Intra Block Copy Candidate Lists for Low-Overhead Coding

The patent segments the candidate block vectors into multiple categories (spatial neighbors, temporal neighbors, and history-based candidates) and organizes them in a structured candidate list. This segmentation allows the decoder to efficiently search through organized groups rather than unstructured candidates, improving compression efficiency while maintaining manageable complexity through systematic organization.

The patent performs preliminary actions by pre-constructing the block vector candidate list with multiple types of candidates before the actual decoding process. The candidate list is prepared in advance with spatial, temporal, and history-based block vectors, allowing the decoder to quickly select from pre-organized options rather than generating candidates on-the-fly, thus improving compression efficiency without proportionally increasing complexity.

The patent applies partial action by signaling only the necessary information to identify the selected candidate from the candidate list, rather than transmitting complete block vector data for all candidates. The decoder reconstructs the full candidate list locally using the same rules as the encoder, and only the index of the selected candidate needs to be signaled, achieving good compression efficiency with minimal signaling overhead.

The patent enables self-service by allowing the decoder to autonomously reconstruct the block vector candidate list using the same generation rules as the encoder. The decoder uses locally available reference blocks and history information to build its own candidate list, eliminating the need for the encoder to signal detailed candidate information, thus reducing signaling overhead while maintaining compression efficiency.