Performance evaluation of a JPEG-based multiple description image coder

Abstract

In this paper we design and evaluate the performance of a JPEG-based multiple description image coder. We add zeros to DCT transformed of the original image to create an over-sampled image, and split the over-sampled image into two partitions by separating image blocks. For each partition we use a baseline JPEG coder. In order to evaluate the system performance, we assume that encoded bit-streams passes through two noisy channels, and we calculate the PSNR of reconstructed images in different situations. Results show that the quality of reconstructed images in highly noisy channel improves significantly with a small increase in the transmission bit-rate.