Inter-Intra Prediction Weighting for Video Coding Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques, such as those in the Versatile Video Coding (VVC) standard, do not fully utilize neighboring block coding information to enhance inter-intra prediction performance, leading to potential artifacts along region boundaries and suboptimal coding efficiency.
Innovation Solution
The proposed solution involves a position-independent weight scheme for inter and intra prediction samples, where weights are adaptively determined based on the number of intra-coded and inter-prediction coded neighbor blocks, as well as the size of the current block, to form a combined prediction block for chrominance blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional inter-prediction or intra-prediction is used separately, then the coding process is simple, but the prediction accuracy is insufficient and artifacts appear along region boundaries
Solution Approach 1:
The patent combines inter-prediction and intra-prediction into a unified inter-intra prediction framework. The prediction block is formed by merging the inter-prediction block (from reference picture) and intra-prediction block (from current picture neighbors) with position-dependent weights, allowing the system to leverage both temporal and spatial redundancy for improved prediction accuracy while avoiding artifacts at region boundaries.
Solution Approach 2:
The patent applies position-dependent weighting where different spatial positions within the prediction block receive different weight combinations. Specifically, samples closer to intra-coded neighboring blocks receive higher intra-prediction weights, while samples farther away receive higher inter-prediction weights. This local adaptation of prediction quality resolves the contradiction by improving accuracy where needed without uniformly increasing complexity.
2Productivity
If equal weights are applied to inter and intra prediction blocks, then the processing is straightforward, but the coding efficiency is suboptimal
Solution Approach 1:
The patent introduces dynamic weight determination based on the actual coding modes of neighboring blocks. Rather than using fixed equal weights, the system adaptively adjusts weights according to whether neighboring blocks are intra-coded or inter-coded, and their distances from the current block. This dynamic adaptation improves coding efficiency by matching the weight distribution to the local characteristics of the video content.
Solution Approach 2:
The weight determination process utilizes information already available in the coding process (neighboring block coding modes and positions) without requiring additional complex calculations or external data. The system serves itself by leveraging existing coding information to derive optimal weights, improving efficiency while avoiding excessive complexity.
3Reliability
If neighboring block information is not utilized, then the processing is simple, but artifacts appear along region boundaries and coding efficiency decreases
Solution Approach 1:
The patent performs preliminary analysis of neighboring block coding modes before forming the final prediction block. By examining the coding modes (intra or inter) of neighboring blocks in advance, the system can pre-determine the appropriate weight distribution for each position in the prediction block. This preliminary action prevents artifacts at region boundaries by ensuring that the prediction smoothly transitions between different coding regions without requiring complex post-processing.
Data Source
AI summary
An example device for coding (encoding or decoding) video data includes a memory for storing video data and one or more processors implemented in circuitry and configured to form an inter-prediction block for a current chrominance block of the video data; form an intra-prediction block for the current chrominance block of the video data; determine a number of neighboring blocks to a luminance block corresponding to the current chrominance block that are intra-prediction coded; determine a first weight and a second weight according to the number of neighboring blocks that are intra-prediction coded; apply the first weight to the inter-prediction block and the second weight to the intra-prediction block; combine the first weighted inter-prediction block and the second weighted intra-prediction block to form a prediction block for the current chrominance block; and code the current chrominance block using the prediction block.