An Improved Codeword Search Algorithm Based on Subvector Features

An improved fast codeword search algorithm based on sub vector features is presented in this paper. First divide every input vector into two sub vectors with the same dimension. Then compute the sum and variance of every input vector and its two sub vectors in turn, get a series of space division inequalities which are used as the elimination criterions. The result of the simulation shows that in the same coding quality compared with full search algorithm, the proposed algorithm has better data in the average number of distance calculations compared with several classic algorithms.