Intra-Inter Prediction Blending for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in efficiently compressing video data due to the limitations of traditional Inter and Intra prediction methods, which do not effectively handle complex motions and variations in video content, leading to suboptimal coding efficiency and increased storage/transmission bandwidth requirements.
Innovation Solution
A new Inter prediction mode is introduced that combines Intra and Inter prediction by blending predictors using weighting coefficients, allowing for adaptive prediction in boundary regions and enabling efficient encoding/decoding of video data, particularly through the use of Intra-Inter predictors generated by combining Intra and Inter predictors with weighted coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional Inter prediction is used, then motion compensation is provided based on reference pictures, but coding efficiency is suboptimal for complex motions
Solution Approach 1:
The patent combines Intra prediction and Inter prediction into a unified Intra-Inter prediction mode. The encoder selects reference pictures from both intra-picture (current picture) and inter-picture (reference pictures) lists, allowing the decoder to reconstruct blocks using a blended approach that leverages both spatial and temporal correlations, thereby improving prediction accuracy for complex motions while maintaining coding efficiency.
Solution Approach 2:
The patent creates a composite prediction signal by merging Intra and Inter predictors. The Intra predictor provides spatial redundancy exploitation from neighboring blocks in the current picture, while the Inter predictor provides temporal redundancy exploitation from reference pictures. This composite approach creates a more robust prediction that adapts to various motion patterns and content types.
2Reliability
If Intra prediction is used, then spatial redundancy is exploited from neighboring blocks, but it cannot effectively handle temporal variations and complex motions
Solution Approach 1:
The patent merges Intra prediction (spatial redundancy exploitation) with Inter prediction (temporal redundancy exploitation) into a single Intra-Inter prediction mode. This allows the system to simultaneously leverage spatial correlations from neighboring blocks and temporal correlations from reference pictures, effectively handling both spatial redundancy and temporal variations in a unified framework.
3Ease of manufacture
If separate Intra and Inter prediction modes are used, then each mode can be optimized independently, but device complexity increases and management becomes difficult
Solution Approach 1:
The patent merges Intra and Inter prediction modes into a single Intra-Inter prediction mode with unified syntax and processing. The encoder selects reference pictures from both intra-picture and inter-picture lists using a single reference picture index, and the decoder reconstructs blocks using a unified Intra-Inter reconstruction process. This integration simplifies mode management and reduces device complexity while preserving the optimization benefits of both approaches.
4Reliability
If video data is stored or transmitted in uncompressed form, then perfect quality is maintained, but storage and transmission bandwidth requirements become formidable
Solution Approach 1:
The patent segments the prediction process into multiple components (Intra predictor, Inter predictor, and their blended Intra-Inter predictor) that work together to achieve better compression. By dividing the prediction task into spatial and temporal components and combining them, the system achieves higher coding efficiency, reducing the quantity of data that needs to be stored or transmitted while maintaining video quality.
Data Source
AI summary
Method and apparatus of video coding are disclosed. According to the method, a boundary region in the current block is determined. For each pixel in the boundary region, an Intra predictor and an Inter predictor care derived. An Intra-Inter predictor is generated by blending the Intra predictor and the Inter predictor. Pixels in the boundary region of the current block are encoded or decoded using the Intra-Inter predictor. For the remaining pixels in the current block not belonging to the boundary region, the pixels are encoded or decoded using the Inter predictor or the Intra predictor. In another method, the Intra-Inter prediction process is implemented by scaling the sum of the Intra weighting factor and the Inter weighting factor is scaled to a power-of-2 number to avoid the need for the division operation. In yet another method, the Intra-Inter prediction is applied depending on the current block size.