Variable M×N Sample Pattern for Video Deblocking Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in deblocking filtering due to the use of fixed sample patterns, which can lead to inaccurate estimation of local luma levels and misalignment with other parameters, particularly for chroma components and longer deblocking filters.
Innovation Solution
Implementing a variable M×N sample pattern based on block size to estimate local luma levels, allowing for improved computational efficiency with minimal loss in coding accuracy, and adapting deblocking filter strengths accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed 4×4 sample pattern is used for estimating local luma level, then the deblocking filtering process is simple and consistent, but the estimation accuracy deteriorates for large blocks and chroma components
Solution Approach 1:
The sample pattern size is made dynamic rather than fixed. The patent applies different M×N sample pattern sizes (e.g., 4×4, 8×8, 16×16) depending on the block size being processed. For larger blocks, larger sample patterns are used to capture sufficient local variation, while for smaller blocks, smaller patterns are used to avoid excessive smoothing. This dynamic adaptation resolves the contradiction by making the pattern size contingent on the specific processing context.
Solution Approach 2:
Different sample patterns are applied to different regions or block types within the video data. The patent specifically applies different M×N patterns to luma and chroma components, and different patterns to different block sizes. This local differentiation ensures that each region receives the appropriate level of detail in estimation, improving accuracy without uniformly increasing complexity across the entire processing system.
2Manufacturing precision
If a fixed sample pattern is used regardless of block size, then the processing is computationally efficient, but the coding accuracy deteriorates for large blocks
Solution Approach 1:
The sample pattern size dynamically adapts to block size. Larger blocks use larger M×N sample patterns (e.g., 16×16) to maintain coding accuracy, while smaller blocks use smaller patterns (e.g., 4×4) to preserve computational efficiency. This dynamic scaling ensures that computational resources are allocated proportionally to the actual processing needs of each block.
Solution Approach 2:
The patent changes the parameter of sample pattern size based on block size. By adjusting the M and N dimensions of the sample pattern according to the block dimensions being processed, the system optimizes the balance between accuracy and efficiency for each specific case rather than using a one-size-fits-all approach.
3Reliability
If deblocking filtering is applied uniformly to all blocks, then the implementation is straightforward, but the filtering strength misaligns with local signal characteristics
Solution Approach 1:
The deblocking filtering process is localized to match block characteristics. Different filtering strengths and sample patterns are applied to different block sizes and component types (luma vs. chroma). This ensures that filtering effectiveness is optimized for each local region's signal characteristics rather than applying a uniform approach that may over-smooth or under-smooth certain areas.
Solution Approach 2:
The filtering parameters (sample pattern size M×N, filtering strength) are changed based on local signal characteristics such as block size and component type. This parameter adaptation allows the filtering process to reliably achieve appropriate smoothing for each case while maintaining a systematic approach that doesn't excessively complicate the overall implementation.
Data Source
AI summary
A method of decoding video data includes generating a prediction block for a current block of the video data and decoding a residual block for the current block of the video data. The method includes combining the prediction block and the residual block to generate a reconstructed block of the video data and determining a variable M×N sample pattern based on a block size of the current block. The method includes estimating a local luma level for samples of the reconstructed block using the variable M×N sample pattern and applying, based on the estimated local luma level, a deblocking filter to the samples of the reconstructed block to generate filtered samples. The method includes generating final samples of the video data based on the filtered samples.