Journal
IEEE SENSORS JOURNAL
Volume 22, Issue 14, Pages 14415-14423Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSEN.2022.3183868
Keywords
Superresolution; Imaging; Image reconstruction; Dictionaries; Detectors; Training; Testing; Compressive sensing; super-resolution; non-destructive testing; infrared image; bonded joints
Funding
- Fuzhou University [GXRC-18066]
- Department of Education, Fujian Province [JAT190005]
Ask authors/readers for more resources
In this paper, a post-processing method based on compressive sensing for uncooled IR sensors is proposed to perform super-resolution NDT thermography. The method generates super-resolution IR phase images by using a sparse representation of low-resolution IR images and a sparse dictionary generated from high-resolution raw training image patches. The method is verified using pulsed phase thermography and shows improved visual effect and reduced mean square error compared to other methods. The proposed method is expected to enable wider application of uncooled IR sensors in NDT.
To compensate for the inability of inexpensive uncooled infrared (IR) sensors to perform high-speed thermography detection with low noise and high sensitivity in the field of non-destructive testing (NDT), a post-processing method for uncooled IR sensors based on compressive sensing is proposed in this paper to perform super-resolution NDT thermography. Super-resolution IR phase images can be generated by first using a sparse representation of the low-resolution (LR) IR images and a sparse dictionary generated from randomly sampled high-resolution (HR) raw training image patches to generate an HR IR image sequence, and, subsequently, implementing IR thermography NDT. To verify the reliability of the proposed method, super-resolution IR images were combined with pulsed phase thermography (PPT) to verify bonded joint structures in carbon fibre-reinforced polymer (CFRP). Compared to other methods, the proposed method reduces the mean square error (MSE) by 11.13% and yields a better visual effect on the PPT phase image reconstructed using three-fold super-resolution. The proposed method not only produces sharp defect edges, but also retains the original texture, which is expected to enable wider application of uncooled IR sensors in NDT applications in the future.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available