Reduced Memory SPIHT Coding Using Wavelet Transform with Post-Processing

Abstract

Traditional wavelet-based image coding applies the discrete wavelet transform (DWT) on an image using filter banks over rings of characteristic zero. As the level of the DWT decomposition increases, the number of bits needed to represent the wavelet coefficients also increases. A significant amount of memory is needed to store these wavelet coefficients especially when the level of DWT decomposition is high. In this paper, a post-processing method is proposed to set the amplitude of the wavelet coefficients to pre-defined N-bits. The set-partitioning in hierarchical trees (SPIHT) coding is then performed to encode these coefficients to achieve compression. The main advantage of our proposed work is the significant reduction in memory requirements for wavelet coefficients storage during bit-plane coding. Simulation results show that our proposed SPIHT coding using wavelet transform with post-processing gives an equally good compression performance when M-3 les N les M-1 where M and N are the number of bits needed to represent the largest wavelet coefficient without and with post-processing respectively.