Embedded denoising for the H.264/AVC extension-spatial SVC

Recently, many schemes that embed the denoising process into the video encoding have been proposed. They are mainly implemented into the single-layer encoder, like the H.264 reference encoder joint model (JM). In this paper, we propose to embed the denoising process into the multi-layer spatial SVC encoder. Since either for the base layer or the enhancement layer our proposed filter could achieve linear minimum mean square error (LMMSE) and meanwhile the additional inter-layer prediction tools could be utilized, the encoder performance can be largely improved. The experimental results show that the multi-layer embedded approach can remove most of the noise while achieving spatial scalability. And the R-D curve shows the optional use of inter-layer prediction tools can help yield better denoising performance and coding efficiency.