DCT-domain translation for composing better pictures in the MDSQ-based multiple description video coding

Multiple description coding (MDC) is an effective technique for transmission of image and video signals over error-prone channels. Naturally, a good MDC system needs to guarantee a quality gain when more descriptions are received. To this goal, a transform-domain overlapping technique can be used in the popular multiple description scalar quantization (MDSQ) scheme. However, this overlapping technique often becomes inapplicable in multiple description video coding (MDVC) because different reference frames are used in different descriptions at the motion-compensation stage. In this paper, we introduce a DCT-domain translation to solve this problem, i.e., after the translation, the overlapping technique can be applied to all predictive frames. We demonstrate that frames with a better quality can be composed upon receiving more descriptions. In particular, a remarkable objective quality gain (in terms of PSNR) has been observed.