FGS is the video coding scheme that can control the picture quality with high scalability, but FGS has lower efficiency than the single layer video coding scheme because of hierarchical structure. In this paper, we propose three techniques for efficiency improvement of FGS. First, significant bit classification is introduced into the bit plane coding. Second, motion compensation prediction is introduced not only to base layer but also to enhancement layer. This technique can raise prediction accuracy and cut down the amount of prediction errors. However, 2nd technique causes occurrence of drifting errors at low transmission rate. Therefore, third, reference picture is complemented from both base layer and enhancement layer to reduce the propagation of drifting error. Proposed techniques show about 1dB better PSNR performance than the conventional FGS.
View full abstract