Inspection Applications Research Articles

Class-Patch Similarity Weighted Embedding for Few-Shot Infrared Image Classification

Infrared imaging plays a vital role in critical surveillance, military reconnaissance, and industrial inspection applications due to its advantages such as strong concealment and the ability to operate around the clock. However, the combination of low infrared image resolution and complex background scenarios poses significant challenges for traditional deep learning models in accurately extracting the most discriminative features for classification. These models are often disrupted by irrelevant features, especially when data for new classes is scarce. Current few-shot learning approaches heavily rely on comparing image patches, but the scarcity of data can significantly degrade the performance of recognition algorithms. To address these challenges, we propose the Class Patch Similarity Weighted Embedding (CPSWE) framework for few-shot infrared target classification. The CPSWE framework employs a ViT architecture for feature extraction. By introducing class embeddings and calculating similarity-based weights for each patch, CPSWE reweights the patch features to enhance their relevance to the target class. This approach improves the discriminability of class-related features, leading to better generalization in few-shot settings. Furthermore, we introduce an infrared dataset specifically designed for few-shot learning, combining multiple open-source datasets to support research in this area. Extensive experiments on the few-shot learning benchmark dataset miniImageNet and the infrared dataset miniIRNet show that CPSWE outperforms existing few-shot learning methods, achieving significant improvements in classification accuracy on infrared image datasets with limited labeled samples.

Review of Short-Wavelength Infrared Flip-Chip Bump Bonding Process Technology.

Short-wave infrared (SWIR) imaging has a wide range of applications in civil and military fields. Over the past two decades, significant efforts have been devoted to developing high-resolution, high-sensitivity, and cost-effective SWIR sensors covering the spectral range from 0.9 μm to 3 μm. These advancements stimulate new prospects across a wide array of fields including life sciences, medical diagnostics, defense, surveillance, security, free-space optics (FSO), thermography, agriculture, food inspection, and LiDAR applications. In this review, we begin by introducing monolithic SWIR image sensors and hybrid SWIR image sensors and indicate that flip-chip bump bonding technology remains the predominant integration method for hybrid SWIR image sensors owing to its outstanding performance, adaptable integration with innovative epitaxial SWIR materials, long-term stability, and long-term reliability. Subsequently, we comprehensively summarize recent advancements in epitaxial thin-film SWIR sensors, encompassing FPAs and flip-chip bump bonding technology for epitaxial InGaAs and Ge (Sn) thin-film SWIR sensors. Finally, a summary and outlook regarding the development of InGaAs and Ge (Sn) SWIR sensors are provided and discussed. The ongoing evolution of epitaxial thin-film SWIR sensors with flip-chip bump bonding technology is poised to foster new applications in both academic and industry fields.

Dynamic Monolithic X-ray Imager with Enhanced Performance via Strain Relaxation in Metal-Halide Scintillator

Scintillation X-ray detectors have widespread applications in medical imaging, non-destructive testing, and security inspection. In recent years, metal halide materials have gained attention as scintillators due to their superior properties,...

An agile multimodal microrobot with architected passively morphing wheels.

Multimodal microrobots are of growing interest due to their capabilities to navigate diverse terrains, with promising applications in inspection, exploration, and biomedicine. Despite remarkable progress, it remains challenging to combine the attributes of excellent maneuverability, low power consumption, and high robustness in a single multimodal microrobot. We propose an architected design of a passively morphing wheel that can be stabilized at distinct geometric configurations, relying on asymmetric bending stiffness of bioinspired tentacle structures. By integrating such wheels with electromagnetic motors and a flexible body, we develop a highly compact, lightweight, multimodal microrobot (length ~32 mm and mass ~4.74 g) with three locomotion gaits. It has high motion speed (~21.2 BL/s), excellent agility (relative centripetal acceleration, ~206.9 BL/s2), low power consumption (cost of transport, ~89), high robustness, and strong terrain adaptabilities. Integration of batteries and a wireless control module enables developments of an untethered microrobot that maintains high motion speed and excellent agility, with capabilities of traveling in hybrid terrains.

Flexible Piezoresistive Film Pressure Sensor Based on Double-Sided Microstructure Sensing Layer.

Flexible thin-film pressure sensors have garnered significant attention due to their applications in industrial inspection and human-computer interactions. However, due to their ultra-thin structure, these sensors often exhibit lower performance, including a narrow pressure response range and low sensitivity, which constrains their further application. The most commonly used microstructure fabrication methods are challenging to apply to ultra-thin functional layers and may compromise the structural stability of the sensors. In this study, we present a novel design of a film pressure sensor with a double-sided microstructure sensing layer by adopting the template method. By incorporating the double-sided microstructures, the proposed thin-film pressure sensor can simultaneously achieve a high sensitivity value of 5.5 kPa-1 and a wide range of 140 kPa, while maintaining a short response time of 120 ms and a low detection limit. This flexible film pressure sensor demonstrates considerable potential for distributed pressure sensing and industrial pressure monitoring applications.

Multiple Applications of Chiral Self-Assembled H8-BINOL Nanoprobes with AIE-ESIPT Characteristics: Chiral Switching and pH Visualization Detection.

R-BTOBD exhibited significant aggregation-induced Emission (AIE) characteristics, including high brightness (αAIE ≈ 40), robust light stability, a substantial Stokes shift (128 nm), and a high signal-to-noise ratio, effectively overcoming aggregation-caused quenching (ACQ). Derived from the axially chiral R-H8-BINOL, R-BTOBD was synthesized via nucleophilic cyclization and exhibited pronounced self-assembly properties. Through robust intra- and intermolecular hydrogen bonding interactions, R-BTOBD formed diverse supramolecular structures, including spherical flower-like aggregates, hollow-core triangular tubules, hexagonal tubules, and irregular white block-like stacks. The specific morphology was influenced by factors such as solvent and pH. For fluorescence applications, R-BTOBD can be employed as an effective visual pH detector through an AIE-enhanced excited-state intramolecular proton transfer (AIE-ESIPT) mechanism. When incorporated into pH paper, it exhibited a bright yellow fluorescence across a wide pH range (pH 1-11), transitioning to an intense bright blue fluorescence under strongly basic conditions (pH 12-14). In chiral applications, R-BTOBD demonstrated potential as a chiral switching agent, leveraging the pronounced changes and inversions in its circular dichroism (CD) signals under varying aggregation states and pH conditions. Additionally, it serves as a specific chiral enantioselective fluorescence probe for l-lysine, demonstrating high efficiency and a low detection limit. Given its exceptional fluorescence and chiral properties, this fluorescent material has significant potential for diverse applications in fluorescence, chirality, and visual inspection.

Development and validation of genome-informed and multigene-based qPCR and LAMP assays for accurate detection of Dickeya solani: a critical quarantine pathogen threatening the potato industry.

Dickeya solani one of the most aggressive pectinolytic phytopathogens, causes blackleg disease in potatoes, resulting in significant economic losses and adversely impacting one of the world's most important food crops. The diagnostics methods are critical in monitoring the latent infection for international trade of potato seed tubers and in implementation of control strategies. Our study employed a whole-genome comparative approach, identifying unique target gene loci (LysR and TetR family of transcriptional regulators gene regions) and designing loop-mediated isothermal amplification (LAMP) and a multi-gene-based multiplex TaqMan qPCR assays for specific detection and differentiation of D. solani. Both methods underwent meticulous validation with extensive inclusivity and exclusivity panels, exhibiting 100% accuracy and no false positives or negatives. The LAMP method demonstrated the detection limit of 100 fg and 1 CFU per reaction using pure genomic DNA and crude bacterial cell lysate, respectively. The qPCR detection limit was 1 pg, 100 fg and 10 fg with quadplex, triplex, and singleplex, respectively. None of the assays were impacted by any inhibitory or competitive effects after adding host DNA (in qPCR) or crude lysate (in LAMP). The assays proved robust and reproducible in detecting the target pathogen in infected samples, with the LAMP assay being field-deployable due to its simplicity and rapid results acquisition within approximately 9 min. The reproducibility was confirmed by performing the assay in two independent laboratories. These assays offer a robust, rapid, and reliable solution for routine testing, with applications in phytosanitary inspection, disease diagnosis, and epidemiological studies.IMPORTANCEDickeya solani, one of the most aggressive soft rot causing bacteria and a quarantine pathogen, poses a severe threat to food security by causing substantial economic losses to the potato industry. Accurate and timely detection of this bacterium is vital for monitoring latent infections, particularly for international trade of potato seed tubers, and for implementing effective control strategies. In this research, we have developed LAMP and multi-gene-based multiplex TaqMan qPCR assays for specific detection of D. solani. These assays, characterized by their precision, rapidity, and robustness, are crucial for distinguishing D. solani from related species. Offering unparalleled sensitivity and specificity, these assays are indispensable for phytosanitary inspection and epidemiological monitoring, providing a powerful tool enabling management of this threatening pathogen.

A review of computer vision applications for asset inspection in the oil and gas Industry

A review of computer vision applications for asset inspection in the oil and gas Industry

Multivariate image analysis for assessment of textural attributes in transglutaminase-reconstituted meat

The control of sensorial textural attributes has high interest to the meat industry focused on the recovery of the value of meat by-products by developing reconstituted meat pieces with added sensory and nutritional values. Sensorial analysis of foods is still a quite subjective methodology, highly dependent of a well-trained team of inspectors, which is simulated by textural analysis in order to measure objective physical properties. This work presents a non-destructive and contactless experimental methodology to predict the physical properties of a reconstituted meat product, based on integrating multispectral imaging and multivariate image analysis (MIA). The experiment was based on reconstituting grounded meat with different concentrations of transglutaminase (0.1, 1, 3, 6 and 10 %), from which textural properties and multispectral imaging data were measured. Multispectral images (UV, VIS and NIR wavelengths) were processed with chemometric procedures to obtain the distribution maps and score images, from which different blocks of features were extracted to generate feature vectors (basic statistics and co-occurrence matrix) for each image. The obtained regression models built with these features predicted all physical properties of the meat with Q2 > 0.90, after feature selection using VIPs. These results evidenced the capacity of multispectral imaging, combined with chemometric procedures, to capture the variability of physical properties induced by transglutaminase in a derivate meat product. It could represent the base of a potential contactless application for a meat industrial inspection, where work environments have strong hygienic requirements.

Long-Tune Natural Logarithmic Wavelength Modulation Spectroscopy for Gas Sensing.

This article presents a gas sensing method based on long-tune natural logarithmic wavelength modulation spectroscopy (long-tune ln-WMS) and explores means to improve its accuracy. The long-tune spectrum can detect multiple gases with high precision. In ln-WMS, due to the natural logarithm algorithm, the harmonic magnitude which is related to gas concentration would not be affected by the light intensity fluctuations. However, the background signal of the harmonic will become strong and nonlinear in the long-tune spectrum. Three CO2 absorption lines and one H2O line near 2004 nm are applied to verify the proposed theory. The effects of light intensity, modulation depth, gas concentration, and phase shift on the harmonics are tested separately through both simulations and experiments. The results reveal that our proposed method can always keep the harmonics at their maximum which ensures high measurement precision. Moreover, the background signal only varies with the modulation depth, not the concentration and light intensity. Even the mechanical vibrations cannot disturb the harmonics, which enables the proposed method to be suitable for gas detection in harsh environments, especially for heavy dust and severe mechanical vibrations. The CO2 concentration detection results indicate that when the background is eliminated, the accuracy can be achieved with a relative error of below 0.5%, while the error would be greater than 5% with background presence. The proposed long-tune ln-WMS method is effective for trace gas detection (weak absorption) or over-modulation conditions and has potential applications in field inspection.

Analytical modeling and non-dimensionalization study for endoscopic ultrasound acoustic field in tubular structure

The concept of endoscopic ultrasound (EUS) has been introduced to non-destructive testing for the internal inspection of tubular structure. It is a typical inspection application on layered media with a non-planar interface. Existing research in this area is hindered by two significant limitations: i) the absence of an EUS acoustic field model that is adaptable to the layered medium with non-planar interfaces; ii) the lack of a generalizable analysis and regularity of the EUS acoustic field distribution. This paper derives a refraction model for the tube interface and delay law. By combining these with the angular spectrum model, an analytical acoustic field model for EUS in tubular structure is developed, providing the theoretical framework for acoustic field analysis and interpretation. Subsequently, the geometric and acoustic parameters of the transducer and the tubular object are non-dimensionalized to normalize diverse EUS inspection scenarios. A parametric analysis is then conducted to identify and comprehend the regularity of the EUS acoustic field distribution, which is the main object of this study. The derived EUS acoustic field regularity is validated through an EUS inspection experiment of a tube using control variate method. The quantitative analysis result of the EUS imaging aligns with the expectations from beam analysis. The proposed analytical model is applicable to any scenarios with multi-layered media and with arbitrary interfaces. Its primary strength lies in its intuitive ability to demonstrate the impact of various parameters on the acoustic field distribution. The methodology of non-dimensionalization provides a paradigm for deriving generalizable regularities in the field of ultrasonic testing and imaging.

Design and Computational Modelling of AUV Tunnel Thruster Covers for Efficient Operation

Autonomous underwater vehicles have seen widespread adoption across industrial, scientific, and defence applications. They are typically utilized to perform oceanic mapping, surveillance, and inspection-type missions. Hovering AUVs, used for inspection applications, are over-actuated vehicles incorporating multiple thrusters to enable multiple degrees of freedom control at a low velocity. These vehicles, however, are extremely energy-limited, owing to their restrictive structural design that prohibits large batteries. This necessitates careful hydrodynamic design to best utilize this limited energy storage. Of particular importance are the hydrodynamic propulsion efficiencies of these vehicles. Whilst the external structure of AUV platforms is relatively well-defined and hydrodynamically optimized, one area has seen limited focus and optimization. This is the immediate surroundings of the propulsion geometry and housing. In this body of work, we propose an adaptation to the traditional through-body tunnel thruster geometry of an over-actuated AUV platform. The modification is the inclusion of a retractable internal thruster cover. Subsequently, a comparison is provided between a clean-hull AUV configuration, one with open through-body thrusters, and one fitted with the designed cover geometry. A comprehensive computational fluid dynamics analysis is then converged and assessed using the Reynolds-Averaged Navier–Stokes equations. The drag and local flow fields are determined, where the covers are found to reduce the drag coefficient and total drag of the AUV by 9.51%, primarily due to a reduction of 9.91% in the pressure drag. These findings highlight the increased operational efficiency of the cover geometry and support the adoption of such covers for energy-constrained AUVs.

High‐Resolution X‐Ray Imaging With 0D Organic–Metal Halide Scintillator Featuring Reversed Exciton Trapping

AbstractScintillators, essential for applications in nuclear medicine, radiation detection, and industrial inspection, convert high‐energy radiation into visible light. Manganese (Mn)‐based inorganic–organic hybrid materials are distinguished by their thermal stability, mechanical strength, and flexibility. However, the effects of temperature on Mn(II)‐based hybrid scintillators have not been clearly analyzed, making the elucidation of their temperature‐dependent luminescence mechanisms particularly important. A notable advancement is the synthesis of Mn‐1 nanocrystals (NCs) using methyltriphenylphosphonium chloride (mtppCl) and MnCl₂. These NCs exhibit distinctive temperature‐dependent photoluminescence luminescence: the intensity decreases from 77 to 150 K but paradoxically increases at higher temperatures due to anomalous thermal exciton behavior in the [MnCl₄]2⁻ tetrahedra. Besides, Mn‐1 NCs achieve a detection limit of 1.01 µGyair/s, surpassing medical diagnostic standards and outperforming commercial scintillators such as Bi₄Ge₃O₁₂ (BGO). Additionally, they show exceptional stability under continuous irradiation and can be incorporated into a flexible scintillating film with a resolution of 11.3 lp/mm at an MTF of 0.2. The current study has further refined the luminescence mechanism of Mn(II)‐based materials and optimizes their properties for a wider range of applications.

Machine Vision AI in Railroad Safety: Advanced Inspection Techniques

This article explores the transformative impact of machine vision AI in railroad safety, focusing on advanced inspection techniques. It examines the integration of technologies such as line scan cameras, LiDAR, and IoT sensors in enabling comprehensive 360-degree inspections of railway infrastructure and rolling stock. The article discusses key innovations in continuous monitoring, data processing, and AI-driven analysis, highlighting their potential to significantly enhance defect detection accuracy, reduce maintenance costs, and prevent accidents. Real-world applications in structural integrity assessment and overhead line inspection are presented, along with an analysis of current challenges and future developments in the field. The article underscores the paradigm shift from reactive to proactive maintenance strategies, promising a safer and more efficient future for rail transport.

Image enhancement methods for inspection of planar and non-planar FRP structures using a noise-based microwave NDT inspection system

Microwave nondestructive testing (MNDT) includes inspection techniques that assess a particular material’s health status using low-power and contactless inspection systems. In near-field microwave inspections, imaging results are heavily influenced by the standoff distance parameter, i.e., the physical separation between the microwave sensor and the sample under test (SUT). Variations in the standoff distance during an inspection tend to cause defect masking of disbonds and delaminations in fiber-reinforced polymer (FRP) materials, causing defects to go undetected frequently. An MNDT near-field inspection system using noise waveforms is used to identify engineered internal defects within carbon fiber-reinforced polymer (CFRP) samples. Tactics utilizing Principal Component Analysis (PCA), Stacked Sparse Autoencoders (SSAEs), and an autoencoder network trained in a manner for anomaly detection are used to minimize the effects of standoff distance, reduce defect masking, and increase the ability to identify hidden defects. The samples tested are constructed to possess planar and non-planar geometries, such that the viability of the data-driven image enhancement and standoff distance correction methods are demonstrated with respect to a wide variety of in situ inspection applications.

Comparison of Immersion and Portable Ultrasonic Housing to Quantify the Adhesive Bond Thickness and Sizing of Foreign Objects.

High-performance materials, such as carbon fiber laminates, are costly to manufacture and are often used in demanding environments requiring the use of high-resolution non-destructive testing (NDT) methods to confirm the integrity of the parts. One NDT method that has shown promise for qualifying carbon fiber laminates is the use of immersion ultrasound with spherically focused probes. However, many parts may not be submersible in an immersion tank due to size or material constraints. These parts must be scanned with contact transducers with inferior resolutions or with expensive and messy systems such as bubblers. This research presents the use of a novel housing system that allows for the use of focused immersion transducers in an out-of-tank portable ultrasonic scanning application. This work presents a comparison between scans taken using a custom high-resolution immersion system and scans taken using the presented housing. There are a wide variety of potential inspection applications for this novel system, and the present work focused on two specific applications: the quantification of the spatially varying adhesive thickness in bonded carbon fiber laminates and the quantification of foreign object inclusions in carbon fiber laminates. The results presented show that scans using the portable housing are comparable in quality to scans performed using an immersion system. Specifically, both inspection approaches had an average error of 0.04 mm when quantifying the adhesive thickness of a bonded composite, and for the foreign object detection, the error in quantifying the dimensions of the embedded foreign object was 0.1 mm and 0.2 mm for the immersion system and the portable inspection system, respectively. The demonstration was performed in a laboratory setting, but a discussion is provided for the necessary improvements needed to extend the system for use in field applications.

Enhancing the Sensitivity and Spatial Imaging Resolution of a Hybrid X-Ray Imaging Screen via Energy Transfer at the ZnS (Ag) and a Thermally Activated Delayed Fluorescence Interface.

Novel scintillation materials have played an indispensable role in the recent remarkable progress witnessed for X-ray imaging technology. Herein, a high-performance X-ray scintillation screen was developed based on a highly efficient hybrid system combining inorganic ZnS (Ag) with thermally activated delayed fluorescence (TADF) scintillator materials via an interfacial energy transfer (EnT) mechanism. ZnS (Ag) has a high X-ray absorption capacity and functions as the initial layer for efficiently converting high-energy X-ray photons into low-energy visible light (acting as a sensitizer) while also serving as an energy donor. The TADF component, on the contrary, is an energy acceptor and forms an active scintillating layer. By harnessing TADF chromophores that can efficiently capture both singlet and triplet excitons, our composite material offers a remarkable spatial imaging resolution of 24 line pairs per millimeter, surpassing those of the majority of existing organic and inorganic scintillators. Further, our interfacial energy transfer strategy effectively amplifies the radioluminescence intensity of the TADF scintillator by a factor of 75, offering an outstanding light yield of 38,000 photons/MeV. This advancement represents a remarkable breakthrough in organic X-ray scintillation technology and is a notable achievement within the X-ray imaging field, paving the way for novel applications in medical imaging and security inspection.

Insights into deformation-induced heating: Temperature-strain prediction in SCM440 steel under deformation scenarios

Recently, deformation-induced temperature was found to be predictable based on corresponding strain, enabling broader applications in temperature prediction and strain measurement. This prediction relies on the temperature-strain relationship during deformation. However, the influence of deformation rate has not been addressed, despite most materials exhibiting rate-dependent behavior that affects both material properties and deformation-induced temperature. This study aims to explore the rate-dependent behavior of the temperature-strain relationship and develop a prediction model. Thermal-strain analysis reveals a linear correlation between deformation-induced temperature and plastic strain, with rate-dependent behavior governed by the deformation rate. Relationships among change in temperature difference per unit strain, heat generation rate, and strain rate are utilized to establish the prediction model. The developed model is primarily applied under various deformation conditions in uniaxial tensile testing. Prediction results for temperature and strain closely match measured values, with an overall difference within 5 % in most cases, except at slow deformation rates where strain estimation is less effective due to dominant heat convection. The model was also applied to three-point bending tests with different artificial-defect specimens, revealing predictable temperature results with a maximum difference of 7 % and estimated strains agreeing with a 5 % error band in the initial 20 % of plastic strain development. Both findings underscore the crucial influence of heat transfer on prediction accuracy, suggesting its necessity in further improvements. This study provides insights into temperature-strain prediction driven by deformation-induced heating, offering guidance for temperature-strain prediction methods and paving the way for applications in thermal-based inspection and temperature-based strain measurement.

Bright compact ultrabroadband source by orthogonal laser-sustained plasma

Laser-sustained plasma (LSP) source featuring high brightness and broadband spectral coverage is found to be powerful in various fields of scientific and industrial applications. However, the fundamental limit of low conversion efficiency constrains the system compactness and widespread applications of such broadband light sources. In this paper, we propose an innovative orthogonal LSP to break through the conversion efficiency limitation. Driven by the elevated conversion efficiency from absorbed laser power to ultraviolet (UV) emission, a compact broadband source (250–1650 nm) with UV spectral radiance exceeding 210 mW/(mm2⋅sr⋅nm)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${mW}/({{mm}}^{2}\\,\\cdot\\, {sr}\\,\\cdot\\, {nm})$$\\end{document} is achieved with >100 W pump laser. With the plot of a two-dimensional refractive index model, we report an important conceptual advance that the orthogonal design eliminates the influence of the negative lensing effect on laser power density. Experimental results unambiguously demonstrate that we achieve a bright compact UV-VIS-NIR source with negligible thermal loss and the highest conversion efficiency to our knowledge. Significant enhancement of 4 dB contrast-to-noise ratio (CNR) in spectral single-pixel imaging has been demonstrated using the proposed ultrabroadband source. By establishing the quantitative link between pumping optics design and plasma absorption, this work presents a compact broadband source that combines superior conversion efficiency and unprecedented brightness, which is essential to high-speed inspection and spectroscopy applications.

Efficient and accurate semi-supervised semantic segmentation for industrial surface defects

Deep learning-based defect detection methods have gained widespread application in industrial quality inspection. However, limitations such as insufficient sample sizes, low data utilization, and issues with accuracy and speed persist. This paper proposes a semi-supervised semantic segmentation framework that addresses these challenges through perturbation invariance at both the image and feature space. The framework employs diverse perturbation cross-pseudo-supervision to reduce dependency on extensive labeled datasets. Our lightweight method incorporates edge pixel-level semantic information and shallow feature fusion to enhance real-time performance and improve the accuracy of defect edge detection and small target segmentation in industrial inspection. Experimental results demonstrate that the proposed method outperforms the current state-of-the-art (SOTA) semi-supervised semantic segmentation methods across various industrial scenarios. Specifically, our method achieves a mean Intersection over Union (mIoU) 3.11% higher than the SOTA method on our dataset and 4.39% higher on the public KolektorSDD dataset. Additionally, our semantic segmentation network matches the speed of the fastest network, U-net, while achieving a mIoU 2.99% higher than DeepLabv3Plus.