A terahertz time-domain super-resolution imaging method using a local-pixel graph neural network for biological products

Lei, Tong; Tobin, Brian; Liu, Zihan; Yang, Shuyi; Sun, Da‐Wen

doi:10.1016/j.aca.2021.338898

Cited by 18 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep neural networks (DNNs) have also been used to ameliorate THz data using robust restoration approaches, but rarely considering the frequency-dependence of the THz-TDS images during the training process. They were mostly used for 2-D THz image deblurring [22] and super-resolution [23], [24] or for specific applications, such as biological product analysis [25]. The main limitation of DNN-based methods is the need for a large number of THz images (hundreds or thousands) for accurately training the network: the unavailability of a sufficient number of THz images has, as a consequence, limited their application.…”

Section: Image Processing Methods For Thz Datamentioning

confidence: 99%

Beam-Shape Effects and Noise Removal From THz Time-Domain Images in Reflection Geometry in the 0.25–6 THz Range

Ljubenovic

Artesani

Bonetti

et al. 2022

IEEE Trans. THz Sci. Technol.

View full text Add to dashboard Cite

The increasing need of restoring high-resolution hyper-spectral (HS) images is determining a growing reliance on computer vision-based processing to enhance the clarity of the image content. HS images can, in fact, suffer from degradation effects or artefacts caused by instrument limitations. This article focuses on a procedure aimed at reducing the degradation effects, frequency-dependent blur and noise, in Terahertz time-domain spectroscopy (THz-TDS) images in reflection geometry. It describes the application of a joint deblurring and denoising approach that had been previously proved to be effective for the restoration of THz-TDS images in transmission geometry, but that had never been tested in reflection modality. This mode is often the only one that can be effectively used in most cases, for example, when analyzing objects that are either opaque in the THz range, or that cannot be displaced from their location (e.g., museums), such as those of cultural interest. Compared to transmission mode, reflection geometry introduces, however, further distortion to THz data, usually neglected in existing literature. In this work, we successfully implement image deblurring and denoising of both uniform-shape samples (a contemporary 1 Euro cent coin and an inlaid pendant) and samples with the uneven reliefs and corrosion products on the surface, which make the analysis of the object particularly complex (an ancient Roman silver coin). The study demonstrates the ability of image processing to restore data in the 0.25-6 THz range, spanning over more than four octaves, and providing the foundation for future analytical approaches of cultural heritage using the far-infrared spectrum still not sufficiently investigated in the literature.

show abstract

Section: Image Processing Methods For Thz Datamentioning

confidence: 99%

Beam-Shape Effects and Noise Removal From THz Time-Domain Images in Reflection Geometry in the 0.25–6 THz Range

Ljubenovic

Artesani

Bonetti

et al. 2022

IEEE Trans. THz Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Deep neural networks (DNNs) have also been used to ameliorate THz data using robust restoration approaches, but rarely considering the frequency-dependence of the THz-TDS images during the training process. They were mostly used for 2D THz image deblurring [21] and super-resolution [22], [23] or for specific applications, such as biological product analysis [24]. The main limitation of DNN-based methods is the need for a large number of THz images (hundreds or thousands) for accurately training the network: unavailability of a sufficient number of THz images has, as a consequence, limited their application.…”

Section: Image Processing Methods For Thz Datamentioning

confidence: 99%

Beam-Shape Effects and Noise Removal from THz Time-Domain Images in Reflection Geometry in the 0.25-6 THz Range

Ljubenovic¹,

Artesani²,

Bonetti³

et al. 2022

Preprint

View full text Add to dashboard Cite

The increasing need of restoring high-resolution Hyper-Spectral (HS) images is determining a growing reliance on Computer Vision-based processing to enhance the clarity of the image content. HS images can, in fact, suffer from degradation effects or artifacts caused by instrument limitations. This paper focuses on a procedure aimed at reducing the degradation effects, frequency-dependent blur and noise, in Terahertz Time-Domain Spectroscopy (THz-TDS) images in reflection geometry. It describes the application of a joint deblurring and denoising approach that had been previously proved to be effective for the restoration of THz-TDS images in transmission geometry, but that had never been tested in reflection modality. This mode is often the only one that can be effectively used in most cases, for example when analyzing objects that are either opaque in the THz range, or that cannot be displaced from their location (e.g., museums), such as those of cultural interest. Compared to transmission mode, reflection geometry introduces, however, further distortion to THz data, neglected in existing literature. In this work, we successfully implement image deblurring and denoising of both uniform-shape samples (a contemporary 1 Euro cent coin and an inlaid pendant) and samples with the uneven reliefs and corrosion products on the surface which make the analysis of the object particularly complex (an ancient Roman silver coin). The study demonstrates the ability of image processing to restore data in the 0.25 -6 THz range, spanning over more than four octaves, and providing the foundation for future analytical approaches of cultural heritage using the farinfrared spectrum still not sufficiently investigated in literature.

show abstract

“…Very recently, the use of deep learning for superresolution purposes based on terahertz imaging images is attracting the attention of many researchers in the NDT&E field. In particular, in [133,131,134] tailored deep convolutional neural network architectures have been proposed to enhance the resolution of Terahertz images based on measurements performed on different kind of structures. In [135,136], the super-resolution task have been tackled by considering generative adversarial network adapted to a dataset of experimental Terahertz images.…”

Section: Infrared Thermography Testing and Terahertz Wave Testingmentioning

confidence: 99%

Deep learning techniques for non-destructive testing and evaluation

Miorelli¹,

Skarlatos²,

Vienne³

et al. 2022

Applications of Deep Learning in Electromagnetics: Teaching Maxwell's Equations to Machines

View full text Add to dashboard Cite

A terahertz time-domain super-resolution imaging method using a local-pixel graph neural network for biological products

Cited by 18 publications

References 41 publications

Beam-Shape Effects and Noise Removal From THz Time-Domain Images in Reflection Geometry in the 0.25–6 THz Range

Beam-Shape Effects and Noise Removal From THz Time-Domain Images in Reflection Geometry in the 0.25–6 THz Range

Beam-Shape Effects and Noise Removal from THz Time-Domain Images in Reflection Geometry in the 0.25-6 THz Range

Deep learning techniques for non-destructive testing and evaluation

Contact Info

Product

Resources

About