Video Coding Mode Selection Using BIO and DMVR Cost Criteria
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in optimizing compression ratios and decoding efficiency, particularly in handling bi-directional optical flow (BIO) and decoder-side motion vector refinement (DMVR) techniques, which incur high computational complexity and inefficiencies.
Innovation Solution
The proposed methods enable dynamic enabling or disabling of BIO and DMVR techniques based on cost criteria, such as mean removed sum of absolute differences (MRSAD), sum of absolute transformed differences (SATD), and gradient values, to optimize video processing, and include modifications to reference blocks and gradients for early termination and refined motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If BIO and DMVR techniques are applied to the current block, then video quality and compression ratios are improved, but computational complexity increases
Solution Approach 1:
The patent implements dynamic enabling or disabling of BIO and DMVR techniques based on cost criteria evaluation. The system adaptively selects whether to apply these computationally intensive techniques on a per-block basis, transforming a static always-on approach into a dynamic conditional approach that balances quality and complexity
Solution Approach 2:
The patent changes the parameter of technique application from binary (always on/off) to conditional (based on cost criteria). By evaluating cost metrics and comparing against thresholds, the system dynamically adjusts which blocks receive BIO/DMVR processing, optimizing the trade-off between video quality improvement and computational complexity
2Manufacturing precision
If BIO and DMVR techniques are applied to all blocks, then compression ratio is improved, but decoding efficiency decreases
Solution Approach 1:
The patent applies BIO and DMVR techniques selectively to specific blocks based on their individual characteristics and cost criteria, rather than uniformly to all blocks. This local differentiation allows the system to concentrate computational resources on blocks that benefit most from these techniques while skipping blocks where they would be redundant or inefficient
Solution Approach 2:
The patent implements partial application of BIO and DMVR techniques by evaluating cost criteria and applying the techniques only when the benefit justifies the cost. This partial action approach avoids the excessive computation that would result from applying these techniques to every block, thereby maintaining decoding efficiency while still achieving compression ratio improvements where needed
3Adaptability or versatility
If cost criterion evaluation is performed for each block, then technique selection is optimized, but processing time increases
Solution Approach 1:
The patent performs cost criterion evaluation and technique selection decisions during the encoding phase, preparing the information needed for optimal technique selection in advance. This preliminary action during encoding allows the decoder to simply follow the predetermined decisions without performing complex evaluations, thereby avoiding additional processing time at decoding
Data Source
AI summary
A method for processing a video includes performing a conversion between a current block of visual media data and a corresponding coded representation of the visual media data. The conversion of the current block includes determining whether a use of one or both of a bi-directional optical flow (BIO) technique or a decoder-side motion vector refinement (DMVR) technique to the current block is enabled or disabled. Determining the use of the BIO technique or the DMVR technique is based on a cost criterion associated with the current block.