Classified Vector Spiht Forwavelet Image Coding

Wavelet based image coding has been shown to be an effective method, especially at low bit rate. Two successful approaches, space-frequency quantization (SFQ) and set partitioning in hierarchical trees (SPIHT) were proposed to predict insignificant wavelet coefficients. However, both of them employ a simple scalar quantization scheme instead of powerful vector quantization for coding of the significant wavelet coefficients. In this paper we propose a new image coding method that combines the SPIHT type set partitioning technique with classified vector quantization. That is, the set partitioning technique is used to locate significant wavelet coefficient vectors based on vectors´ energies. In addition, the same information is used as classification indices to select the correct energy sub-codebook for vector quantization. The main idea of the proposed method is that the SPIHT type information is shared by set partitioning for locating significant vectors and classified vector quantization. In this way, many bits are saved for coding the side information required by classified vector quantization. Experimental results show that our proposed method achieves better performance, as compared to SPIHT and SFQ.