JVET Planar Prediction Plane Construction Using Least Squares Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding standards, such as JVET, face challenges in constructing a predicted plane for the planar mode, as the final planar predictor is not actually a plane, leading to inefficiencies in video compression.
Innovation Solution
A method is introduced to construct a plane for planar prediction in JVET using the least squares method, determining lines based on neighboring pixels and averaging y-intercepts to define a plane, which can be used as a predictor for current coding units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the existing planar prediction method in JVET is used, then the coding process is simple, but the predicted plane is not actually a plane leading to lower coding efficiency
Solution Approach 1:
The patent changes the parameters used in plane construction from simple gradient calculations to least squares method parameters (slope and intercept). By fitting a proper mathematical plane equation z = ax + by + c to the pixel data, the invention ensures the predictor is actually a plane while maintaining coding efficiency through systematic parameter determination.
Solution Approach 2:
The patent replaces the existing mechanical calculation system (simple gradient-based plane construction) with a more robust mathematical system (least squares fitting). This substitution ensures the predicted surface satisfies the mathematical definition of a plane while improving prediction accuracy and coding efficiency.
2Measurement precision
If a proper plane is constructed using least squares method, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The invention determines plane parameters (slope a, slope b, and intercept c) using the least squares method, which provides the optimal fit for the data. This mathematical approach ensures maximum prediction accuracy by minimizing the sum of squared errors between the predicted and actual pixel values.
Solution Approach 2:
The patent uses neighboring pixels to construct the prediction plane, copying spatial information from already-decoded adjacent blocks. This approach leverages the redundancy in video data while maintaining accuracy through systematic least squares fitting of the copied pixel values.
Data Source
AI summary
A system and method of defining a plane for planar coding in JVET in which first and second lines can be defined based upon pixels in left-adjacent and top-adjacent coding units. In some embodiments, the least squares method can be employed to define the relevant lines. One point along each of the lines can then be identified and the y-intercepts of the two lines can be averaged to obtain a third point. The three points can then be used to identify and define a plane for planar coding in JVET.