In-Loop Adaptive Filter Signaling Through Predictive Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies face challenges in efficiently signaling and applying adaptive loop filters (ALFs) due to high syntax costs, which affect compression efficiency and artifact reduction.
Innovation Solution
Implementing methods and devices that encode/decode a mapping between classes and filter indices using predictive coding, and arithmetically decoding filter indices based on neighboring blocks to reduce syntax overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional signaling methods are used for adaptive loop filters, then filter application can be achieved, but syntax overhead is high
Solution Approach 1:
The patent applies preliminary action by constructing a candidate list of filter indices from neighboring blocks before encoding the current block's filter index. This pre-computation of candidates based on spatial correlation allows the encoder to select from a limited set of likely values, reducing the number of bits needed to signal the filter index while ensuring accurate reconstruction at the decoder side.
Solution Approach 2:
The patent changes the parameter representation by transitioning from direct signaling of filter indices to differential coding based on a candidate list. Instead of encoding absolute filter indices, the system encodes differences or selections from pre-determined candidates, thereby reducing the syntax overhead while maintaining the ability to represent all necessary filter configurations.
2Manufacturing precision
If more filter indices are signaled explicitly, then filtering accuracy is improved, but bitstream size increases
Solution Approach 1:
The patent applies local quality by adapting the filter index selection to local characteristics of neighboring blocks. Instead of using a uniform signaling approach for all blocks, the system constructs candidate lists based on the specific filter indices of adjacent blocks, allowing each block to be processed with locally optimized parameters that reflect the local image structure and correlation properties.
Solution Approach 2:
By pre-constructing candidate lists from neighboring block information before encoding the current block, the system ensures that the most probable filter indices are available for selection. This preliminary action based on spatial correlation allows accurate filtering to be achieved with fewer bits, as the candidate list is tailored to the local context of each block.
3Object-affected harmful factors
If adaptive loop filters are applied to all blocks, then artifact reduction is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by selectively applying adaptive loop filtering only to blocks where it is most beneficial, rather than uniformly to all blocks. By using candidate lists derived from neighboring blocks and differential coding, the system can identify and process only those blocks that require filtering, reducing computational complexity while maintaining effective artifact reduction in critical areas.
Solution Approach 2:
The system changes the approach from universal filter application to selective filter application by introducing a candidate list mechanism. This allows the encoder to determine, based on local characteristics and neighboring block information, which blocks warrant filtering and with which specific filter indices, thereby reducing overall computational complexity while maintaining effective artifact reduction where needed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In various implementations, methods and devices are disclosed that encode/decode a mapping between a class and a filter index, for instance for signaling ALF usage information with predictive coding. Besides, methods and devices are disclosed that encode/decode a filter index of a set of filters, for instance for signaling ALF usage information with predictive coding.