Video Coding With Wavefront Parallelism to Reduce Processing Time
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high efficient video coding (HEVC), the sequential coding of LCUs results in low resource utilization and long data processing times due to the need to complete processing of one LCU before proceeding to the next, leading to waste of computing and storage resources.
Innovation Solution
The method involves partitioning an image into coding tree units (CTUs) and distributing coding tasks of coding blocks into multiple task sequences, which are executed in parallel using wavefront parallel processing (WPP) to optimize resource utilization and improve coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sequential coding of LCUs is used, then coding quality is maintained through complete processing of each LCU, but resource utilization is low and processing time is long
Solution Approach 1:
The patent divides the image into multiple CTU rows and further segments coding tasks into independent units that can be processed in parallel. Each CTU row can be coded independently, and within each row, multiple CUs can be processed simultaneously using wavefront parallel processing, thereby improving resource utilization while maintaining coding quality through structured segmentation.
Solution Approach 2:
The patent transitions from sequential processing (one-dimensional time sequence) to parallel processing across multiple dimensions. By introducing wavefront parallel processing that operates on multiple CTU rows simultaneously and uses task sequences with dependency management, the system achieves multi-dimensional parallelism that improves productivity without compromising coding quality.
2Reliability
If sequential coding of LCUs is used, then coding quality is maintained, but data processing time is long
Solution Approach 1:
The patent segments the coding process into independent CTU rows and task sequences that can be executed in parallel. This segmentation allows multiple coding operations to proceed simultaneously, significantly reducing total processing time while maintaining coding quality through structured processing of segmented units.
Solution Approach 2:
The patent implements continuous parallel processing where multiple CTU rows are processed simultaneously and task sequences are executed in parallel. This continuous useful action eliminates idle waiting time between sequential operations, reducing overall processing time while maintaining quality through coordinated parallel execution.
3Device complexity
If sequential coding of LCUs is used, then processing is simple, but computing and storage resources are wasted
Solution Approach 1:
The patent introduces dynamic task sequence management where the processing system adapts to data dependencies dynamically. The wavefront parallel processing framework with task sequences allows the system to manage complexity dynamically, allocating computing and storage resources efficiently across parallel operations, thereby reducing resource waste while handling the increased processing complexity through intelligent scheduling.
Data Source
AI summary
A video coding method and apparatus including acquiring an image, partitioning the image into a plurality of coding tree units (CTUs), each CTU including one or more coding units (CUs), distributing coding tasks of coding blocks of CUs in a plurality of CTU rows in the image into a plurality of task sequences, each task sequence being a sequence of coding tasks of a plurality of coding blocks, and executing the plurality of task sequences in parallel to obtain coding results for the coding blocks of the CUs in the plurality of CTU rows.