Sparse Sensing and Machine Learning for Rapid Calibration of Defect Detection with rf Atomic Magnetometers

Area of Science:

• Physics
• Materials Science
• Computer Science

Background:

• Radio-frequency atomic magnetometers (RF-AMs) are crucial for sensitive magnetic field detection.
• Traditional defect detection methods using RF-AMs can be time-consuming due to image acquisition and calibration.
• Improving the speed and efficiency of these processes is essential for practical applications.

Purpose of the Study:

• To enhance the speed of image acquisition and calibration for defect detection using RF-AMs.
• To explore the application of machine learning and sparse sensing for faster data reconstruction.
• To develop image processing algorithms for improved defect characterization and experimental parameter adjustment.

Main Methods:

• Utilized machine learning algorithms to optimize image acquisition and calibration.
• Implemented sparse sensing techniques for efficient data sampling.
• Developed image processing algorithms for result qualification and parameter adjustment.

Main Results:

• Achieved a reconstructed image with an average relative error of approximately 5% using only 1.25% of sampled pixels.
• Demonstrated the effectiveness of image processing in qualifying results and enhancing image contrast.
• Showcased the potential for significantly faster defect detection compared to traditional raster scanning.

Conclusions:

• Machine learning and sparse sensing offer a viable approach to accelerate defect detection with RF-AMs.
• The developed image processing techniques improve the quality and interpretability of defect data.
• This work represents a foundational step towards a tool for rapid, arbitrary defect detection.