Modified Listless Set Partitioning in Hierarchical Trees (MLS) for Memory Constrained Image Coding Applications

Abstract

SPIHT image coding algorithm is a very effective state-of-art technique for compression of wavelet transformed images. However, the use of three continuously growing linked lists, limit its applications to achieve high quality images in memory constrained environments. The No List SPIHT (NLS) is an attempt to implement SPIHT algorithm without lists, but it uses 4 bits per coefficient state markers to keep track of significant/insignificant information of pixels/sets. This paper presents a listless implementation of SPIHT algorithm that uses state markers with only 2.5 bits/coefficients(on an average) to keep track of pixels, set and subsets information. Simulation results show that the proposed modified listless SPIHT (MLS) coder has nearly the same coding performance as that of SPIHT and NLS, but it requires significantly smaller memory compared to these coders. 

Keywords: Image coding, wavelet, SPIHT, NLS, MLS, memory efficient image coder