Position-Dependent Prediction Combination for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently combining intra and inter prediction methods, particularly in scenarios where bi-directional optical flow refinement is not applicable, leading to suboptimal compression efficiency.
Innovation Solution
The proposed solution involves an apparatus and method for video encoding/decoding that utilizes a position-dependent prediction combination (PDPC) to reconstruct samples by combining intra and inter predictions, with optional filtering of neighboring samples, and includes a flag-based mechanism to determine the usage of PDPC, while excluding bi-directional optical flow refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bi-directional optical flow refinement is applied to inter prediction, then prediction accuracy is improved, but computational complexity increases and the method cannot be applied in all scenarios
Solution Approach 1:
The patent applies position-dependent prediction combination selectively based on the location of samples within the block. Different weighting schemes are used for different positions (e.g., corner samples vs. edge samples), allowing the system to optimize prediction accuracy locally without applying complex bi-directional optical flow refinement uniformly across all samples, thus reducing overall computational complexity while maintaining accuracy where most needed.
Solution Approach 2:
The patent segments the prediction process into multiple stages: first generating inter prediction samples, then applying position-dependent combination with neighboring samples. This segmentation allows the system to avoid the computationally intensive bi-directional optical flow refinement while still achieving improved prediction accuracy through the multi-stage approach.
2Productivity
If only inter prediction is used, then computational complexity is reduced, but compression efficiency deteriorates in scenarios where bi-directional optical flow refinement is not applicable
Solution Approach 1:
The patent merges inter prediction with intra prediction by combining inter prediction samples with neighboring samples using position-dependent weighting. This merging allows the system to achieve compression efficiency comparable to bi-directional optical flow refinement while avoiding its computational complexity, as the combined approach leverages both temporal (inter) and spatial (intra) correlations without requiring full bi-directional optical flow processing.
3Productivity
If position-dependent prediction combination is applied, then compression efficiency is improved, but additional processing steps are required
Solution Approach 1:
The patent performs preliminary actions by first generating inter prediction samples and identifying relevant neighboring samples before applying the position-dependent combination. This preliminary organization of data allows the subsequent combination step to proceed efficiently with pre-computed weighting factors, reducing the overall processing burden despite the additional computational step.
Data Source
AI summary
Aspects of the disclosure provide methods and apparatuses for video encoding/decoding. In some examples, an apparatus for video decoding includes processing circuitry. For example, the processing circuitry decodes prediction information of a current block from a coded video bitstream. The prediction information indicates that a prediction of the current block is at least partially based on an inter prediction. Then, the processing circuitry reconstructs at least a sample of the current block as a combination of a result from the inter prediction and neighboring samples of the block that are selected based on a position of the sample.