Region-Based In-Loop Filtering for Low-Latency Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding standards like HEVC introduce delays and increased memory requirements due to picture-based processing of sample adaptive offset (SAO) and adaptive loop filter (ALF) parameters, which are not optimized for low-latency applications such as video conferencing or cloud computing.
Innovation Solution
Implement region-based SAO and ALF parameter determination and signaling techniques that allow parallel processing and interleaving of parameters with region data, reducing encoder delay and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If picture-based processing is used for SAO and ALF parameters, then filtering quality is improved, but encoder delay and memory requirements increase
Solution Approach 1:
The patent divides the picture into multiple regions (e.g., slices or coding tree units) and processes SAO/ALF parameters independently for each region. This segmentation allows parallel processing of different regions, reducing the encoder delay while maintaining filtering quality through region-specific parameter determination.
2Manufacturing precision
If picture-based processing is used for SAO and ALF parameters, then filtering quality is improved, but memory requirements increase
Solution Approach 1:
By segmenting the picture into regions and processing parameters locally for each region, the patent reduces the memory footprint compared to picture-based processing. Region-specific parameter storage requirements are significantly smaller than storing parameters for the entire picture.
Solution Approach 2:
The patent performs preliminary determination of SAO/ALF parameters for each region before actual filtering application. This allows parameters to be computed and stored efficiently, reducing the need for large memory buffers during the filtering process.
3Loss of time
If region-based processing is implemented, then encoder delay is reduced, but parameter determination complexity increases
Solution Approach 1:
The patent segments the parameter determination process into independent region-level operations. Each region can be processed in parallel, and the complexity is distributed across multiple smaller operations rather than one large complex operation, making it more manageable and implementable.
Data Source
AI summary
A method for in-loop filtering in a video encoder is provided that includes determining filter parameters for each filtering region of a plurality of filtering regions of a reconstructed picture, applying in-loop filtering to each filtering region according to the filter parameters determined for the filtering region, and signaling the filter parameters for each filtering region in an encoded video bit stream, wherein the filter parameters for each filtering region are signaled after encoded data of a final largest coding unit (LCU) in the filtering region, wherein the in-loop filtering is selected from a group consisting of adaptive loop filtering and sample adaptive offset filtering.