Vector quantization fast search algorithm using hyperplane based k-dimensional multi-node search tree

A vector quantization fast search algorithm using a hyperplane based k-dimensional multi-node search tree is presented. The misclassification problem associated with hyperplane decision is eliminated by a multi-level backtracking algorithm. The vector quantization complexity is further lowered by a novel relative distance quantization rule. Triangular inequality is applied to lower bound the search distance, thus eliminating all the sub-trees in the k-dimensional search tree during backtracking. Vector quantization image coding results are presented which show the proposed vector quantization algorithm outperforms other vector quantization algorithms in the literature both in PSNR and computation time.