Abstract

A terahertz imaging system is considered to be an effective method to study the thermal barrier coating defects in gas turbine engines. However, due to the influence of the system hardware and terahertz wavelength, the imaging system has slow acquisition efficiency, low image resolution, and serious edge blur, which cannot meet the demand for defect detection. To overcome the above defects, a model-driven terahertz image reconstruction method is proposed, which uses simulation data to build datasets, reduces the dependence on experimental data, and has a good reconstruction effect on experimental images. A fusion loss function based on the edge intensity was designed to optimize the edge effect of reconstructed images. Compared with the bicubic, SRCNN, and VDSR methods, the proposed method can achieve better results in terms of visual and evaluation indices for the reduced terahertz images. It is proved that this method can effectively restore the defect contour in the terahertz image, sharpen the edge of the image, and improve the image quality. It has a good application value in the industry.

© 2023 Optica Publishing Group

Super-resolution reconstruction of terahertz images based on a deep-learning network with a residual channel attention mechanism

Xiuwei Yang, Dehai Zhang, Zhongmin Wang, Yanbo Zhang, Jun Wu, Biyuan Wu, and Xiaohu Wu
Appl. Opt. 61(12) 3363-3370 (2022)

Terahertz image super-resolution based on a deep convolutional neural network

Zhenyu Long, Tianyi Wang, ChengWu You, Zhengang Yang, Kejia Wang, and Jinsong Liu
Appl. Opt. 58(10) 2731-2735 (2019)

Detection method and instrument of interfacial debonding inside a thermal barrier coating based on fluorescence imaging

Zong Wang, Yanheng Zhang, Jinli Guo, Shijie Zhang, and Wei Qiu
Appl. Opt. 61(18) 5401-5408 (2022)

Super-resolution terahertz synthetic aperture image reconstruction algorithm

Ningbo Wang and Feng Qi
Appl. Opt. 63(1) 186-192 (2024)

Super-resolution reconstruction for terahertz imaging based on sub-pixel gradient field transform

Youdong Guo, Furi Ling, He Li, Siyan Zhou, Jie Ji, and Jianquan Yao
Appl. Opt. 58(23) 6244-6250 (2019)