HMVP Table Update via Parallel Child Block Processing

The patent segments the video unit into multiple child blocks and processes them in parallel using independent processing paths. Each child block's motion information is derived independently without relying on others, allowing selective updating of HMVP table while maintaining overall prediction accuracy.

The patent performs preliminary derivation of motion information for each child block independently before updating the HMVP table. This preliminary processing allows the system to prepare motion candidates in advance and selectively update the table with only relevant information, reducing unnecessary processing complexity.

The patent divides the video unit into multiple child blocks that can be processed simultaneously in parallel. This segmentation enables independent derivation of motion information for each block, eliminating sequential processing delays while maintaining comprehensive motion prediction accuracy.

The patent maintains continuous processing by allowing multiple child blocks to be processed simultaneously without waiting for previous blocks to complete. This continuous parallel processing reduces overall processing time while ensuring all motion information is captured for accurate prediction.

The patent segments the processing into independent child block units that can execute in parallel without interdependencies. Each segment processes its own motion information independently, maximizing parallel processing efficiency while the collected results collectively improve overall motion prediction accuracy.

Each child block performs self-service by deriving its own motion information independently without requiring information from other child blocks during the derivation process. This self-contained approach enables true parallel processing while the aggregated motion information from all blocks improves prediction accuracy.