Video Signal Decoding With Adaptive Merge Candidate Lists

The coding block is divided into multiple sub-blocks, and a second transform is selectively applied to specific sub-blocks based on threshold criteria. This segmentation allows the patent to apply enhanced compression only where needed (when residual signal energy exceeds the threshold) while avoiding unnecessary processing elsewhere, thus improving overall compression performance without proportionally increasing data volume.

The patent applies different processing strategies to different regions of the coding block. By evaluating the residual signal energy of each sub-block and applying the second transform only to those exceeding a threshold, the patent creates locally optimized compression. This local quality approach ensures that regions requiring enhanced compression receive it, while other regions use standard processing, balancing compression performance with data efficiency.

Instead of applying the second transform to all sub-blocks (excessive action) or no sub-blocks (partial action), the patent selectively applies it only to sub-blocks where the residual signal energy exceeds a predetermined threshold. This partial action approach optimizes the balance between compression performance improvement and encoding complexity by performing the additional transform only where it provides meaningful benefit.

The patent introduces a threshold parameter that controls when the second transform is applied. By comparing the residual signal energy of each sub-block against this threshold parameter, the system dynamically adjusts processing complexity. This parameter-based control allows the encoder to adapt the level of processing to the actual content characteristics, improving compression performance while managing device complexity through threshold-based decision making.

The patent extracts and signals index information only for sub-blocks that meet specific criteria (those where the residual signal energy exceeds the threshold). By taking out only the necessary index information rather than signaling for all sub-blocks, the patent reduces bitstream overhead while maintaining sufficient accuracy to correctly identify which sub-blocks require the second transform.

Instead of signaling index information for all sub-blocks (excessive action), the patent signals index information only for sub-blocks where it is actually needed (partial action). This selective signaling approach reduces bitstream overhead by omitting redundant index information for sub-blocks that will not receive the second transform, while still maintaining the accuracy needed to correctly apply the transform where required.