Multiple Description Image Coding Using Dual-Tree Discrete Wavelet Transform

A novel multiple description image coding scheme, called dual-tree discrete wavelet transform based multiple description image coding is proposed. First, the 2D dual-tree discrete wavelet transform is applied to an image to produce 2 wavelet trees with complementary information. A noise-shaping algorithm is then employed to remove the redundancies of both trees and form the basic parts of two descriptions. A side noise shaping algorithm is proposed to draw the supplementary information for side decoding from the other tree. Basic information from one tree and supplementary from the other tree are grouped to form a description. Two descriptions are then dispatched from different channels. Because every description includes critical information of the picture, the reconstructed image from any single channel can have acceptable fidelity. The reconstructed image is of much better quality if both channels are available. Simulation results have demonstrated the performance of the proposed algorithm.