Adaptive Loop Filtering for Video Coding Using LCU Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face challenges in achieving an optimal tradeoff between reconstructed video quality and compression quality, particularly in efficiently filtering video blocks during the encoding and decoding processes.
Innovation Solution
The implementation of adaptive loop filtering (ALF) techniques, where filters are applied at the encoder and encoded in the bitstream, allowing the decoder to apply the same filtering, using methods such as block adaptive and region adaptive classifications to select filters based on activity metrics and region divisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If adaptive loop filtering is applied to improve reconstructed video quality, then the filtering accuracy and visual quality improve, but the computational complexity and processing time increase
Solution Approach 1:
The video data is divided into multiple largest coding units (LCUs), and each LCU is further divided into sub-LCUs. Different filtering operations can be applied to different sub-LCUs based on their local characteristics, allowing the system to process complex video content in manageable segments rather than applying uniform filtering to the entire frame.
Solution Approach 2:
The patent implements block adaptive filtering and region adaptive filtering where different filter types (e.g., strong filtering, weak filtering, no filtering) are applied to different regions of the video based on local activity metrics. This allows high filtering accuracy in regions that need it while reducing or skipping filtering in regions where it would be unnecessary, thus improving overall filtering accuracy while managing computational complexity.
2Manufacturing precision
If multiple filtering operations are tested and selected based on rate-distortion analysis, then the reconstructed video quality improves, but the encoding complexity and processing time increase
Solution Approach 1:
The patent performs rate-distortion analysis and selects the optimal filtering operation for each LCU during the encoding phase. By determining and storing the optimal filter selection in advance, the system avoids performing multiple filtering operations during decoding, significantly reducing processing time while maintaining high reconstructed video quality.
Solution Approach 2:
The patent encodes filter information in the bitstream so that the decoder can replicate the exact same filtering operations that were applied during encoding. This copying approach ensures that the reconstructed video quality matches the original encoded quality without requiring the decoder to perform complex analysis or multiple filtering trials.
3Stability of the object's composition
If filter information is encoded in the bitstream for decoder to apply the same filtering, then the consistency between encoder and decoder improves, but the bitstream size increases
Solution Approach 1:
The patent uses a unified filtering framework where the same set of filtering operations is used for both encoding and decoding. The filter information encoded in the bitstream serves multiple purposes: it specifies which filters to apply, indicates whether filtering should be applied at all, and ensures consistency between encoder and decoder without requiring separate signaling mechanisms.
Solution Approach 2:
The patent varies the filtering parameters (such as filter strength, filter type, and whether to apply filtering) based on local video characteristics like activity metrics. By adapting the filter parameters to local regions rather than using fixed parameters for the entire frame, the system achieves consistent filtering results while minimizing the amount of information that needs to be encoded in the bitstream.
4Measurement precision
If block adaptive and region adaptive classifications are used to select filters, then the filtering precision improves, but the device complexity and processing overhead increase
Solution Approach 1:
The patent implements dynamic filter selection where the filtering operation for each LCU is determined based on local video characteristics such as activity metrics. The system can dynamically switch between different filtering modes (strong filtering, weak filtering, no filtering) and adjust filter parameters based on the local content, achieving high filter selection precision while adapting to varying video conditions.
Solution Approach 2:
The patent uses local video content characteristics (such as activity metrics calculated from the video data itself) to automatically determine the appropriate filtering operation for each region. This self-service approach allows the system to make intelligent filter selection decisions based on the video content without requiring external guidance or complex pre-processing, thereby improving precision while managing processing overhead.
Data Source
AI summary
Video encoders and video decoders filter reconstructed video data in a video encoding and/or video decoding processes and, more particularly, may perform adaptive loop filtering. Instead of implementing region adaptive (RA) classification using a fixed set of regions, a variable number of regions in a picture can be used. Additionally or alternatively, a global set of filters can be signaled in a parameter set from an encoder to a decoder, but instead of selecting filters from the global set based on an RA classification or a block adaptive (BA) classification, a filter from the global set can be explicitly signaled, for a given block, using a filter ID associated with the filter.