3D-TV: Coding of Disocclusions for 2D+Depth Representation of Multi-View Images

C. Vázquez and W.J. Tam (Canada)


3D-TV, Multi-view coding, disocclusions, 2D+Depth


The 2D+depth (2D+D) format has recently emerged as a good candidate for the transmission of multi-view content in a 3D-TV broadcasting environment because it allows the rendering of new views with minimum processing require ments. Its main drawback is the appearance of disoccluded areas in the rendered views. We explore a way to encode these disoccluded areas as an enhancement layer in a 2D+D representation. We exploit the fact that the disocclusion is generally an extension to the farther side in a depth edge. We used an interpolating wavelet decomposition of the hor izontal line formed by embedding the disoccluded area be tween the pixels defining the depth edge. This strategy pre serves the value of pixels in the transmitted view. Our re sults show that the proposed method reduces the amount of information needed if compared to the simultaneous trans mission of all views while improving the quality of the new views rendered from the 2D+D representation.

Important Links:

Go Back