Intra Prediction Filtering Tabs for Video Residual Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding methods face challenges in minimizing the calculation amount during the transform coding process without degrading codec performance, particularly in intra prediction where efficient transform methods are needed to reduce the residual signal data quantity.
Innovation Solution
The method involves generating prediction pixels through interpolation using discrete cosine transform (DCT) or discrete sine transform (DST)-based filtering with adjustable filtering tabs, where the number of tabs is determined based on intra prediction direction, edge information, and pixel value changes, allowing for optimal filtering in various regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DCT or DST-based filtering with multiple tabs is used for intra prediction, then prediction accuracy and compression performance are improved, but calculation complexity increases
Solution Approach 1:
The patent dynamically adjusts the number of filtering tabs based on the intra prediction direction. Different prediction directions use different numbers of tabs (e.g., vertical direction uses different tab count than horizontal direction), making the filtering process adaptive rather than static. This resolves the contradiction by applying higher complexity only when needed for specific directions while using lower complexity for others.
Solution Approach 2:
The patent applies different filtering tab configurations to different regions or directions of the prediction block. By tailoring the filtering parameters (number of tabs) to local characteristics of the prediction direction, it achieves high prediction accuracy where needed while reducing unnecessary calculation complexity in other areas.
2Measurement precision
If the number of filtering tabs is increased to improve prediction accuracy, then residual signal intensity decreases, but encoding complexity increases
Solution Approach 1:
The patent changes the parameter of filtering tab count based on prediction direction requirements. By optimizing this parameter for each direction rather than using a fixed high value, it achieves good prediction accuracy while controlling encoding complexity and maintaining efficiency.
3Reliability
If different filtering methods are applied to different prediction directions, then compression performance is improved, but device complexity increases
Solution Approach 1:
The patent segments the prediction process by direction, applying different filtering tab configurations to vertical, horizontal, and diagonal predictions. This segmentation allows optimized performance for each direction while keeping the overall system manageable by using simple direction-based classification rather than complex adaptive methods.
Data Source
AI summary
A method for performing an intra prediction according to the present invention may comprise the steps of: generating predicted pixels by performing interpolation using N reference pixels including two adjacent reference pixels according to an intra prediction direction; and generating a residual signal by using the generated predicted pixels. Accordingly, when a prediction for generating the residual signal is performed according to each prediction direction (mode), an optimal filtering method is determined and a predicted value is acquired, so that a size of the residual signal can be minimized and compression performance can be improved.