QUANTIZATION

摘要

A method of compressing an original signal which is converted into a plurality of binary signal vectors. Each signal vector contains a plurality of data elements of the original signal being compressed. The method includes the step of hashing each signal vector to identify one of at least three different subsets collectively constituting one of several different provided sets of code words. Each code word of each set has a unique, associated binary code shorter than the hashed signal vector. The binary code of the one code word of the one identified subset having the least distortion with respect to the hashed signal vector represents the compressed signal vector. By providing plural sets of code words, each set can be optimized for different component data of the composite signal. For example, luminance and chrominance data in a composite color video signal can be compressed with separate sets of code words for less distortion, greater compression or both.