Intra Prediction Mode Adaptation for High-Resolution Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding/decoding methods and apparatuses face inefficiencies in performing intra prediction, particularly in handling high-resolution videos and efficiently encoding/decoding intra prediction modes.
Innovation Solution
The proposed method involves obtaining intra prediction information from a bitstream, deriving an intra prediction mode for a current block, and performing intra prediction using a reference area determined based on the mode and block characteristics. This includes generating an MPM list that adapts to neighboring block modes, reference area locations, and split information, with specific configurations for directional and non-directional modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional intra prediction methods are used, then the encoding process is simple, but the compression efficiency is insufficient for high-resolution videos
Solution Approach 1:
The patent segments the intra prediction process into distinct phases: generating candidate modes, selecting the optimal mode, and reconstructing the block. This segmentation allows for more sophisticated prediction algorithms to be applied in each phase, improving overall compression efficiency without overwhelming system complexity.
Solution Approach 2:
The patent introduces dynamic adaptation in intra prediction by adjusting prediction modes and parameters based on block characteristics, neighboring block modes, and reference line positions. This dynamic approach enables the system to optimize compression efficiency for each specific block while managing complexity through context-dependent decision-making.
2Measurement precision
If more intra prediction modes are used, then prediction accuracy improves, but encoding complexity increases
Solution Approach 1:
The patent applies local quality by using different prediction modes and reference lines based on the specific characteristics of each block and its neighbors. Rather than applying a uniform prediction approach, the system selects from multiple modes (including planar, DC, and directional modes) based on local block properties, achieving higher prediction accuracy without uniformly increasing complexity across all blocks.
Solution Approach 2:
The patent changes parameters such as the number and type of candidate modes, reference line positions, and prediction mode selections based on block characteristics. This parameter adaptation allows the system to achieve higher prediction accuracy for complex blocks while using simpler parameters for easier blocks, effectively managing the trade-off between accuracy and complexity.
3Adaptability or versatility
If multiple reference lines are used for intra prediction, then prediction flexibility increases, but processing time increases
Solution Approach 1:
The patent applies partial action by using a limited number of reference lines (typically up to 4) rather than all possible reference lines. The system selectively activates reference lines based on block position, size, and neighboring block characteristics, achieving sufficient prediction flexibility without the processing overhead of evaluating all possible reference lines for every block.
Solution Approach 2:
The patent performs preliminary actions by pre-identifying and ordering candidate reference lines based on their likelihood of providing good prediction results. Reference lines are prioritized and prepared in advance, allowing the decoder to efficiently select from pre-organized candidates rather than evaluating all reference lines during the actual prediction process, thus reducing processing time while maintaining flexibility.
Data Source
AI summary
The present invention provides an image encoding method and an image decoding method. The image decoding method of the present invention may comprise: obtaining intra-frame prediction information of a current block from a bitstream; inducing an intra-frame prediction mode of the current block on the basis of the intra-frame prediction information; determining a reference region for intra-frame prediction of the current block; and performing intra-frame prediction of the current block on the basis of the intra-frame prediction mode and the reference region.