Dynamic FPN Correction for Thermal Imaging Motion
Solution Overview
Problem
Fixed Pattern Noise (FPN) in thermal imaging systems, particularly in mass-produced infrared cameras, exceeds acceptable levels due to design requirements different from complex military or industrial systems, necessitating new methods for noise filtration.
Innovation Solution
A method for reducing FPN in imaging systems with occasional motion involves developing a pixel-by-pixel correction term through blurring, identifying excluded pixels, detecting motion, and updating or decaying the FPN correction term based on previous frame information, using a series of image processing steps including blurring, edge filtering, and motion detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional FPN filtering methods are used in mass-produced infrared cameras, then manufacturing cost is reduced, but Fixed Pattern Noise exceeds acceptable levels
Solution Approach 1:
The FPN correction approach transitions from static factory calibration to dynamic continuous correction. The system continuously updates FPN correction terms during operation using current frame data, allowing the correction to adapt to changing conditions while maintaining low manufacturing costs through software-based processing rather than expensive hardware modifications.
Solution Approach 2:
The imaging system performs self-correction of FPN by using its own captured frame data to generate correction terms. The system identifies excluded pixels, calculates motion, and updates FPN correction terms autonomously during operation, eliminating the need for expensive external calibration equipment or complex manufacturing processes.
2Measurement precision
If FPN correction is continuously updated using current frame data, then FPN reduction effectiveness is improved, but image motion causes inaccuracies in correction terms
Solution Approach 1:
The system dynamically adjusts the FPN correction process based on detected image motion. When motion is detected through comparison of blurred pixel sets, the system identifies excluded pixels that should not be used for correction calculations, preventing motion-induced inaccuracies while still allowing continuous correction during stable periods.
Solution Approach 2:
The system uses feedback from motion detection to control the FPN correction process. By comparing blurred pixel sets between frames and identifying excluded pixels based on motion thresholds, the system provides feedback that prevents inaccurate correction terms from being generated during motion, thereby maintaining correction accuracy despite varying image stability.
3Measurement precision
If excluded pixels are identified and removed from FPN calculations, then correction accuracy for moving regions is improved, but processing complexity increases
Solution Approach 1:
The image frame is segmented into excluded pixels and non-excluded pixels based on motion detection. This segmentation allows the FPN correction to be applied selectively only to stable regions, improving accuracy for moving regions while the processing complexity is managed through efficient algorithms that identify excluded pixels using simple motion thresholding on blurred data.
Solution Approach 2:
Different processing approaches are applied to different regions of the image based on local motion characteristics. Excluded pixels in moving regions are handled differently from non-excluded pixels in stable regions, with FPN correction applied only where appropriate. This local quality approach improves overall correction accuracy while maintaining reasonable processing complexity through region-specific handling.
4Stability of the object's composition
If blurred pixel sets are saved for multiple frames, then FPN correction continuity is improved, but memory requirements increase
Solution Approach 1:
Instead of saving complete high-resolution frame data for multiple frames, the system saves only the blurred pixel sets at lower resolution. This partial action approach maintains the essential information needed for FPN correction continuity while significantly reducing memory requirements by storing only the necessary processed data at reduced detail levels.
Data Source
AI summary
An imaging system whose Field of View FOV experiences occasional motion in relation to viewed scenes may be configured to reduce Fixed Pattern Noise (FPN) of acquired image data. FPN may be reduced by developing a pixel by pixel FPN correction term through a series of steps including blurring the image, identifying pixels to exclude from some calculations, a motion detector and an FPN updater for frames under motion and an FPN decay element for frames that are still.