Scalable Video Encoding Motion Information Mapping

The patent extracts and removes unnecessary motion information mapping operations by identifying that motion information should only be mapped when layers actually refer to each other. The solution separates essential mapping operations from redundant ones by checking layer reference relationships before performing mapping, thereby reducing computational complexity while preserving prediction accuracy for necessary mappings.

The patent changes the parameter of motion information mapping by introducing conditional logic that modifies the mapping behavior based on layer reference relationships. When layers do not refer to each other, the mapping parameter is effectively disabled or set to identity, avoiding unnecessary computational operations while maintaining correct prediction when mapping is needed.

The patent introduces an intermediary mechanism (layer reference relationship checking) that mediates between collocated picture usage and layer compatibility. This intermediary layer verifies whether layers actually refer to each other before allowing collocated picture usage, ensuring that prediction capability is enhanced only when appropriate, thereby preventing encoding efficiency degradation from incorrect picture selection.

The patent performs preliminary verification of layer reference relationships before utilizing collocated pictures for prediction. By checking compatibility conditions in advance, the system prevents incorrect picture selection that would degrade encoding efficiency, while still allowing enhanced prediction capability when the preliminary check confirms appropriateness.