Non-uniformity Correction via Shifted Image Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Infrared imaging devices face challenges in noise removal, particularly high spatial frequency fixed pattern noise (FPN), which is difficult to address due to mechanical failures in shutters and increased size and cost in small form factor devices, and existing methods may obscure fine details in scenes.
Innovation Solution
The technique involves capturing image frames with frame-to-frame shifts caused by controlled or detected motion patterns, allowing for the distinction of noise from scene information and updating non-uniformity correction terms to reduce noise without the need for a shutter.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a shutter is used to provide a substantially uniform scene for calibration, then non-uniformity correction can be achieved, but the device size and cost increase and mechanical failure risk increases
Solution Approach 1:
The patent extracts the shutter component from the system and replaces it with a computational method. Instead of using a physical shutter to create uniform scenes for calibration, the system uses software algorithms to simulate and remove non-uniformity effects, thereby eliminating the mechanical shutter while maintaining correction capability.
Solution Approach 2:
The patent replaces the mechanical shutter system with a digital image processing system. The mechanical shutter that physically blocks light is substituted by computational algorithms that digitally remove non-uniformity effects through frame differencing and noise modeling, converting a mechanical problem into a computational one.
2Reliability
If a shutter is used for calibration, then non-uniformity correction is achieved, but mechanical failure risk increases
Solution Approach 1:
The patent removes the mechanical shutter from the system entirely, extracting the problematic component while retaining the essential function of non-uniformity correction through computational methods. This eliminates mechanical failure risks while maintaining calibration effectiveness.
Solution Approach 2:
The mechanical shutter system is replaced with a software-based correction system that uses image frame differencing and noise modeling to achieve the same calibration objectives without any moving mechanical parts, thereby eliminating mechanical failure modes.
3Reliability
If intentional blurring is used to remove FPN, then noise removal is achieved, but fine details of the scene are obscured
Solution Approach 1:
Instead of applying full blurring to remove noise, the patent applies partial correction by calculating and applying only the necessary non-uniformity correction terms. The system processes multiple image frames to compute correction factors that target specific noise patterns while preserving the original image's fine details, avoiding excessive processing that would blur legitimate scene information.
Data Source
AI summary
Various techniques are provided to reduce noise in captured images using frame-to-frame shifts of scene information. In one example, a system includes an imager and a processor. The imager is configured to capture a plurality of image frames comprising scene information. The image frames exhibit frame-to-frame shifts of the scene information caused by a motion pattern associated with the imager. The processor is configured to distinguish noise in the image frames from the frame-to-frame shifts of the scene information. The processor is also configured to update non-uniformity correction terms to reduce the noise. Additional systems and methods are also provided.