Intra-Prediction Boundary Filtering Skip for Image Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intra-prediction methods using multiple reference samples for image processing often result in excessive smoothing at block boundaries, reducing prediction accuracy and efficiency due to excessive boundary filtering.
Innovation Solution
A method that adaptively determines whether to apply boundary filtering based on the intra-prediction mode, skipping filtering when bi-directional intra-prediction is applied and using weighted reference samples, and applying uni-directional prediction with boundary filtering when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If boundary filtering is applied to reduce block boundary artifacts, then smoothing effect is improved, but prediction accuracy deteriorates due to excessive smoothing
Solution Approach 1:
The patent changes the parameter of boundary filtering application by introducing a flag (intra_bdlf_flag) that controls whether boundary filtering is applied. This flag is determined based on the intra-prediction mode, allowing the system to adaptively adjust the filtering parameter to avoid excessive smoothing while maintaining block boundary smoothness where needed.
Solution Approach 2:
The patent makes the boundary filtering process dynamic by conditionally applying it based on the intra-prediction mode. When bi-directional intra-prediction is detected, boundary filtering is skipped; otherwise, it is applied. This dynamic approach allows the system to adapt to different prediction scenarios and avoid excessive smoothing.
2Measurement precision
If bi-directional intra-prediction with multiple reference samples is used, then prediction accuracy is improved, but unnecessary boundary filtering increases processing complexity
Solution Approach 1:
The patent extracts and removes the boundary filtering step from the processing chain when bi-directional intra-prediction is detected. By taking out this unnecessary processing step, the system maintains high prediction accuracy while reducing processing complexity and avoiding redundant operations.
Solution Approach 2:
The patent implements a skipping mechanism where boundary filtering is omitted (skipped) when the intra-prediction mode indicates bi-directional prediction. This is controlled by the intra_bdlf_flag, which causes the system to rush through the prediction process without the additional filtering step, thereby reducing complexity.
3Reliability
If adaptive boundary filtering is applied selectively, then prediction performance is improved, but additional mode detection increases processing overhead
Solution Approach 1:
The patent performs preliminary action by detecting the intra-prediction mode and setting the intra_bdlf_flag before the actual prediction and filtering processes. This advance determination allows the system to know whether boundary filtering should be applied, enabling efficient processing without redundant operations and improving overall prediction performance.
Data Source
AI summary
The present invention provides an image processing method on the basis of an intra prediction mode and an apparatus therefor. Specifically, a method for processing an image on the basis of an intra prediction mode may comprise the steps of: inducing a prediction mode applied to an intra prediction of a current block; determining whether a bi-directional intra prediction using reference samples located at both boundaries of the current block with respect to a prediction direction of the prediction mode is applied to the current block; and generating a prediction sample by using a reference sample neighboring the current block on the basis of the prediction mode, wherein when the bi-directional intra prediction is applied, boundary filtering for the current block is skipped.