Shutterless FIR Camera Nonuniformity Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Shutter-based far-infrared (FIR) cameras face challenges in automotive applications due to unacceptable downtime, mechanical failures, and inadequate correction of residual nonuniformity, which affects image quality and safety in advanced driver assistance systems and autonomous vehicles.
Innovation Solution
A shutterless FIR camera system that uses a noise estimation algorithm and weight mask matrix to incrementally correct nonuniformity in images, facilitating noise estimation in smooth regions and inhibiting noise estimation in areas with strong edges or temporal changes, thereby improving image quality without the need for mechanical shutters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a shutter is used for flat-field correction in FIR cameras, then ambient drift and nonuniformity are corrected, but image capture is interrupted causing unacceptable downtime for automotive applications
Solution Approach 1:
The patent extracts the shutter mechanism from the FIR camera system entirely. Instead of using a mechanical shutter to block radiation during calibration, the system uses computational methods to estimate and remove nonuniformity from continuous video frames, eliminating the need for shutter-induced interruptions in image capture.
Solution Approach 2:
The patent replaces the mechanical shutter system with a computational algorithm. The noise estimation algorithm processes video frames mathematically to correct nonuniformity, substituting mechanical action (shutter opening/closing) with digital signal processing operations.
2Measurement precision
If a shutter is used for calibration, then nonuniformity correction is achieved, but mechanical wear and potential failure occur
Solution Approach 1:
The patent replaces the mechanical shutter system with a computational algorithm. The noise estimation algorithm processes video frames mathematically to correct nonuniformity, substituting mechanical action (shutter opening/closing) with digital signal processing operations, thereby eliminating wear and mechanical failure risks.
Solution Approach 2:
The system performs self-calibration using its own video output. The noise estimation algorithm analyzes the camera's captured footage to identify and correct nonuniformity patterns, allowing the system to maintain image quality without external calibration equipment or mechanical intervention.
3Measurement precision
If noise estimation is applied to all pixels, then correction is comprehensive, but estimation accuracy decreases in regions with strong edges or temporal changes
Solution Approach 1:
The patent applies different processing strategies to different image regions. The weight mask matrix identifies smooth regions suitable for noise estimation and applies correction there, while excluding regions with strong edges or temporal changes from the estimation process, ensuring high accuracy where applicable.
Solution Approach 2:
The patent applies noise estimation selectively rather than universally. By using the weight mask to limit correction to appropriate regions (smooth areas), the system achieves higher overall accuracy by concentrating computational effort where it is most effective rather than applying uniform processing everywhere.
Data Source
AI summary
A system and method for correcting nonuniformity in far-infrared (FIR) images captured by a shutterless FIR camera. The method includes determining a noise of a current image based on updating a noise estimate of a previous image with a noise estimate of a current image; determining a weight mask matrix of the current image, where the weight matrix includes high values corresponding to pixels of the current image in which noise estimation is facilitated, and low values corresponding to pixels of the current image in which noise estimation is inhibited; applying the weight mask matrix to the current image; and correcting the nonuniformity of the current image incrementally based on the determined noise of current image and the applied weight mask matrix.