Multiple description image coding using contourlet transform

Abstract

In our work, we present multiple description coding (MDC) scheme using contourlet transform (CT) for greyscale images. The principle is to prove performance of our MDTC algorithm using CT in contrast to wavelet, which is widely used for image coding. The main goal of our approach is to encode directional sub-bands resulting from the directional stage decomposition to form the four descriptions compressed and transmitted over different packet loss channels. Most of multiple description scenarios are implemented in transform domain. A controlled amount of correlation is appended to the formed descriptions using a pairwise correlating transform (PCT). The experimental results of CT compared to discrete wavelet transform (DWT) show that our method achieves better performance in term of delivered peak signal to noise ratio (PSNR) and visual image quality reconstruction.