Abstract

In this paper, we demonstrate, both numerically and experimentally, a method for the detection of defects on structured surfaces having optically unresolved features. The method makes use of synthetic reference data generated by an observational model that is able to simulate the response of the selected optical inspection system to the ideal structure, thereby providing an ideal measure of deviation from nominal geometry. The method addresses the high dynamic range challenge faced in highly parallel manufacturing by enabling the use of low resolution, wide field of view optical systems for defect detection on surfaces containing small features over large regions.

Automatic optical inspection platform for real-time surface defects detection on plane optical components based on semantic segmentation

Jules Karangwa, Linghua Kong, Dingrong Yi, and Jishi Zheng
Appl. Opt. 60(19) 5496-5506 (2021)

Metrology of optically-unresolved features using interferometric surface profiling and RCWA modeling

Peter de Groot, Xavier Colonna de Lega, Jan Liesener, and Michael Darwin
Opt. Express 16(6) 3970-3975 (2008)

Improved STMask R-CNN-based defect detection model for automatic visual inspection of an optics lens

Haiyang Tang, Shan Liang, Dan Yao, and Yongjie Qiao
Appl. Opt. 62(33) 8869-8881 (2023)

Unsupervised defect detection in textiles based on Fourier analysis and wavelet shrinkage

Guang-Hua Hu, Qing-Hui Wang, and Guo-Hui Zhang
Appl. Opt. 54(10) 2963-2980 (2015)

FOC winding defect detection based on improved texture features and low-rank representation model

Xiaole Chen, Ruifeng Yang, Chenxia Guo, and Qianchuang Zhang
Appl. Opt. 61(19) 5599-5607 (2022)