DMVR Inter-Prediction Refinement for Lower-Complexity Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images and videos, such as UHD images and videos of 4K or 8K, poses a challenge in efficiently compressing and transmitting or storing these data due to the increased amount of information required, leading to higher transmission and storage costs. Additionally, the need for efficient compression technologies that can handle various characteristics of immersive media, like VR and AR content, further complicates the issue.
Innovation Solution
The proposed solution involves a method and apparatus for enhancing image coding efficiency through inter prediction techniques, specifically by employing Decoder-side Motion Vector Refinement (DMVR) and Bi-directional optical flow (BDOF). These techniques refine motion vectors and improve prediction performance by applying specific conditions to determine when to use DMVR and BDOF, thereby optimizing coding efficiency and reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-resolution and high-quality images and videos are transmitted or stored, then image and video quality is improved, but transmission costs and storage costs are increased
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting motion vector refinement based on block characteristics and prediction modes. The system changes parameters such as refinement precision and computational complexity according to the specific requirements of different video blocks, achieving high quality where needed while reducing overall data volume through intelligent parameter adaptation.
Solution Approach 2:
The patent implements local quality by applying different levels of motion vector refinement to different blocks within a video frame. Instead of uniformly processing all blocks at maximum precision, the system identifies blocks that benefit most from refinement and applies computational resources selectively, improving local prediction accuracy while reducing overall processing load and data transmission requirements.
2Productivity
If Decoder-side Motion Vector Refinement (DMVR) is applied to improve prediction accuracy, then coding efficiency is improved, but calculation complexity is increased
Solution Approach 1:
The patent applies dynamics by making the motion vector refinement process adaptive rather than static. The system dynamically adjusts the refinement level based on real-time conditions such as block size, prediction mode, and motion characteristics. This allows the system to achieve high coding efficiency for complex motion scenarios while reducing computational complexity for simpler cases through dynamic adaptation.
Solution Approach 2:
The patent implements partial action by applying motion vector refinement only to the extent necessary for achieving good prediction accuracy, rather than uniformly applying maximum refinement everywhere. The system identifies when refinement is beneficial and when it is excessive, applying computational effort partially only where it provides meaningful improvement in coding efficiency.
3Productivity
If Bi-directional optical flow (BDOF) is applied to enhance prediction performance, then image coding efficiency is improved, but processing time is increased
Solution Approach 1:
The patent applies periodic action by implementing BDOF in a structured, step-wise manner rather than as a continuous process. The system performs optical flow calculation in discrete stages, evaluating motion vectors at specific intervals and only when necessary based on block characteristics. This periodic approach maintains coding efficiency for complex motion while reducing overall processing time compared to continuous refinement methods.
Data Source
AI summary
An image decoding method comprises the steps of: determining whether or not an application condition of decoder-side motion vector refinement (DMVR) for applying motion vector refinement for a current block is satisfied; deriving a minimum sum of absolute differences (SAD) on the basis of L0 and L1 motion vectors of the current block if the application condition of the DMVR is satisfied; deriving refined L0 and L1 motion vectors of the current block based on the minimum SAD; deriving prediction samples of the current block based on the refined L0 and L1 motion vectors; and generating reconstructed samples of the current block based on the prediction samples. With respect to whether or not the application condition of the DMVR is satisfied, the application condition of the DMVR is determined to be satisfied if a prediction mode, in which inter-prediction and intra-prediction are combined, is not applied to the current block.