Combined Loop Filtering for Video Compression Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques face inefficiencies due to the implementation challenges of multiple loop filter stages, which can lead to latency and increased hardware complexity, while also compromising compression efficiency when trying to minimize these issues.
Innovation Solution
Combining bilateral loop filtering or Hadamard loop filtering with Sample Adaptive Offset (SAO) or Adaptive Loop Filtering (ALF) to reduce the number of loop filter stages, allowing these filters to operate in parallel and maintain compression efficiency gains.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If multiple loop filter stages are implemented separately, then compression efficiency is improved, but device complexity and latency increase
Solution Approach 1:
The patent combines multiple loop filter stages (deblocking filter, SAO filter, and ALF) into a unified filtering pipeline where filters operate in sequence but share common infrastructure. The combining loop filter merges bilateral loop filtering with SAO or ALF, allowing them to process data through a unified structure rather than completely separate stages, thus maintaining compression efficiency while reducing overall hardware complexity and latency.
Solution Approach 2:
The filtering process is divided into distinct but integrated stages: deblocking filter stage, SAO filter stage, and ALF stage. Each stage processes the output of the previous stage in a segmented manner, allowing independent optimization of each filter type while maintaining overall system efficiency. This segmentation enables parallel processing within each stage while keeping inter-stage dependencies manageable.
2Loss of energy
If multiple loop filter stages are implemented separately, then compression efficiency is improved, but processing speed decreases
Solution Approach 1:
The deblocking filter is applied as a preliminary step before SAO and ALF filtering stages. By performing deblocking first, the patent prepares the data in advance for subsequent filtering operations, reducing the computational burden on later stages and improving overall processing speed. This preliminary action ensures that blocking artifacts are addressed before more complex filtering is applied.
Solution Approach 2:
The patent implements a continuous filtering pipeline where the output of one filter stage immediately becomes the input of the next stage without interruption. The deblocking filter feeds continuously into SAO, which feeds into ALF, maintaining a steady flow of data processing. This continuity eliminates idle time between stages and maximizes processing throughput while preserving compression efficiency.
3Device complexity
If loop filter stages are minimized, then device complexity is reduced, but compression efficiency deteriorates
Solution Approach 1:
The patent creates a universal filtering framework that can accommodate multiple filter types (deblocking, SAO, ALF) within a single integrated structure. The combining loop filter serves multiple functions by merging bilateral filtering with SAO or ALF, allowing the system to maintain comprehensive filtering capability while using a unified hardware architecture rather than completely separate filter units.
Solution Approach 2:
The patent introduces intermediate filtering stages and combined loop filter structures that act as mediators between the deblocking filter and final output. These intermediate structures preserve the beneficial effects of multiple filtering approaches while reducing the need for fully separate, complex filter implementations, thus maintaining compression efficiency with reduced hardware complexity.
4Adaptability or versatility
If filters operate at different speeds, then processing flexibility is improved, but decoder drift occurs
Solution Approach 1:
The patent implements feedback mechanisms within the filtering pipeline that monitor and adjust processing to prevent decoder drift. By incorporating feedback loops that track filtering outcomes and adjust subsequent processing accordingly, the system maintains reliability even when filters operate at different speeds. The feedback ensures that variations in processing speed do not accumulate into drift errors.
Solution Approach 2:
The patent creates a dynamic filtering system where the processing speed and intensity of each filter stage can be adjusted based on input characteristics and system conditions. This dynamic approach allows filters to operate at optimal speeds for their specific functions while maintaining synchronization through adaptive control mechanisms, preventing decoder drift while preserving processing flexibility.
Data Source
AI summary
In an image processing device (i.e. encoder or decoder), the number of loop filter stages is lowered by combining bilateral loop filtering (or Hadamard loop filtering) with either sample Adaptive Offset Filtering (SAO) or Adaptive Loop Filtering (ALF). This avoids the implementation problems associated with too many loop filter stages and provides approximately the same compression efficiency gain as having separate loop filter stages.