Dyadic Spatial Re-sampling Filters for Scalable Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing dyadic spatial re-sampling filters in scalable video coding face challenges in balancing visual quality, power consumption, and memory usage, particularly due to complex phase calculations and additional low-pass filtering required for interpolation of sub-pixels in Extended Spatial Scalability (ESS) systems.
Innovation Solution
The implementation of dyadic spatial re-sampling filters using a Kaiser window function-based design, with specific tap values for down and up sampling filters, such as [−1, 17, 32, 17, −1]/64 for down sampling and [−5, 0, 21, 32, 21, 0, −5]/64 for up sampling, to simplify the filtering process and improve scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If Extended Spatial Scalability (ESS) is used for dyadic spatial re-sampling, then spatial scalability is improved, but phase calculations become complex and additional low-pass filtering is required
Solution Approach 1:
The patent extracts and eliminates the complex phase calculation component from the ESS re-sampling process by using a simplified filter bank approach. Instead of performing complex phase calculations for each sub-band, the invention uses a set of fixed filters that inherently handle the re-sampling operation, thereby reducing computational complexity while maintaining spatial scalability.
Solution Approach 2:
The patent changes the parameters of the re-sampling process by using a fixed filter bank with specific tap values (e.g., [−1, 17, 32, 17, −1]/64) instead of variable phase calculations. This parameterization approach allows the system to achieve dyadic spatial scalability through simple filter application rather than complex adaptive phase rotation.
2Manufacturing precision
If additional low-pass filtering is applied for sub-pixel interpolation in ESS, then interpolation accuracy is improved, but power consumption increases
Solution Approach 1:
The patent merges the low-pass filtering operation with the re-sampling filter application. Instead of applying separate low-pass filtering stages for sub-pixel interpolation, the invention combines these operations into a single filter bank application, thereby reducing the total number of operations and power consumption while maintaining interpolation accuracy.
Solution Approach 2:
The patent uses a partial filtering approach where the filter bank taps are designed to provide sufficient interpolation accuracy without requiring excessive filtering. The specific tap values (e.g., [−5, 0, 21, 32, 21, 0, −5]/64) are optimized to achieve the required precision with minimal computational effort, avoiding over-filtering and associated power costs.
3Adaptability or versatility
If conventional re-sampling filters are used for dyadic spatial scalability, then compatibility with existing systems is maintained, but visual quality and coding efficiency are insufficient
Solution Approach 1:
The patent changes the filter parameters to optimized values (e.g., [−1, 17, 32, 17, −1]/64 for downsampling and [−5, 0, 21, 32, 21, 0, −5]/64 for upsampling) that are specifically designed for dyadic spatial scalability. These parameter optimizations improve visual quality and coding efficiency while maintaining compatibility with the H.264/SVC framework through standard filter application procedures.
Data Source
AI summary
A dyadic spatial down sampling filter having tap values configured according to a Kaiser window function a beta factor of approximately 2.5, having approximately 1.5 side lobes, and having a down sampling ratio of approximately 1.9. The dyadic spatial down sampling filter may have tap values [−1, 17, 32, 17, −1]/64. A dyadic spatial up sampling filter having tap values configured according to a Kaiser window function having a beta factor of approximately 1.5, having approximately 2 side lobes, and having an up sampling ratio of approximately 2. The dyadic spatial up sampling filter may have tap values [−5.44, 0, 20.71, 33.46, 20.71, 0, −5.44]/64.0, or tap values [−5, 0, 21, 32, 21, 0, −5]/64, or tap values [−5, 21, 21, −5]/32.


