Adaptive vector quantization of wavelet coefficient for image compression

To enhance the performance of nonlinear interpolative vector quantization for wavelet transform image coding, we propose a directional vector classification scheme that results in an adaptive interpolative vector quantization method suitable for wavelet-based image compression. Vectors consisting of wavelet-transformed coefficients are classified into different categories according to their directional activity energy. An adaptive bit assignment algorithm effectively allocates the bit budget among different categories. Simulation results show that the new vector quantization method gives superior compression performance and also reduced computational complexity in codebook training and quantization compared to the conventional interpolative vector quantization technique. For standard test images, the PSNR improvement is about 1 dB at the compression ratio of 50.