Tile-Independent Inter Prediction for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face efficiency losses due to tile dependency disruptions in block-based codecs, particularly in applications like 360° video services and multi-core computing, where maintaining tile-independence is crucial for encoding and decoding operations.
Innovation Solution
Implement a block-based video decoder and encoder that utilize motion-compensated prediction and bi-directional optical flow tools, with mechanisms to enforce tile-independence constraints by adapting motion information derivation and padding or deactivating tools as needed, ensuring that prediction blocks and patches remain within tile boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If motion-compensated prediction is used to improve coding efficiency, then bitrate usage is optimized, but tile-independence is disrupted due to motion vectors crossing tile boundaries
Solution Approach 1:
The picture is divided into multiple tiles that are coded independently. Motion compensation is performed separately for each tile, ensuring that motion vectors and prediction blocks do not cross tile boundaries. This segmentation allows parallel processing while maintaining coding efficiency through motion-compensated prediction within each tile.
Solution Approach 2:
Different coding approaches are applied to different regions: inside tile boundaries, full motion-compensated prediction is used for high coding efficiency; at tile boundaries, the prediction is constrained to maintain independence. This local differentiation optimizes both coding efficiency and tile-independence.
2Measurement precision
If bi-directional optical flow tool is activated to improve motion-compensated bi-directional prediction, then prediction accuracy is enhanced, but tile-independence is compromised when patches cross tile boundaries
Solution Approach 1:
The bi-directional optical flow tool is applied separately to each tile. Prediction patches are constrained to remain within tile boundaries, preventing cross-tile dependencies. This segmentation maintains prediction accuracy while preserving tile-independence for parallel processing.
Solution Approach 2:
The bi-directional optical flow tool is selectively activated only when prediction patches remain within tile boundaries. When patches would cross boundaries, the tool is deactivated or adjusted. This partial application maintains accuracy where possible while ensuring tile-independence.
3Device complexity
If encoder relies on decoder to enforce tile-independence constraints, then encoding complexity is reduced, but decoding robustness is compromised
Solution Approach 1:
The decoder is equipped with built-in mechanisms to automatically enforce tile-independence constraints. It independently verifies that motion vectors and prediction blocks adhere to tile boundaries, and performs boundary padding when necessary. This self-service approach at the decoder enhances robustness without significantly increasing encoder complexity.
Data Source
AI summary
Different concepts for improving video coding efficiency are described, many of them allow for video coding in a manner realizing tile-independent coding with reducing, however, the coding efficiency losses otherwise associated with the tile-dependency disruptions, with nevertheless merely marginally, if all, modifying the codec behavior alongside the tile boundaries.