Image Block Boundary Blending for More Accurate Inter Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images leads to higher data volumes, resulting in increased transmission and storage costs, necessitating improved high-efficiency image encoding/decoding technologies, particularly in inter prediction methods where accurate prediction of partitioning boundaries is challenging.
Innovation Solution
An image encoding/decoding method that partitions a current block into two partitions, determines a blending area around the partitioning boundary, and calculates weights for each partition to improve prediction accuracy, using methods such as geometric partitioning and adaptive weight determination based on the blending area's width.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If inter prediction with partitioning is used to improve encoding efficiency, then encoding speed and compression ratio improve, but prediction accuracy at partitioning boundaries deteriorates
Solution Approach 1:
The patent applies local quality by differentiating the treatment of boundary samples versus non-boundary samples. Specifically, samples adjacent to the partitioning boundary (within the blending area) are processed differently through weighted averaging of prediction values from both partitions, while non-boundary samples use direct partition prediction. This local differentiation resolves the contradiction by maintaining high encoding efficiency through partitioning while improving prediction accuracy specifically at the problematic boundary regions through adaptive blending with position-dependent weights.
2Measurement precision
If blending area width is increased to improve boundary prediction accuracy, then prediction quality improves, but encoding complexity and computational load increase
Solution Approach 1:
The patent applies dynamics by making the blending area width variable rather than fixed. The width is dynamically adjusted based on the block size, with different block sizes corresponding to different blending area widths. This dynamic adaptation resolves the contradiction by optimizing the balance between boundary prediction quality and encoding complexity for each specific block, avoiding excessive computational load on small blocks while ensuring sufficient blending for larger blocks where boundary effects are more pronounced.
3Device complexity
If fixed blending area width is used to simplify encoding process, then encoding complexity decreases, but prediction accuracy for different block sizes deteriorates
Solution Approach 1:
The patent applies parameter changes by making the blending area width a variable parameter that changes according to the block size. Different block sizes (e.g., 4x4, 8x8, 16x16, 32x32) are associated with different blending area widths, allowing the encoding process to adapt to the specific characteristics of each block. This resolves the contradiction by maintaining encoding process simplicity through a systematic parameter mapping while achieving optimal prediction accuracy for each block size category.
Data Source
AI summary
An image decoding method may comprise partitioning a current block into a first partition and a second partition according to a partitioning boundary, determining a first prediction block for the first partition and a second prediction block for the second partition, determining a width of a blending area around the partitioning boundary, determining a first weight and a second weight for determining a final prediction sample of the blending area based on the width of the blending area, and determining a final prediction block from the first prediction block and the second prediction block based on the first weight and the second weight.