Adaptive Multiple Description Depth Image Coding Based on Wavelet Sub-band Coefficients

Jingyuan Ma, Huihui Bai, Meiqin Liu, Dongxia Chang, Rongrong Ni, Yao Zhao

Abstract

With the development of multi-view video plus depth technology, the coding algorithm at the depth image has become one of hot research directions. As we know, after wavelet transform, the energy of the image smoothing region is concentrated in low frequency sub-band, while the edge information of the texture region is concentrated in high frequency sub-bands. However, the edge information is very important to the synthetic viewpoint. In order to improve the edge decoding quality and ensure the transmission reliability, we propose an adaptive multiple description depth image coding scheme based on wavelet sub-band coefficients. The low frequency sub-band is encoded by optimized multiple description lattice quantization (OMDLVQ), while the high frequency sub-bands are encoded by embedded block coding with dead-zone. Finally, two streams of the vector quantization are combined with the embedded block coding stream respectively. The experimental results show that this scheme has good performance in transmission reliability and reconstructed image quality.
BibTeX
@inproceedings{ma2017adaptive,
  author = {Jingyuan Ma and Huihui Bai and Meiqin Liu and Dongxia Chang and Rongrong Ni and Yao Zhao},
  title = {{Adaptive Multiple Description Depth Image Coding Based on Wavelet Sub-band Coefficients}},
  booktitle = {IIH-MSP},
  year = {2017},
  doi = {10.1007/978-3-319-63859-1\_5},
}

Loading related content...

Figure