Refined Intra Prediction Angles for Non-Square Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in intra prediction, particularly in handling non-square rectangles and non-standard aspect ratios, leading to suboptimal compression and reconstruction of video data.
Innovation Solution
Implementing an angle-to-offset mapping lookup table with derivative values that are powers of two, allowing for refined intra prediction angles and adaptive mapping for non-square rectangles, and using shifting operations to reduce computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional intra prediction methods are used for non-square rectangles, then the coding process is simple, but the prediction accuracy and compression efficiency deteriorate
Solution Approach 1:
The patent changes the parameter representation by using angle-to-offset mapping with derivative values stored in lookup tables. Instead of directly using conventional prediction parameters, the system transforms angle information into offset values that better represent non-square rectangular blocks, improving prediction accuracy while managing complexity through pre-computed tables.
Solution Approach 2:
The patent applies preliminary action by pre-computing and storing derivative values in angle-to-offset mapping lookup tables before the actual prediction process. This allows the decoder to quickly retrieve appropriate offset values for given angles without performing complex calculations in real-time, thus improving prediction accuracy for non-square rectangles while keeping the decoding complexity manageable.
2Measurement precision
If refined intra prediction angles are implemented, then the prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent substitutes complex mechanical computation with a lookup table-based approach. Instead of performing real-time trigonometric calculations or complex iterative optimization to find refined prediction angles, the system uses pre-computed angle-to-offset mapping tables that provide refined angle information through simple index-based retrieval, significantly reducing computational power requirements while maintaining high prediction accuracy.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of angle-to-offset mapping lookup tables that act as a bridge between the coded angle information and the actual prediction process. These tables store pre-computed derivative values that mediate the transformation from angle parameters to pixel offset calculations, enabling refined prediction without direct complex computation.
3Ease of manufacture
If conventional lookup tables are used for angle mapping, then the implementation is simple, but the precision for non-standard aspect ratios deteriorates
Solution Approach 1:
The patent applies local quality by creating separate angle-to-offset mapping lookup tables tailored for different block aspect ratios. Instead of using a single generic lookup table, the system implements distinct tables optimized for specific aspect ratios (e.g., 1:2, 2:1, 1:4, 4:1), ensuring that each table provides high precision for its intended use case while maintaining implementation simplicity through a systematic table structure.
4Measurement precision
If more derivative values are stored in the lookup table, then the prediction precision improves, but the memory requirements increase
Solution Approach 1:
The patent applies partial action by storing only the most essential derivative values in the angle-to-offset mapping lookup tables. Instead of pre-computing and storing all possible derivative values for every potential angle and block configuration, the system selectively stores a subset of critical values that provide sufficient prediction precision for the majority of cases, thereby reducing memory requirements while maintaining acceptable prediction accuracy.
Data Source
AI summary
Processing circuitry receives, from a coded video bitstream, coded information of a current coding block in a current picture. The coded information is indicative of a directional intra prediction of the current coding block. The processing circuitry determines an index to an entry in an angle-to-offset mapping lookup table according to the coded information of current coding block. The angle-to-offset mapping lookup table stores a set of derivative values that are associated with angles, the set of derivative values includes two or more integers that are powers of two. The processing circuitry accesses the entry to obtain a derivative value, determines, for a pixel in the current coding block, an offset indicative of a reference location in the current picture according to the derivative value and a location of the pixel, and reconstructs the pixel in the current coding block based on the reference location in the current picture.