Abstract: In order to improve the compression ratio of digital image, an new image compression algorithm was proposed. In the algorithm, wavelet transform was combined with vector quantization. Firstly, the relativity of different scales of the wavelet coefficients was used to decompose the wavelet. Then, the classified vector quantization was used in the different scale of the wavelet subbands and different sizes of subcode book were used in different classes. In order to improve the efficiency in high dimensions, the non-linear interpolated vector quantization (NLIVQ) was used to reduce the number of dimensions of the code vector. Experimental results showed that the compression ratio of the image and the efficiency of the algorithm were improved in this new algorithm. In the meantime, the time complexity of the algorithm was discussed.