Template-Based Video Block Reconstruction Under Latency Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face challenges in efficiently managing latency constraints during template-based operations, particularly in video decoding and encoding processes, leading to inefficiencies in reconstruction and encoding of video blocks.
Innovation Solution
A device performs fast template reconstruction by obtaining unrefined motion information for a first block, partially reconstructing it, and using this information to reconstruct a second block independently or in parallel with the full reconstruction of the first block, employing techniques like adaptive reordering of motion vector candidates and template-based coding tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template-based operations are performed using refined motion information, then reconstruction accuracy is improved, but processing latency increases
Solution Approach 1:
The patent applies preliminary action by performing template-based operations using unrefined motion information before the motion information is fully refined. This allows the system to proceed with reconstruction tasks that would otherwise be blocked until refined motion data is available, thereby reducing latency while maintaining the ability to achieve high accuracy through subsequent refinement steps
Solution Approach 2:
The patent segments the motion information refinement process into distinct stages: unrefined motion information is used for initial template-based operations, and refined motion information is applied for subsequent reconstruction. This segmentation allows different levels of precision to be applied at different stages, optimizing both latency and accuracy
2Manufacturing precision
If blocks are reconstructed sequentially using refined motion information, then reconstruction quality is improved, but processing efficiency deteriorates
Solution Approach 1:
The patent enables parallel processing by performing template-based operations on different blocks simultaneously using unrefined motion information. This preliminary action allows multiple blocks to be processed in parallel without waiting for motion refinement, significantly improving processing efficiency while maintaining reconstruction quality through subsequent refinement steps
Solution Approach 2:
The patent divides the reconstruction process into independent segments that can be processed in parallel. Different blocks can use template-based operations with unrefined motion information independently, and the refinement process is segmented to update motion information for specific blocks without blocking others, thereby improving overall processing efficiency
Data Source
AI summary
Systems, methods, and instrumentalities are disclosed for latency constrained template-based operations. A device (e.g., video decoding device, video encoding device) may perform fast template reconstruction. The device may obtain motion information associated with a first block. The device may obtain a motion vector candidate list for the first block. The motion information may be determined based on the motion vector candidate list. The device may partially reconstruct the first block based on the motion information associated with the first block. The motion information used to partially reconstruct the first block may be unrefined motion information. The device may obtain a template of a second block based on the partially reconstructed first block. The device may reconstruct the second block using a template-based coding tool. The reconstruction of the second block may be performed before or in parallel (e.g., independent) with reconstructing the first block using refined motion information.