Video Intra Prediction Mode Selection via Gradient Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The High Efficiency Video Coding (HEVC) standard faces inefficiencies in intra prediction mode selection due to the complexity of Rate-Distortion optimization (RDO) across 35 intra prediction modes, leading to high computational workload and suboptimal prediction mode determination.
Innovation Solution
A method that classifies prediction blocks based on texture smoothness, adjusting the number of prediction modes for each type, reducing the number of candidates for RDO calculation, and pre-calculating gradient amplitudes and angles for efficient intra prediction mode selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If RDO is conducted on all 35 intra prediction modes, then prediction accuracy is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the 35 intra prediction modes into different groups based on block characteristics. For smooth blocks, only DC and planar modes are considered; for textured blocks, only angular modes corresponding to gradient histogram peaks are evaluated. This segmentation reduces the number of modes requiring full RDO computation while maintaining prediction accuracy for each block type.
Solution Approach 2:
The patent applies different prediction mode selection strategies to different regions of the prediction block based on local texture characteristics. Smooth regions use simplified DC/planar mode selection, while textured regions use gradient-based angular mode selection. This local differentiation reduces overall computational complexity while preserving prediction quality in each region.
2Productivity
If the number of prediction modes is reduced, then calculation efficiency is improved, but prediction accuracy may deteriorate
Solution Approach 1:
The patent changes the parameter of prediction mode selection by using gradient amplitude and gradient angle thresholds to dynamically determine which modes to evaluate. By calculating gradient histograms and identifying peak directions, the system adapts the set of candidate modes based on the actual block characteristics, ensuring that the reduced mode set still includes the optimal mode for each block type.
Solution Approach 2:
The patent performs preliminary gradient calculation and histogram analysis before the final mode selection. This preliminary action identifies the dominant texture directions and filters out unlikely candidate modes, so that subsequent RDO computation is only performed on the most promising modes, maintaining accuracy while improving efficiency.
3Measurement precision
If gradient calculations are performed for all blocks, then mode selection accuracy is improved, but computational workload increases
Solution Approach 1:
The patent applies partial gradient calculations only where necessary. Instead of performing exhaustive gradient analysis on all blocks, the method calculates gradients selectively based on block size and texture characteristics. For small blocks or smooth blocks, simplified criteria are used, while gradient histograms are computed only for blocks where texture analysis is beneficial, reducing overall computational workload.
Data Source
AI summary
This application discloses methods and apparatuses for intra prediction mode selection, one method comprising: acquiring a gradient amplitude and a gradient angle of each element in a prediction block; analyzing statistics on the gradient angles and generating a gradient angle histogram of the prediction block; if the texture smoothness of the prediction block is greater than a first pre-determined threshold value, determining that the prediction block is a first-type prediction block, and setting a prediction mode of the first-type prediction block to comprise a direct current prediction mode and a planar prediction mode; and if the texture smoothness of the prediction block is less than a second pre-determined threshold value, determining that the prediction block is a second-type prediction block, and setting an angular prediction mode of the second-type prediction block to comprise an angular prediction mode corresponding to the maximum peak value in the gradient angle histogram.