Layered multiple description video coding

In this paper, a two description layered video coding scheme is proposed to provide robust video transmission over unreliable networks based on the previous multiple description image coding framework, layered multiple description coding. A wavelet transformed frame is first decomposed to two subframes. These two subframes are then separately performed motion estimation using low-band-shift (LBS) technique in wavelet domain. The predicted errors of each subframe and related motion vectors are independently coded to form one description. Both descriptions are transmitted from different channels. The advantage of this scheme is in its mismatch free and easy control of redundancy. Experimental results have proved its error resilient ability and coding efficiency.