Identifying laser induced damage on the surface of optical components for the purpose of tracking its growth over time and repairing it is an important part of the economical operation of the National Ignition Facility (NIF). Optics installed on NIF are monitored in situ for damage growth and can be removed as needed for repair and re-use. An ex-situ automated microscopy system is used to inspect full sized NIF optics allowing for the detection of damage sites >10 µm in diameter. Due to the various morphology of laser damage, several algorithms are used to analyze the microscopy data and identify damage regardless of size, while ignoring features not related to laser damage. This system has significantly increased the lifetime of NIF final optics (∼2.3x) thereby extending beyond the capabilities of the in-situ inspection by itself.

Accurately classifying microscopic damage helps automate the repair and recycling of National Ignition Facility optics and informs the study of damage initiation and growth. This complex 12-class problem previously required human experts to distinguish and label the various damage morphologies. Finding image analysis methods to extract and calculate distinguishing features would be time consuming and challenging, so we sought to automate this task by using convolutional neural networks (CNNs) pretrained on the ImageNet database to take advantage of its automated feature discovery and extraction. We compared three model architectures on this dataset and found the one with highest overall accuracy, 99.17%, was a novel hybrid architecture, one in which we removed the final decision-making layer of the deep learner and replaced it with an ensemble of decision trees (EDT). This combines the power of feature extraction by CNNs with the decision-making strength of EDT.The accuracy of the hybrid architecture over the deep learning alone is shown to be significantly improved. Furthermore, we applied this novel hybrid architecture to an entirely different dataset, one containing images of repaired damage sites, and improved on the previously published findings, also with a demonstrably significant increase in accuracy over using the deep learner alone. K E Y W O R D Sautomation, deep learning, laser optic damage, machine learning