Adaptive Offset Filter LCU Processing for HEVC Buffer Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In HEVC, the adaptive offset filter parameters are collectively defined in sao_param() for one frame, requiring data to be held in a buffer until the completion of the adaptive offset filter process, which hinders processing efficiency in encoding and decoding.
Innovation Solution
An image processing device acquires adaptive offset filter parameters in units of a largest coding unit from an encoded stream and performs adaptive offset filtering on the decoded image, allowing for efficient processing by reducing the need for a large buffer and enabling concurrent filtering with the adaptive loop filter.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If adaptive offset filter parameters are collectively defined in sao_param() for one frame, then the filter can be applied to the entire frame, but data must be held in a buffer until completion of the adaptive offset filter process, reducing processing efficiency
Solution Approach 1:
The patent divides the frame into multiple largest coding units (LCUs) and processes the adaptive offset filter parameters in units of each LCU rather than for the entire frame. This segmentation allows the decoder to process and apply filter parameters incrementally as each LCU is decoded, eliminating the need to hold entire frame data in buffer and enabling continuous processing without waiting for frame-level parameter completion.
2Reliability
If adaptive offset filter parameters are sent for the entire frame, then comprehensive filtering can be applied, but the amount of data to be buffered and processed increases, complicating the decoding process
Solution Approach 1:
The patent segments both the image data and the filter parameters into LCU-level units. Each LCU has its own associated filter parameters that are independently decoded and applied. This segmentation reduces buffer management complexity by allowing the decoder to handle smaller, manageable units of data and parameters simultaneously, rather than managing large frame-level buffers.
Solution Approach 2:
The adaptive offset filter parameters for each LCU are decoded and prepared in advance before the actual filtering operation. This preliminary action allows the parameters to be ready immediately when each LCU is decoded, enabling continuous processing without delays and simplifying the timing coordination between parameter decoding and filter application.
3Stability of the object's composition
If adaptive offset filtering is performed after deblocking filtering for the entire frame, then the filtering order is maintained, but processing latency increases due to sequential processing requirements
Solution Approach 1:
The patent segments the filtering process into LCU-level operations where deblocking filtering and adaptive offset filtering are performed in sequence for each individual LCU. This segmentation enables pipelined processing where different LCUs can be at different stages of the filtering process simultaneously, maintaining the correct filtering order within each LCU while reducing overall processing latency through parallel processing of multiple LCUs.
Data Source
AI summary
Aspects of the present disclosure provide an image processing device that includes circuitry configured to perform a decoding process on an encoded stream for generating an image. The encoded stream is arranged into largest coding units (LCUs), and filtering parameters applicable to each LCU are provided at a beginning portion of the corresponding LCU. The circuitry is further configured to perform adaptive offset filtering on portions of the image that correspond to the LCUs using the filtering parameters set at the beginning portions of the LCUs, respectively.