Performance Analysis of Set Partitioning in Hierarchical Trees (SPIHT) Algorithm for a Family of Wavelets Used in Color Image Compression

With the spurt in the amount of data (Image, video, audio, speech,&text) available on the net, there is a huge demand for memory&bandwidth savings. One has to achieve this, by maintaining the quality&fidelity of the data acceptable to the end user. Wavelet transform is an important and practical tool for data compression. Set partitioning in hierarchal trees (SPIHT) is a widely used compression algorithm for wavelet transformed images. Among all wavelet transform and zero-tree quantization based image compression algorithms SPIHT has become the benchmark state-of-the-art algorithm because it is simple to implement&yields good results. In this paper we present a comparative study of various wavelet families for image compression with SPIHT algorithm. We have conducted experiments with Daubechies, Coiflet, Symlet, Bi-orthogonal, Reverse Bi-orthogonal and Demeyer wavelet types. The resulting image quality is measured objectively, using peak signal-to-noise ratio (PSNR), and subjectively, using perceived image quality (human visual perception, HVP for short). The resulting reduction in the image size is quantified by compression ratio (CR).