High Performance DWT for Hybrid Image Compression System

Uncompressed multimedia data occupy more storage space and very high data rates for transmission. Innovation is always under process to derive new high speed and efficient image compression methodologies. The main aim of this paper is to determine suitable wavelet for image compression using hybrid wavelet multi neural network architecture. A detailed comprehensive experimental analysis is carried out on the proposed hybrid architecture to obtain better fidelity metrics. The process involved applying different hand designed wavelet transforms on the proposed hybrid architecture and computing the objective fidelity metrics by performing necessary changes in different bands of frequency coefficients obtained, every time the wavelet is changed. This process is repeated with various wavelets such as the Haar Wavelet, Daubechies Wavelets, Coiflets, Symlets and the Biorthogonal wavelets etc. Wide range of gray scale and colour images of varying details are considered and Performance metrics obtained are tabulated and analyzed graphically.