IBC Pairwise Average Candidates for Flipped Block Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies face inefficiencies in handling screen content, particularly in accurately predicting and encoding/decoding blocks within the same picture, leading to suboptimal compression and quality trade-offs.
Innovation Solution
The implementation of Intra Block Copy (IBC) merge and Advanced Motion Vector Prediction (AMVP) techniques, combined with a pairwise average (PA) candidate list construction based on flipping types, enhances block prediction by utilizing reference blocks within the same picture, improving prediction accuracy and compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional video encoding techniques are used, then the encoding process is simple, but prediction accuracy for screen content is insufficient and compression efficiency is suboptimal
Solution Approach 1:
The patent segments the prediction process into multiple candidate lists (initial IBC merge candidate list, initial AMVP candidate list, flipped candidate lists) and processes blocks in specific scan orders. This segmentation allows systematic exploration of multiple prediction sources without overwhelming complexity, improving prediction accuracy through structured multi-stage candidate generation and selection.
Solution Approach 2:
The patent performs preliminary actions by pre-establishing candidate lists and pre-calculating prediction candidates before final encoding decisions are made. The encoder constructs initial candidate lists and flipped candidate lists in advance, allowing the final prediction to be selected from pre-computed options, thereby improving prediction accuracy while managing computational complexity through preparation rather than real-time calculation.
2Productivity
If IBC and AMVP techniques are applied, then prediction accuracy improves, but the encoding process becomes more complex
Solution Approach 1:
The patent merges IBC (Intra Block Copy) and AMVP (Advanced Motion Vector Prediction) techniques into a unified prediction framework. By combining these methods and integrating them with flipped candidate lists, the system achieves higher compression efficiency through synergistic effects, where the complexity of individual methods is offset by their combined effectiveness in capturing both spatial and temporal correlations in screen content.
Solution Approach 2:
The patent creates a universal prediction mechanism that can handle multiple scenarios (IBC mode, AMVP mode, flipped modes) through a single integrated architecture. The candidate list construction and selection processes work universally across different prediction modes, allowing the system to achieve improved compression efficiency without proportionally increasing complexity, as the same framework serves multiple prediction needs.
3Measurement precision
If multiple candidate lists are constructed, then prediction accuracy increases, but computational load increases
Solution Approach 1:
The patent applies partial action by constructing and evaluating only the necessary candidate lists and prediction modes for each specific block, rather than exhaustively processing all possible predictions. The system builds initial candidate lists and flipped candidate lists selectively, evaluating only those that provide the most benefit for each block's prediction, thereby achieving high prediction accuracy while controlling computational load through targeted rather than comprehensive processing.
Solution Approach 2:
The patent implements feedback mechanisms where prediction results from candidate lists are evaluated and used to guide further processing. The encoder uses feedback from prediction error analysis to determine which candidate lists and prediction modes warrant further computation, allowing the system to allocate computational resources dynamically based on actual prediction needs, thus improving accuracy where needed while reducing overall computational load through feedback-driven optimization.
Data Source
AI summary
A video coder (encoder or decoder) determines a first flipping type of a first candidate block vector predictor (BVP) in a list of candidate BVPs. The coder further determines, based on the first flipping type of the first candidate BVP, a second candidate BVP from the list of candidate BVPs. The coder adds, to the list of candidate BVPs, a pairwise average (PA) candidate BVP determined based on the first candidate BVP and the second candidate BVP and codes a current block (CB) based on a candidate BVP selected from the list.