Video Template Matching Near Picture Boundaries

The patent applies dynamics by making the search area adaptive rather than fixed. The search area is dynamically adjusted based on the position of the current block relative to picture and reconstructed area boundaries. Blocks closer to boundaries utilize expanded search areas that wrap around to opposite sides, while blocks in the center use traditional limited search areas. This dynamic adaptation resolves the contradiction by allowing complex search patterns only where necessary (near boundaries) while maintaining simple search patterns in the majority of the picture area.

The patent applies local quality by implementing different search area strategies for different spatial locations within the picture. Rather than uniformly expanding the search area for all blocks, the patent selectively expands search areas only for blocks located near picture or reconstructed area boundaries where such expansion provides benefit. Blocks in the central region of the picture continue to use the traditional, simpler search area. This localized approach improves coding efficiency where needed while avoiding unnecessary complexity in regions where it would not help.

The patent makes the search area dynamic based on block position. For blocks near picture or reconstructed area boundaries, the search area is dynamically expanded to include blocks from the opposite side of the picture, allowing the algorithm to find better reference blocks that would otherwise be inaccessible. This dynamic adjustment maintains algorithm simplicity for central blocks while improving reference block selection quality for boundary blocks through the wrap-around search capability.