Detection of laser-induced optical defects based on image segmentation

Abstract

A number of vision-based methods for detecting laser-induced defects on optical components have been implemented to replace the time-consuming manual inspection. While deep-learning-based methods have achieved state-of-the-art performances in many visual recognition tasks, their success often hinges on the availability of a large number of labeled training sets. In this paper, we propose a surface defect detection method based on image segmentation with a U-shaped convolutional network (U-Net). The designed network was trained on paired sets of online and offline images of optics from a large laser facility. We show in our experimental evaluation that our approach can accurately locate laser-induced defects on the optics in real time. The main advantage of the proposed method is that the network can be trained end to end on small samples, without the requirement for manual labeling or manual feature extraction. The approach can be applied to the daily inspection and maintenance of optical components in large laser facilities.

Data-driven science and machine learning methods in laser–plasma physics

“…Another example for the use of both SVM-and CNN-based classification in high-power laser systems was recently presented by Pascu [269] , who used both techniques for (supervised) anomaly detection in a laser beam profile at the ELI-NP facility. Chu et al [270] presented a first application of image segmentation to locate laser-induced defects on optics in real time using a U-Net. Ben Soltane et al [271] recently presented a deep learning pipeline to estimate the size of damages in glass windows at the Laser Mégajoule (LMJ) facility, using a similar U-Net architecture for segmentation.…”

Continuous gradient fusion class activation mapping: segmentation of laser-induced damage on large-aperture optics in dark-field images

“…Semantic segmentation algorithms based on deep convolutional neural networks do not require custom-built parameters for the above multifactor interference scenes and can automatically extract effective damage features and robustly segment damage sites. Chu et al [ 17 ] of CAEP constructed a fully convolutional network with a U-shaped architecture (U-Net). Through fully supervised training, this model achieves higher damage detection accuracy than conventional methods.…”

Potential damage threats to downstream optics caused by Gaussian mitigation pits on rear KDP surface

“…to determine whether laser damage to the optics surface has occurred [5,9] . Tiny defects such as surface fractures and laser ablation would then be replaced with specific smooth contours [7,10] .…”