Selection of Lifting Scheme based Wavelet Filters for Image Compression in Resource Constrained Wireless Multimedia Sensor Networks

Abstract

Sensor nodes in wireless multimedia sensor networks have very limited resources like low memory capacity, low processing capability, and low data rate. Most of the applications in WMSN require image data which is naturally large in size and it has to be compressed before transmission in WMSN with very small bandwidth. Lifting scheme for computing discrete wavelet transform and SPIHT image coding algorithm are suitable to implement in wireless sensor nodes which have some limitations in power consumption, processing capability, memory and bandwidth. This paper addresses selection of suitable wavelet filter type for image compression using SPIHT coding algorithm. In this paper we have implemented SPIHT algorithm based on the lifting scheme of different wavelets (Haar, Daubechies 4 (DB4), Daubechies 6 (DB6), Cohen-Daubechies-Feveau wavelet (CDF 9/7), and Cubic B-splines) in MATLAB. The experimental subjective and objective results show the lifting CDF 9/7 wavelet filter gives excellent results for bit rates 0.1 bpp, 0. 2 bpp, 0.5 bpp, 0. 8 bpp, and 1 bpp. CDF 9/7 wavelet filter gives an acceptable image quality even for very low bit rates which makes it peferrable for implementation in low data rate applications of WMSN.