Given a finite subset W and a vector x both in k-dimension space

the method to fast locate the vector in W closest to x is of great value in applications. The fast search method plays an important role in the fields such as vector quantization method and pattern recognition using neural network. Based on the analysis of the fast search algorithm in Ref.[6]

a new fast search algorithm is presented. Using the codebook addresses of neighboring image blocks as the predicting codebook address of current image block

the search space can be reduced more dramatically.