Symmetrical Polyphase Video Scaling via Farrow Interpolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video scaling algorithms using polyphase filters require a large number of phases to mitigate approximation errors, leading to increased system cost and visible phase noise, with limited size reduction and undesirable effects on scaling quality.
Innovation Solution
A system and method for implementing graphics and video scaling using second-order interpolation based on symmetrical polyphase filtering, employing a Farrow structure with symmetric polyphase filters to generate in-phase and out-of-phase filtered pixels and an interpolator for scaled output, supporting upscaling and downsampling with reduced hardware complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the number of phases of the polyphase filter is increased to mitigate approximation errors, then the scaling quality is improved, but the system cost and hardware complexity increase
Solution Approach 1:
The patent divides the polyphase filter into a small number of phases (2-4 phases) and uses interpolation to achieve the desired scaling quality. Instead of using many phases, the system segments the filtering operation into fewer phases and compensates through interpolation between the outputs of these phases, thereby reducing hardware complexity while maintaining scaling quality.
Solution Approach 2:
The patent changes the approach from using many phases with simple filtering to using few phases with interpolation. By changing the parameter of the number of phases from large to small (2-4 phases), the system achieves the same scaling quality through a different mechanism (interpolation between phases) that requires less hardware resources.
2Manufacturing precision
If interpolation between sub-phase filters is used to alleviate phase accuracy limits, then the scaling quality is improved, but the number of required multiplications increases
Solution Approach 1:
The patent uses a small number of phases (2-4 phases) which is fewer than the traditional large number of phases, and combines this with interpolation to achieve the desired accuracy. This partial action approach uses only the necessary minimum number of phases and compensates through interpolation, avoiding the excessive complexity of using many phases or many multiplications.
3Device complexity
If simplified methods are used to reduce hardware size, then the device complexity is reduced, but phase noise becomes visible and undesirable
Solution Approach 1:
The patent introduces interpolation as an intermediary mechanism between the few polyphase filter outputs and the final scaled output. This intermediary interpolation step smooths out the phase noise that would otherwise be visible, allowing the system to use fewer phases (reducing hardware size) while maintaining acceptable scaling quality and minimizing harmful phase noise effects.
4Device complexity
If a small number of phases is used in the polyphase filter, then the hardware complexity is reduced, but approximation errors increase
Solution Approach 1:
The patent replaces the mechanical approach of using many filter phases with a computational approach using interpolation. Instead of relying on the mechanical filtering action of many phases, the system uses a small number of phases and substitutes the missing filtering action with interpolation between the phase outputs, thereby reducing hardware complexity while controlling approximation errors through the interpolation process.
Data Source
AI summary
Certain embodiments of the invention may be found in a system and method for implementing graphics and a video scaling algorithm using interpolation based on symmetrical polyphase filtering. A video or graphics scaler may be utilized to scale luma, chroma, and/or alpha information in a video image. The scaler may comprise a first symmetric polyphase sub-filter with zero phase shift that generates an in-phase filtered pixel and a second symmetric polyphase sub-filter that generates an out-of-phase filtered pixel. The video scaler may also comprise an interpolator that may generate a scaled video image pixel based on the generated in-phase and out-of-phase filtered pixels and a scaling factor. The scaling factor may be determined based on an input video size (M) and a desired output video size (N). The interpolation of the generated in-phase and out-of-phase pixels in the video scaler may be implemented by utilizing a Farrow structure.