Video Signal Encoding Using Adaptive Bidirectional Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression technologies, such as HEVC, face limitations in efficiently encoding and decoding high-definition video signals due to increased data amounts, necessitating improved methods for determining partitioning types, using block vectors, and adaptively applying bidirectional optical flow.
Innovation Solution
A video signal encoding/decoding method that derives motion vectors, prediction samples, and determines the application of bidirectional optical flow based on chroma block size, block vector differences, and reference picture directions to optimize encoding/decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If High Efficiency Video Coding (HEVC) is used to improve compression performance, then compression rate is improved (about twice of H.264/AVC), but data amount is greatly increased for high-definition video services
Solution Approach 1:
The video signal is divided into multiple blocks (current block, reference blocks) and further segmented into sub-blocks for motion estimation. The patent applies bidirectional optical flow by segmenting the block into sub-blocks and performing separate motion estimation for each sub-block, allowing more precise motion compensation that improves compression efficiency without proportionally increasing data amount.
Solution Approach 2:
The patent changes the parameter of motion estimation precision by applying bidirectional optical flow at sub-block level instead of whole block level. This parameter change (from block-level to sub-block-level estimation) improves prediction accuracy and compression performance while the selective application (based on block size thresholds) controls the increase in data amount.
2Device complexity
If block size is increased to reduce the number of blocks, then processing complexity is reduced, but motion estimation accuracy deteriorates
Solution Approach 1:
The patent dynamically adjusts the motion estimation approach based on block size. For large blocks (meeting the threshold condition), bidirectional optical flow is applied at sub-block level for high accuracy. For smaller blocks, the method adapts by using coarser estimation or fewer sub-blocks, thus reducing complexity. This dynamic adjustment resolves the contradiction between block size and accuracy.
Solution Approach 2:
The patent applies different levels of motion estimation quality to different regions. By dividing large blocks into sub-blocks and applying bidirectional optical flow selectively, each sub-block receives appropriate estimation precision. This local quality approach ensures high accuracy where needed (large blocks with complex motion) while reducing complexity for smaller or simpler regions.
3Measurement precision
If bidirectional optical flow is applied to all blocks, then motion compensation accuracy is improved, but encoding complexity and data amount increase
Solution Approach 1:
The patent applies bidirectional optical flow partially rather than universally. By using a threshold condition on block size, the method applies the computationally intensive bidirectional optical flow only to blocks that meet the criteria (larger blocks where it provides most benefit). This partial action improves motion compensation accuracy for critical blocks while controlling overall encoding complexity and data amount.
Data Source
AI summary
A video decoding method, according to the present invention, comprises the steps of: deriving an L0 motion vector and an L1 motion vector of a current block; deriving an L0 prediction sample and an L1 prediction sample for a first position inside the current block on the basis of the L0 motion vector and the L1 motion vector; determining whether to apply a bidirectional optical flow to the current block; deriving a refined motion vector for a subblock which comprises the first position, when it is determined to apply the bidirectional optical flow; deriving a refined prediction sample for the first position on the basis of the refined motion vector; and obtaining a prediction sample for the first position by using the L0 prediction sample, the L1 prediction sample, and the refined prediction sample.