For vector quantization (VQ), it is extremely time consuming to extract the similar codeword with input vector during the encoding process. In this paper, we present an efficient algorithm to extract the features of input vector using Haar wavelet transform (HWT) and use these features to early remove impossible codeword in the distortion computations stage. From the experimental results, it is shown that the proposed approach can largely decrease the computation time for achieving VQ coding with the same quality with full search algorithm. More specifically, compared with the DHSS algorithm, the proposed algorithm reduces the computational time by 31% to 61%. Compared with the Panpsilas algorithm, the proposed algorithm reduces the computational time by 62% to 75%. Compared with the Laipsilas algorithm, the proposed algorithm reduces the computational time by 48% to 58%. Compared with the HTPDE algorithm, the proposed algorithm reduces the computational time by 27% to 44%. Compared with the WTPDE algorithm, the proposed algorithm reduces the computational time by 21% to 45%.