Adaptive Upsampling Filters for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems lack flexibility and performance in upsampling video data, particularly when matching the spatial aspect ratio of base and enhancement layers, due to the limited use of a single set of upsampling filters for luma and chroma components.
Innovation Solution
The proposed solution involves an adaptive upsampling process that applies multiple sets of upsampling filters with additional phase shifts and spatial pixel displacement, allowing for fractional accuracy and selective application of filter sets based on enhancement layer information, such as prediction mode and block size, to enhance flexibility and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single set of upsampling filters is used for luma and chroma components, then the device complexity is reduced, but the adaptability and performance in matching spatial aspect ratios between base and enhancement layers deteriorates
Solution Approach 1:
The patent divides the single filter set into multiple separate filter sets, each optimized for specific prediction modes and block sizes. This segmentation allows the system to select appropriate filters for different video content characteristics, improving adaptability while maintaining manageable complexity through organized filter categories.
Solution Approach 2:
The patent introduces dynamic filter selection based on prediction mode and block size parameters. The system adaptively chooses from multiple filter sets depending on the specific coding context, enabling the upsampling process to dynamically adjust to different video content requirements rather than using a static single filter set.
2Adaptability or versatility
If multiple sets of upsampling filters with additional phase shifts are applied, then the upsampling performance and flexibility are improved, but the device complexity increases
Solution Approach 1:
The patent applies different filter characteristics to different local regions of the video data based on prediction mode and block size. Each filter set is tailored with specific phase shifts and coefficients optimized for particular coding scenarios, allowing local optimization of upsampling quality without requiring complete redesign of the entire filtering system.
Solution Approach 2:
The patent modifies filter parameters such as phase shifts and tap coefficients based on prediction mode and block size conditions. By changing these parameters adaptively, the system achieves multiple filter behaviors from a structured set of related filters, improving flexibility while controlling complexity through parameter variation rather than completely independent filter designs.
3Productivity
If adaptive filter selection based on enhancement layer information is implemented, then the coding efficiency is improved, but the processing complexity increases
Solution Approach 1:
The patent pre-organizes multiple filter sets with predetermined characteristics and phase shifts corresponding to different prediction modes and block sizes. This preliminary organization allows the decoding process to efficiently select appropriate filters by simple parameter matching rather than performing complex optimization calculations in real-time, improving coding efficiency while keeping processing complexity manageable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In one embodiment, a video coder for coding video data includes a processor and a memory. The processor selects a filter set from a multiple filter sets for upsampling reference layer video data based at least on a prediction operation mode for enhanced layer video data and upsamples the reference layer video data using the selected filter set. Some of the multiple filter sets have some different filter characteristics from one another, and the upsampled reference layer video data has the same spatial resolution as the enhanced layer video data. The processor further codes the enhanced layer video data based at least on the upsampled reference layer video data and the prediction operation mode. The memory stores the upsampled reference layer video data.