Part 10: Sub-Object Encoder and Decoder

Sub-objects are sorted out according to the SceneProfile associated with an actual service. Each sub-object of arbitrary shape and size is generally represented by rectangular stereoscopic images with the right shape and required spatial and temporal resolution. A mask for the sub-object determines the transparency for each pixel. The motivation for this sub-division is to avoid progressive layering, and obtain equalpriority entities that are independent until viewed on the screen by the user. In the case of progressive coding, basic layers are sent as important AppTraNet packets, while higher layers have decreasing priority. At first sight, progressive coding with successive refinement seems to be an advantage regarding scalability, but in the network where packets are dropped randomly, it is a drawback. Important packets have to be specially protected or prioritized. A flat coding structure is expected to give a simpler and smoother degradation of quality when packets are dropped or lost in the network.