Video Block Decoding Using Extrapolation Merge Candidates

The patent segments the extrapolation filter derivation process into multiple candidate filters with different area types and shape types. Each candidate filter is derived separately based on specific geometric configurations of the current block and its neighboring blocks. This segmentation allows the system to handle complex derivation processes systematically by breaking them down into manageable parts, each corresponding to a specific filter type.

The patent changes parameters such as the area type (e.g., horizontal, vertical, diagonal) and shape type of the extrapolation filter based on the geometric characteristics of the current block and its neighbors. By dynamically adjusting these parameters, the system can adapt the derivation process to different block configurations, maintaining prediction precision while managing complexity through parameterized approaches.

The patent performs preliminary actions by pre-defining multiple candidate filters with different area and shape types before the actual prediction process. These candidate filters are prepared in advance with their respective derivation rules, allowing the decoder to quickly select and apply the appropriate filter without performing complex real-time calculations. This preliminary preparation reduces decoding time while maintaining accuracy.

The patent uses copying by replicating the derivation process for multiple candidate filters, where each candidate is a copy of the base derivation mechanism adapted to specific geometric conditions. This copying approach allows the system to evaluate multiple prediction possibilities efficiently by reusing the same derivation logic across different filter types, thereby reducing overall computational time.