In scalable video coding, inter-layer prediction is used to eliminate the redundancies among layers. Joint Scalable Video Model (JSVM) uses a fixed interpolation scheme to interpolate motion and residual information of the base layer into the enhancement layer. The filter used for luma and chroma information is also the same. In this paper, an adaptive interpolation scheme for inter-layer prediction in scalable video coding is presented. Firstly, the interpolation filter used for luma information is adaptively selected among three filters: bilinear, bicubic, ant 6-tap filter. The selection of interpolation filter is based on the quantization parameter. Secondly, bilinear filter is used for chroma information instead of the 6-tap filter used in JSVM. Experimental results show that, compared with the fixed interpolation scheme used in JSVM, the proposed scheme achieves satisfied results in both subjective and objective quality.
[1]
Itu-T.
Video coding for low bitrate communication
,
1996
.
[2]
Wesley De Neve,et al.
Comparison of prediction schemes with motion information reuse for low complexity spatial scalability
,
2005
.
[3]
H. Schwarz,et al.
Overview of the Scalable H.264/MPEG4-AVC Extension
,
2006,
2006 International Conference on Image Processing.
[4]
Jens-Rainer Ohm,et al.
Advances in Scalable Video Coding
,
2005,
Proceedings of the IEEE.
[5]
Wesley De Neve,et al.
Adaptive Residual Interpolation: a Tool for Efficient Spatial Scalability in Digital Video Coding
,
2006,
IPCV.
[6]
Yuzhuo Zhong,et al.
Improved fine granular scalable coding with interlayer prediction
,
2002,
Proceedings DCC 2002. Data Compression Conference.