Nonuniformity correction for an infrared focal plane array based on diamond search block matching

Area of Science:

• Infrared imaging
• Image processing
• Signal processing

Background:

• Scene-based nonuniformity correction (NUC) algorithms are crucial for infrared imaging.
• Existing algorithms suffer from ghosting and blurring, degrading image quality.
• Artificial ghosting and image blurring are significant challenges in NUC.

Purpose of the Study:

• To propose an improved scene-based nonuniformity correction algorithm.
• To mitigate ghosting artifacts and image blurring in infrared images.
• To enhance the overall quality of nonuniformity correction.

Main Methods:

• Utilizing a diamond search block matching algorithm for accurate transform pair estimation between adjacent frames.
• Applying a gradient descent algorithm to update correction parameters based on transform pair errors.
• Implementing an adaptive learning rate controlled by local standard deviation and a threshold to prevent matching error accumulation.
• Employing a linear model for nonuniformity correction with updated parameters.

Main Results:

• The proposed algorithm significantly reduces nonuniformity in infrared image sequences.
• Demonstrated reduction in ghosting artifacts, particularly in moving image areas.
• Successfully overcame image blurring issues in static image areas.
• Experimental validation on four real infrared image sequences confirmed performance.

Conclusions:

• The developed algorithm offers a superior approach to scene-based nonuniformity correction.
• It effectively balances correction accuracy with artifact reduction.
• The method provides enhanced image quality for infrared applications.