Multiple Description Transform Coding Comparison for Multiresolution Image Representation

Compression is widely used in image processing to improve efficiency transmission over erasure networks, so the design of the compression algorithm depends on the performance of transform. Many approaches have been applied, such that Discret Cosine Transform (DCT) adopted by JPEG standard and Discret Wavelet Transform (DWT) leaded to JPEG2000 which has remarkable success in still signal and image processing, but it fails to capture discontinuities to represent edges directionality, therefore new schemes have been developed to compensate this limitation, such as Contourlet Transform (CT) and it’s over-complete version Nonsubsampled Contourlet Transform (NSCT) to overcome the lacked shift invariant property in CT. In this paper, in the suggested scheme of multiple description transform coding (MDTC), experimental results demonstrate that our method leads to perform image reconstruction for practical MDC design in term of peak signal to noise ratio (PSNR) and visual quality assessment.