Constructed Reference Samples for Image Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-resolution and high-quality image data requires high-efficiency encoding/decoding techniques to reduce the increased costs of transmission and storage associated with conventional methods.
Innovation Solution
The method involves deriving an intra prediction mode using rightmost and bottommost samples of a block, constructing reference samples through interpolation, and performing intra prediction with weighted sums and overlay predictors to enhance compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image encoding/decoding techniques are used for high-resolution and high-quality images, then image quality and resolution are improved, but transmission and storage costs increase
Solution Approach 1:
The image block is divided into multiple regions (e.g., first block, second block, third block, fourth block) with different prediction modes applied to each region. This segmentation allows the patent to use more complex prediction methods for critical areas while using simpler methods for other areas, thereby improving overall image quality while controlling compression complexity and transmission costs.
Solution Approach 2:
Different prediction modes (e.g., planar mode for some regions, directional modes for others) are applied to different regions of the image block based on local characteristics. This local quality approach ensures that each region is encoded with the most appropriate method for its specific features, improving overall image quality while maintaining efficient compression.
2Device complexity
If intra prediction uses only top and left reference samples, then device complexity is reduced, but compression efficiency deteriorates
Solution Approach 1:
The patent extends the reference sample usage from traditional top and left directions to include bottom and right directions as well. By utilizing reference samples from all four directions (top, left, bottom, right), the patent improves prediction accuracy and compression efficiency while maintaining manageable device complexity through systematic processing.
Solution Approach 2:
The patent creates multiple prediction blocks (first, second, third, fourth blocks) using different prediction modes, each serving different functional purposes. This multi-functionality approach allows the system to handle various image characteristics efficiently, improving overall compression efficiency without significantly increasing device complexity.
Data Source
AI summary
A method of decoding an image includes, deriving an intra prediction mode for a current block, decoding at least one original sample that is present in a rightmost column and a bottommost row (a bottom row) of the current block, constructing a reference sample by using the at least one decoded original sample and performing intra prediction on the current block by using the constructed reference sample.


