Extrapolation Filter Intra Prediction with Independent Sample Sets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently utilizing intra prediction methods, particularly in determining gradient information for sample prediction, which affects compression efficiency and quality.
Innovation Solution
The implementation of an extrapolation filter-based intra prediction (EIP) mode that sets the number and positions of input samples independently for horizontal and vertical gradient information, using nonlinear relationships to determine predicted values, enhancing the precision of sample prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional intra prediction methods are used with shared input samples for gradient and prediction, then device complexity is reduced, but prediction accuracy deteriorates
Solution Approach 1:
The patent segments the input sample configuration into two independent sets: one set for gradient calculation and another set for initial prediction. This segmentation allows each set to be optimized independently for its specific purpose, improving overall prediction accuracy without requiring complex shared configurations. The encoder can selectively configure sample numbers and positions for each purpose.
Solution Approach 2:
The patent introduces dynamic configurability where the number and positions of input samples can be independently adjusted based on block characteristics, gradient directions, and prediction modes. This dynamic approach allows the system to adapt to different content types and prediction scenarios, improving accuracy while maintaining manageable complexity through conditional logic rather than fixed complex structures.
2Measurement precision
If the number and positions of input samples are set independently for horizontal and vertical gradient information, then prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by configuring input samples specifically for gradient calculation needs rather than using a uniform configuration for all purposes. Different numbers and positions of samples can be selected based on the specific gradient direction (horizontal or vertical) and block characteristics, optimizing the quality of gradient information locally for each prediction scenario.
Solution Approach 2:
The patent changes parameters (number and positions of input samples) based on prediction mode, block size, and gradient direction. This parameter adaptation allows the system to optimize gradient calculation accuracy for different scenarios without requiring a permanently complex configuration structure, as the complexity is activated only when beneficial.
3Measurement precision
If more input samples are used for gradient calculation, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by using only the necessary number of input samples for gradient calculation based on the specific prediction scenario. Rather than always using the maximum number of samples, the system selectively uses additional samples only when they provide meaningful gradient information, avoiding unnecessary computational complexity while maintaining accuracy where needed.
Data Source
AI summary
Methods and apparatuses for video decoding and video encoding and a method of processing visual media data are described. The apparatus for video decoding comprises processing circuitry configured to: receive predicted information indicating that a current block in a current picture is predicted using an extrapolation filter-based intra prediction (EIP) mode; determine gradient information associated with a current sample in the current block; determine a predicted value of the current sample based on an initial predicted value predicted using the EIP mode and additional information that includes the gradient information; and reconstruct the current sample from the predicted value of the current sample.