Statistical Scene-Based Non-Uniformity Correction Method with Interframe Registration.

Area of Science:

• Infrared imaging technology
• Image processing and computer vision

Background:

• Non-uniform response in infrared focal plane array (IRFPA) detectors causes fixed-pattern noise, corrupting image quality.
• Existing non-uniformity correction (NUC) methods often suffer from high time delay or suboptimal performance.

Purpose of the Study:

• To introduce a novel, adaptive scene-based NUC method, CSIR (Correction method with Statistical scene-based and Interframe Registration).
• To achieve low-delay calculation of correction coefficients for infrared images.
• To enhance NUC performance by combining statistical and registration techniques.

Main Methods:

• CSIR employs a statistical method to estimate gain coefficients, providing an initial value for the registration method.
• The integrated approach reduces the reliance on interactive images, thereby lowering time delay.
• Real non-uniformity infrared image sequences were utilized for validation.

Main Results:

• CSIR demonstrates superior performance compared to standalone statistical or registration methods.
• The method significantly reduces fixed-pattern noise with effective gain and offset correction.
• Validation on real infrared image sequences confirms CSIR's speed and reliability.

Conclusions:

• CSIR offers a fast and reliable solution for fixed-pattern noise reduction in IRFPA imaging.
• The combined statistical and registration approach effectively minimizes NUC time delay.
• The proposed method achieves improved accuracy and robustness for infrared image correction.