Architectures of incorporating MPEG-4 AVC into three-dimensional wavelet video coding

  • Xiangyang Ji ,
  • Jizheng Xu ,
  • Debin Zhao ,
  • Feng Wu

3D wavelet video coding provides an elegant fulldimensional solution for scalable video coding. However, due to the Motion-Compensated Temporal Filtering (MCTF) with open-loop structure and no good rate distortion optimization trade-off between the texture coding and overhead coding of the corresponding mode and motion vectors within narrow bandwidth, its performance is unsatisfactory at low bit-rates. On the other hand, video data below a certain spatial, temporal or SNR resolution are not required in most applications of scalable video coding. The introduction of a base layer in three-dimensional scalable video coding scheme can provide a lot of advantages such as close-loop temporal prediction, in-loop filter and rate-distortion optimization without any constraint occurred in pure 3D wavelet codec and also compatibility with existing video compression standards. This paper presents the architectures about how to efficiently incorporate base layer codec MPEG-4 AVC into 3D wavelet video coding.