Motion Vector Coding Using SMVD and Short-Term References
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and immersive media such as VR and AR content has led to a need for more efficient image/video compression technologies to reduce transmission and storage costs, particularly in handling motion vector differences and inter prediction procedures.
Innovation Solution
The method and device improve image/video coding efficiency by signaling information on motion vector differences, specifically L0 and L1 motion vector differences, using bi-prediction and short-term reference pictures, and deriving SMVD reference indices based on reference picture marking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution, high-quality image/video data is transmitted or stored using existing mediums, then image quality and resolution are maintained, but transmission cost and storage cost increase
Solution Approach 1:
The patent changes the parameter of data representation by using advanced compression techniques including motion vector prediction, bi-prediction, and symmetric motion vector difference (SMVD) to reduce the bit depth and information content while maintaining visual quality at high resolutions
Solution Approach 2:
The patent segments the image/video data into blocks and uses inter prediction with motion vectors to process only the differences between blocks rather than transmitting all pixel data, reducing the total data volume while preserving quality
2Productivity
If motion vector difference information is signaled using traditional methods, then inter prediction is performed, but coding efficiency is insufficient for high-resolution content
Solution Approach 1:
Instead of signaling full motion vector information, the patent uses partial action by transmitting only motion vector differences (MVD) and using prediction from neighboring blocks, reducing the bit depth required while maintaining prediction accuracy
Solution Approach 2:
The patent uses copying by predicting motion vectors from neighboring blocks and using bi-prediction to combine information from multiple reference pictures, avoiding the need to transmit complete motion data for all blocks
3Measurement precision
If bi-prediction with multiple reference pictures is applied, then prediction accuracy improves, but signaling complexity increases
Solution Approach 1:
The patent applies local quality by using bi-prediction selectively based on local image characteristics and motion patterns, applying the more complex bi-prediction mode only where beneficial rather than uniformly across all blocks
Solution Approach 2:
The patent uses preliminary action by pre-marking reference pictures as short-term or long-term references and pre-configuring prediction parameters, reducing the complexity of real-time signaling and processing
Data Source
AI summary
According to embodiments of the present document, symmetric motion vector differences (SMVDs) and merge mode motion vector differences (MMVDs) according to an inter prediction mode may be performed on the basis of reference picture types and, particularly, short term reference pictures may be used. Therefore, prediction performance and coding efficiency in the inter prediction mode can be improved.