Block-Level Pipeline for Super-Resolution and Loop Restoration Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods employ frame-level pipeline architectures, resulting in high processing latency and memory bandwidth requirements due to sequential processing of frames, which limits efficiency in reducing artifacts and optimizing buffer usage.
Innovation Solution
Implementing a block-level pipeline architecture that includes a super-resolution filter and a loop restoration filter, where the loop restoration filter starts processing before the super-resolution filter has finished filtering all pixels, reducing latency and memory bandwidth by allowing parallel processing and efficient buffer management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a frame-level pipeline architecture is used for in-loop filtering, then filtering quality can be maintained, but processing latency and buffer requirements increase significantly
Solution Approach 1:
The patent divides the frame into multiple blocks and processes each block independently through the filtering pipeline. The frame-level filtering operations are segmented into block-level operations, allowing parallel processing of multiple blocks simultaneously. This segmentation enables the LR filter and SR filter to operate on different blocks in parallel, reducing overall processing latency while maintaining filtering quality.
Solution Approach 2:
The patent transitions from a sequential frame-level processing approach to a parallel block-level processing approach by adding a spatial dimension to the pipeline architecture. Multiple filtering pipelines are instantiated for different blocks, enabling concurrent processing. This dimensional change from 1D sequential to 2D parallel processing reduces latency without compromising filtering quality.
2Manufacturing precision
If a frame-level pipeline architecture is used for in-loop filtering, then filtering quality can be maintained, but memory bandwidth requirements and buffer sizes increase
Solution Approach 1:
By segmenting the frame into blocks and processing them independently, the patent reduces the amount of data that needs to be buffered at any given time. Instead of buffering entire frames for sequential processing, only the current and adjacent blocks need to be in the buffer, significantly reducing memory bandwidth requirements and buffer size while maintaining filtering quality.
Solution Approach 2:
The patent performs filtering operations on blocks as they become available in the decoding pipeline, rather than waiting for complete frame processing. This preliminary action on individual blocks allows the system to proceed with filtering before all frame data is available, reducing buffer requirements and memory bandwidth consumption.
3Device complexity
If sequential frame-level processing is used, then implementation simplicity is maintained, but processing efficiency and throughput decrease
Solution Approach 1:
The patent segments the filtering process into independent block-level operations that can be executed in parallel. While this increases implementation complexity compared to sequential frame-level processing, it dramatically improves processing efficiency and throughput. The segmented approach allows multiple filtering pipelines to operate concurrently, achieving higher productivity.
Solution Approach 2:
The patent merges multiple filtering operations into a unified block-level pipeline architecture where LR filtering and SR filtering are combined and executed efficiently. By merging the filtering stages and enabling parallel block processing, the system achieves both improved productivity and managed complexity through systematic integration.
Data Source
AI summary
A post processing apparatus includes a super-resolution (SR) filtering circuit and a loop restoration (LR) filtering circuit. The SR filtering circuit applies SR filtering to a processing result of a preceding circuit. The LR filtering circuit applies LR filtering to a processing result of the SR filtering circuit. Before the SR filtering circuit finishes SR filtering of all pixels of a frame that are generated by the preceding circuit, the LR filtering circuit starts LR filtering of pixels that are derived from applying SR filtering to pixels included in the frame.