Curved Intra Prediction for Higher-Efficiency Image Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding techniques face limitations in enhancing encoding efficiency due to the use of only straight-directional prediction in intra prediction, leading to increased costs in transmitting and storing high-resolution and high-quality image data.
Innovation Solution
Incorporating curved intra prediction modes, which include information on intra prediction modes, direction information, and reference pixel positions, specified in units such as pixel groups, horizontal lines, and diagonal lines, to generate prediction blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If only straight-directional prediction is used in intra prediction, then the intra prediction process is simple, but encoding efficiency cannot be enhanced for high-resolution and high-quality images
Solution Approach 1:
The patent introduces curved intra prediction modes that use curved lines instead of straight lines to generate prediction blocks. This allows the prediction to follow curved edges and contours in the image, improving prediction accuracy for high-resolution and high-quality images while maintaining a manageable level of complexity through mathematical modeling of curved paths.
Solution Approach 2:
The patent changes the fundamental parameter of prediction line geometry from straight to curved by introducing curvature parameters (e.g., radius of curvature, curvature direction). This enables the prediction process to adapt to different edge types in the image, enhancing encoding efficiency without requiring completely new prediction mechanisms.
2Productivity
If curved intra prediction modes are incorporated, then encoding efficiency is improved, but the complexity of intra prediction increases
Solution Approach 1:
The patent segments the prediction process by dividing the current block into multiple pixel groups or sub-blocks, where different curved prediction modes can be applied to different segments. This segmentation allows the system to manage complexity by applying curved prediction only where needed rather than uniformly across the entire block, improving encoding efficiency while controlling computational complexity.
Solution Approach 2:
The patent implements curved intra prediction as an optional enhancement that can be selectively applied to specific blocks or regions based on image content characteristics. By using curved prediction partially (only when beneficial) rather than excessively (always), the system improves encoding efficiency for complex images while avoiding unnecessary complexity for simpler cases.
Data Source
AI summary
An intra prediction method and apparatus are discussed. In an example, an image decoding method can include deriving an intra prediction mode of a current block, deriving neighboring reference samples of the current block, deriving filtered reference samples by performing filtering on the neighboring reference samples, and generating a predicted block based on the filtered reference samples. The neighboring reference samples can be derived based on a substitution process.