Fixed-Filtered Block Decoding With Low-Complexity Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies due to the use of fixed filters with high resource utilization for parameter signaling and adaptive filters, leading to sub-optimal coding accuracy and increased bit cost.
Innovation Solution
Implementing low-complexity filtering of fixed-filtered data by signaling parameters for a low-complexity filter applied to data previously filtered using a fixed filter, reducing redundancy and bit cost while improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If fixed filters with parameter signaling are used, then filtering capability is provided, but resource utilization increases due to parameter signaling overhead
Solution Approach 1:
The patent extracts only the essential low-complexity filter parameters from the encoded bitstream, omitting redundant parameter data. This selective extraction maintains necessary filtering capability while reducing resource utilization by eliminating unnecessary parameter signaling overhead.
Solution Approach 2:
The patent changes the parameter representation by using a simplified parameter structure for low-complexity filters. Instead of signaling complete filter parameters, it uses compact representations that reduce bit cost while maintaining adequate filtering performance for the specific application.
2Measurement precision
If adaptive filters are used, then coding accuracy improves, but bit cost increases
Solution Approach 1:
The patent applies different filtering strategies to different regions or blocks based on their specific characteristics. Low-complexity filters are applied where sufficient, while more sophisticated filtering is used only where genuinely needed, optimizing the balance between coding accuracy and bit cost locally rather than uniformly.
Solution Approach 2:
The patent applies low-complexity filtering as a sufficient partial solution for many cases, recognizing that full adaptive filtering is excessive for numerous blocks. This partial action approach achieves acceptable coding accuracy for the majority of cases without incurring the high bit cost of complete adaptive filtering everywhere.
3Reliability
If redundant parameter data is signaled, then filter configuration is complete, but bitstream size increases
Solution Approach 1:
The patent extracts and signals only the critical parameter data necessary for filter configuration, removing redundant parameter information from the bitstream. This selective signaling maintains reliable filter configuration while reducing bitstream size by eliminating unnecessary data.
Solution Approach 2:
The patent uses universal default parameter values that can be applied across multiple blocks without explicit signaling. These default parameters provide a baseline configuration that works for many cases, reducing the need to signal parameters individually and thereby reducing overall bitstream size while maintaining configuration reliability.
Data Source
AI summary
Decoding using low-complexity filtering of fixed-filtered data includes obtaining reconstructed block data for a current block of a current frame by decoding encoded block data from an encoded bitstream, obtaining filtered reconstructed block data for the current block, and outputting the filtered reconstructed block data. Obtaining the filtered reconstructed block data includes obtaining a first filter for the current block, wherein obtaining the first filter omits accessing parameter data for the first filter from a portion of the encoded bitstream corresponding to the current frame, obtaining first filtered reconstructed block data for the current block by filtering the reconstructed block data using the first filter, accessing, from the portion of the encoded bitstream corresponding to the current frame, low-complexity filtering data for a low-complexity filter, and obtaining second filtered reconstructed block data for the current block by filtering the first filtered reconstructed block data using the low-complexity filter.