Gradient-Based Boundary Filtering for DC and Planar Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding methods face inefficiencies in intra prediction, particularly in DC and PLANAR modes, due to discontinuities at block boundaries, which affect coding performance and visual quality.
Innovation Solution
The proposed solution involves gradient-based boundary filtering in DC and PLANAR prediction modes, where gradients between reference samples and predicted values are computed, and a decay function is used to adjust the predicted values, reducing discontinuities and improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gradient-based boundary filtering is applied in DC and PLANAR prediction modes, then coding performance and visual quality are improved, but computational complexity increases
Solution Approach 1:
The patent applies boundary filtering to reference samples before they are used in prediction, pre-processing the reference data to eliminate discontinuities. This preliminary action ensures that when DC and PLANAR prediction modes generate predicted blocks, the boundary artifacts are already minimized, improving coding performance without requiring complex post-processing operations.
Solution Approach 2:
The patent applies filtering operations specifically at block boundaries where discontinuities occur, rather than uniformly across all reference samples. By targeting only the boundary regions with filtering operations and using decay functions that localize the effect, the solution improves prediction accuracy at critical interfaces while minimizing unnecessary computations in already-smooth regions.
2Manufacturing precision
If boundary filtering is applied to reduce discontinuities at block boundaries, then visual quality is improved, but encoding time increases
Solution Approach 1:
The patent modifies the reference sample values at boundaries by applying filtering operations that adjust local parameters (pixel values) based on neighboring samples. This parameter change approach smoothly transitions values at block boundaries, reducing visual artifacts and improving quality while maintaining a relatively simple computational structure that limits encoding time increase.
Data Source
AI summary
When a block is intra predicted, the predicted samples on the boundary are subject to boundary filtering for certain modes. Various embodiments are proposed to perform boundary filtering for intra prediction in the DC or PLANAR mode, using a gradient based approach. The gradient for a particular reference sample is calculated based on a difference between the particular reference sample and a predictor for predicting the particular reference sample. For the PLANAR mode, the difference between a top (left) reference sample and a linear interpolation using the top (left) reference samples is used as the gradient, a fractional part of which is propagated to the initial predicted samples with a decay function. For the DC mode, the difference between a top (left) reference sample and an average of reference samples is used as the gradient. The average can be computed with the top reference samples and/or left reference samples.