OBMC Blending Selection by Template Matching for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques suffer from blockiness artifacts due to lossy transform and quantization processes, which can be mitigated by improving inter prediction methods, particularly through enhanced overlapped block motion compensation (OBMC) blending processes.
Innovation Solution
The implementation of OBMC blending techniques that select an optimal blending process based on template matching costs between current and neighboring blocks, allowing for varying amounts of blending without increasing signaling overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional inter prediction methods are used, then encoding complexity is low, but blockiness artifacts occur due to lossy transform and quantization processes
Solution Approach 1:
The current block is divided into multiple sub-blocks for independent motion compensation processing. Each sub-block uses its own motion vector to predict from reference blocks, allowing localized optimization and reducing the propagation of artifacts across the entire block while maintaining manageable complexity per sub-block.
Solution Approach 2:
Different blending weights are applied to different spatial locations within the block based on local characteristics. The OBMC technique uses position-dependent weighting where reference blocks closer to block boundaries receive higher blending weights, improving local prediction quality at boundaries while maintaining efficiency elsewhere.
2Manufacturing precision
If OBMC blending processes are applied to improve prediction accuracy, then blockiness artifacts are reduced, but signaling overhead increases
Solution Approach 1:
OBMC blending is applied selectively rather than uniformly across all blocks. The technique determines whether OBMC should be applied based on block characteristics and motion complexity, applying the blending process only where it provides significant benefit, thus avoiding unnecessary signaling overhead while maintaining prediction accuracy where needed.
Solution Approach 2:
The blending weights in OBMC are derived from motion vector differences and spatial positions rather than being explicitly signaled. By changing the parameter representation from explicit weight values to derived weights based on motion information, the system maintains high prediction accuracy while minimizing signaling overhead.
3Manufacturing precision
If multiple OBMC blending processes are available for selection, then prediction quality is improved by varying blending amounts, but processing complexity increases
Solution Approach 1:
The OBMC blending process dynamically adjusts blending weights based on local motion characteristics and spatial position. Rather than using fixed blending parameters, the system computes weights that adapt to the specific block being processed, improving prediction quality for diverse motion patterns while using a unified algorithmic framework that avoids the complexity of maintaining multiple separate processes.
Data Source
AI summary
A device for decoding video data can be configured to determine that a block is coded in an overlapped block motion compensation (OBMC) mode; determine a first prediction block for the block using a first motion vector for the block; determine a second prediction block for the block using a second motion vector, wherein the second motion is a motion vector of a neighboring block of the block; select an OBMC blending process from a plurality of available OBMC blending processes; blend the first prediction block and the second prediction block using the selected OBMC blending process to determine a blended block; and reconstruct the block using the blended block.