| In this thesis, a new algorithm called Adaptive Set Partitioning Wavelet Vector Quantization (ASPWVQ) is presented. ASPWVQ applies the set partitioning concept in the context of our previously developed coding method based upon wavelet vector quantization (VQ) and rate-distortion optimization, which is called WRDADRVQ. By adopting the set partitioning idea, ASPWVQ can use fewer bits to code the ordering information of the VQs. Moreover, a new set of codebooks is designed based upon average energy of wavelet decomposition coefficients. The performance of ASPWVQ and energy codebooks is compared with WRDADRVQ and non-energy-based codebooks by five experiments. Ten commonly used test images are used in each experiment to generate the average experiment results. These results show that ASPWVQ achieves better performance than WRDADRVQ at a compression ratio less than 4:1, and the energy codebooks perform remarkably better than non-energy-based codebooks in the complete compression ratio region. |
