Large-Block Intra Prediction Signaling for Hierarchical Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as MPEG-4 AVC, do not effectively address how to signal intra prediction for large block sizes greater than 32x32, leading to inefficiencies and increased overhead, and lack support for hierarchical intra prediction.
Innovation Solution
A method and apparatus for signaling intra prediction for large blocks by selecting a basic coding unit size and assigning a single spatial intra partition type, allowing hierarchical layer intra prediction through splitting or merging operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If INTRA32x32 or INTRA64x64 prediction are added by extending the mb_type in the MPEG-4 AVC Standard, then large block intra prediction capability is improved, but signaling overhead increases significantly and hierarchical type of intra prediction is not allowed
Solution Approach 1:
The patent segments the intra prediction signaling into multiple layers: a first layer signals the large block type (INTRA32x32 or INTRA64x64), and a second layer signals the specific intra prediction mode. This hierarchical segmentation allows efficient signaling by separating the block size indication from the mode indication, reducing overall overhead while supporting both large block prediction and hierarchical prediction types.
Solution Approach 2:
The patent introduces a new dimension to the signaling structure by adding a block type indication layer before the mode indication layer. This dimensional expansion allows the system to efficiently represent both the block size and prediction mode without excessive overhead, enabling support for INTRA32x32 and INTRA64x64 while maintaining compact signaling through the two-layer approach.
2Productivity
If a larger transform (such as a 16x16 transform) instead of a cascaded transform is used for INTRA16x16, then transform efficiency is improved, but the current signaling cannot be applied
Solution Approach 1:
The patent changes the transform size parameter from the traditional cascaded 4x4 approach to a larger 16x16 transform for INTRA16x16 prediction modes. This parameter change improves transform efficiency by better matching the spatial correlations in larger blocks, while the signaling structure is adapted to accommodate this change through the two-layer approach that can indicate both the transform type and size appropriately.
3Measurement precision
If different priority for intra prediction modes inside one intra partition type is given, then prediction accuracy is improved, but signaling complexity increases
Solution Approach 1:
The patent applies different priority levels to different intra prediction modes within the same partition type by signaling the most probable mode first in the second layer, followed by remaining modes if necessary. This local differentiation of mode priorities allows the encoder to achieve better prediction accuracy by emphasizing the most likely modes, while the signaling complexity is managed through the efficient two-layer structure that only signals what is necessary.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Methods and apparatus are provided for signaling intra prediction for large blocks for video encoders and decoders. An apparatus includes a video encoder (400) for encoding picture data for at least one large block in a picture by signaling intra prediction for the at least one large block. The intra prediction is signaled by selecting a basic coding unit size and assigning a single spatial intra partition type for the basic coding unit size. The single spatial intra partition type is selectable from among a plurality of spatial intra partition types. The at least one large block has a large block size greater than a block size of the basic coding unit. The intra prediction is hierarchical layer intra prediction and is performed for the at least one large block by at least one of splitting from the large block size to the basic coding unit size and merging from the basic coding unit size to the large block size.