Reduced memory zerotree coding algorithm for hardware implementation

Abstract

This paper presents a color image zerotree coding algorithm with a significantly reduced memory requirement, thus making it hardware implementation friendly. The memory reduction was achieved by using the lifted wavelet transform and listless zerotree coding (LZC). The lifted wavelet transform has a lower memory requirement and lower implementation complexity than the traditional filter bank implementation. Unlike other zerotree coding algorithms such as Said and Pearlman's (1996) SPIHT our LZC algorithm uses no lists during encoding and decoding. Without the lists, the memory requirement for the VLSI implementation is reduced significantly. PSNR values for LZC are only slightly lower than SPIHT's. Our approach illustrates well the trade-off between memory requirement and compressed image quality.