This PDF file contains the front matter associated with SPIE Proceedings Volume 12536, including the Title Page, Copyright information, and Conference Committee lists.

Active thermography is a nondestructive evaluation (NDE) method. Active thermography's main advantage is its ease of use. While the technical aspect has evolved with time, it has never become a primary NDE tool; interpretation is lagging. We introduce a new approach based directly on the governing differential equation. By using relationships between wave propagation and thermal excitation's diffusive propagation, one can transform from diffusive propagation to wave propagation. Having wave-like behavior produces simple-to-interpret images and allows to use advance techniques developed for wave propagation. In this paper, we demonstrate the use of active thermography to detect defects in carbon composite, flaws in welding, and breast vascular imaging.

Infrared thermography is a condition monitoring technique that, from a measurement of the radiant heat pattern emitted by a material, is able to determine regions or points of increased or reduced heat emission that can indicate the presence of an imperfection in the investigated material. The result of an infrared thermographic investigation is a sequence of thermograms or thermal images, in other words a picture of temperature, that can be further processed for qualitative and quantitative purposes. Such images can be presented in either false color or black and white format. In the present work, the philosophy and history of thermal–infrared imaging are reviewed. Moreover, the different evaluation approaches (passive and active), as well as many standards related to infrared thermography are discussed. Finally, various applications of the transient thermography approach are briefly presented

The dominant signature of thermal breast images is vascular heat. We analyzed dynamic thermograms as a response to external stimuli. The vascular response to temperature stimulus is affected by vasomodulation. We are using the Brazilian visual lab mastology data set. The recorded data is analyzed by converting the diffusive heat propagation into a virtual wave and identifying the reflection using component analysis. We identified two classes of images: 1. The dominant one is a reflection at the same polarity as the applied temperature stimulus (vasoconstriction). 2. A second reflection at twice the depth and reverse polarity (vascular reflection).

As part of on-going research into filament-based Material Extrusion (MEX; aka Fused Filament Fabrication) polymer additive manufacturing, a modified industrial printer equipped with an infrared (IR) window was used to perform temperature measurements using infrared thermography. The IR window allowed the measurements to be taken without interfering with the printing process; i.e. with the printer door closed, and with build plate and build volume heated to printing conditions. The IR camera, lens and IR window system was calibrated using a black body calibration source, and the effects of the IR window on the temperature measurements were determined. This paper covers the calibration process and steps required to obtained high accuracy temperature measurements, within 3°C of the printer’s build plate. In addition, examples of measurements obtained from prints of various polymers are presented.

Laser Metal Deposition (LMD – DED) is an Additive Manufacturing (AM) process that allows the production, and the repairing of 3-D printing metallic parts. This kind of process is complex because it involves a high number of parameters and variables, some of which are difficult to control. The choice of these process parameters influences the quality of the final component, in terms of its mechanical properties. Furthermore, the production of components free of defects and, therefore, reliable is still a challenge and for these reasons, the online monitoring of this kind of process is becoming essential. In this regard, contactless sensors such as infrared cameras, pyrometers, bolometers, and optical cameras with sensors like CMOS, CCD, or photodiode, are usually used. This work focuses on the online monitoring of the Laser Metal Deposition (LMD – DED) process during the production of different coupons in Nickel-based alloy Inconel 718, using thermal data deriving from the analysis of the signal acquired by IR Focal Plane Array sensors. An experimental plan based on a customized Design of Experiments (DOE), in which the main controlled process parameters were the laser scanning speed, the powder flow rate and the laser power, together with some useful their combinations, was carried out. The experimental setup consisted of two microbolometer thermal sensors, one integral with the scanning laser head and the other one laterally and fixed respect to the deposition platform was adopted to monitor the process during the manufacturing of all coupons. By means of ANOVA and regression models, a correlation between process parameters and extracted thermal features was provided. At the end of the process, some hardness tests and macrographs on specific coupons were also performed to evaluate the mechanical properties of the material in correspondence with different combinations of the process parameters. A statistical approach was also employed to describe the results to directly correlate the process parameters and the final mechanical properties with the extracted thermal features.

Flash thermal diffusivity measurements were obtained on additively manufactured Ti-6Al-4V disk shaped specimens with various process parameters. For additively manufactured metal parts, processing parameters such as laser power and scanning speed are critical to ensure the desired microstructure. For this study, the laser powder bed fusion process parameters were changed at various angular sections on a 21 mm diameter and 3.0 mm thick disk. The measurement of thermal diffusivity was performed by fitting a 1-dimensional thermal model to the data pixel by pixel to produce an inspection image. The image revealed the detection of defects such as lack of fusion porosity and areas of aggregated porosity. The thermal diffusivity imagery was compared to immersion scan ultrasonic and X-ray computed tomography (CT) measurements for validation. Based on these results, additional samples were investigated using a single side thermal inspection technique to detect lack of fusion porosity and near surface voids.

Recent years have seen substantial growth of pulsed thermography as a popular non-destructive testing tool to characterise surface and near-surface defects for both metals and composites. The current research focus is on material characterisation to determine material thermal properties such as thermal diffusivity. However, most of these studies have focused on using the reflection mode of pulsed thermography where data is captured from the front wall of the specimen at which the heat flux is applied. The transmission mode, where the heat flux is applied on the front wall and data is acquired at the back wall, has not been investigated as comprehensively as the reflection mode. Research has indicated that the transmission mode is able to detect defects deeper into the specimen when compared to the reflection mode however, it has mainly been used as an indicator to detect defects and not to quantify the depth of subsurface defects within the specimen. This study develops a finite element model using the commercially available software COMSOL to investigate material characterisation using both the reflection and transmission modes of pulsed thermography. A finite element model of a thin steel plate was first created followed by the application of a heat flux to the front surface for the model. The solvers from the software were used to record the temperature variation on both the front and back surfaces. Results show that the thermal contrast curves on the backwall were less sensitive to changes in depth compared to changes in defect size. Furthermore, this study provides motivation for conducting a more in-depth study of the through transmission thermography to better understand its capabilities.

The study of the response of a temperature field (recorded from IR cameras) to a laser spot heating is increasingly used for NDE (Non Destructive Evaluation) applications. The most classical type of application is to use the flying spot in order to detect vertical cracks and/or to measure the in plane thermal diffusivity in relation to the observation plane of opaque materials. But several other ways of applications are presented here related to tomography and also super resolution. Instead of opaque materials applications, the tomography is using the principles of the flying spot. It consists in an indirect detection on an intermediate layer (the thermoconverter) that can convert a wide range of radiation from the spot. The objective of super-resolution can also be implemented with flying spot in order to circumvent the low spatial resolution of IR imaging systems. Such methods consider spots whose diameter is small compared to the size of the pixel. Some applications of our team will be shown with multiscale considerations.

Fatigue characteristics of carbon fiber reinforced plastics (CFRP) are an important factor in predicting the design life of structures such as wind turbine blades. Due to the anisotropic and heterogeneous structure of CFRP, design life prediction under fatigue stress is a challenging task. The authors proposed an approach to assess fatigue damage at the microscale using a laser-periodic-heating method based on lock-in thermography (LIT). It assumes that the rate of fatigue progression correlates with the number of micro-voids, micro-cracks, and fiber/resin delamination that inhibit heat transfer. The effective thermal diffusivity in the out-of-plane direction of CFRP laminates, which were tested under fatigue loading, showed a decreasing trend with the number of loading cycles.

The growth of application using multi-material assemblies in the transportation industry has resulted in structural adhesive becoming increasingly used. These assemblies, in addition to be subjected to mechanical stresses, are also exposed to harsh environments throughout the life of vehicles with temperature variations, high humidity and exposure to de-icing salts and fluids. While such assemblies are tested for mechanical strength and fatigue resistance, it is also critical to identify the failure mode of adhesive bonds to ensure that proper actions are taken to prevent catastrophic failure. Despite the obvious need to qualify the adhesive failure modes, this task is typically relegated to a semi-quantitative analysis of cohesive/adhesive failure ratios based on the visual inspection of an experienced eye, which along with the use of adhesives of various colors, inevitably introduces variability in the qualifying process. Moreover, the characterization of adhesive performance typically involves analysing hundreds of coupons and while the general failure types are known: bulk of the adhesive (cohesive failure), substrate/adhesive interface (adhesive failure) and near-interface; quantification on each coupon suffers from inaccuracy and better means are needed. In this work, we introduce the use of pulsed thermography (PT) as a repeatable and objective solution to quantify failure modes of metal-adhesive-metal assemblies by harnessing the fundamental differences in thermal properties of the two materials. It is shown that the inspection performed in through transmission mode allows for the distinction of the various adhesion failures.

Scanning inductive thermography is a non-destructive inspection technique, which is suitable for detecting surface defects in long metallic work pieces. The work piece is moved below the inductor and the infrared (IR) camera, which is recording the surface temperature during the motion. To evaluate such measurements via phase image the recorded infrared image sequence must be reorganized according to the scanning speed. If the speed is not constant during the motion (e.g., due to manual scanning), visual fiducials (AprilTags) can be used in the camera’s field of view to register shifts between consecutive images. The main contribution of this work is image fusion, applied to scanning inductive thermography, combining the results of an infrared camera and a visual camera. An uncooled IR µ-bolometer camera with a thermal time constant of 8 ms is used for the infrared spectrum. Information of the motion speed during the scanning is acquired by the image registration, and it is used to deblur the IR image sequence before the evaluation to phase image via Fourier transform is performed. A second camera records the scanning process in the visual range. Using the AprilTags for registration, a panoramic view of the specimen is created. The results from both cameras are superimposed to improve the interpretation and localization of defects.

In this work, a thermographic non-destructive method has been investigated with the aim to estimate the mechanical properties of steels through thermal diffusivity measurements. Considering the anti-correlation between hardness and thermal diffusivity in steels, a thermographic procedure based on pulsed laser thermography has been used to investigate the technique's capability to discern different phase percentages in boron steel. After a discussion between two different methods to select the time window for thermographic measurements, an experimental investigation on nine specimens with three different phase percentages has been carried out, and a correlation among mechanical properties and phase percentages with different values of thermal diffusivity has been presented.

For bonded composite materials, an accurate characterization of the adhesive bond line is needed to predict failure modes and fracture toughness. In this paper, bond line thickness was estimated from data obtained using through transmission flash thermography. The forward model that predicts back surface temperature is based on a three layer heat diffusion equation with varying diffusivity and flux boundary conditions. The corresponding inverse problem of estimating bond line thickness from measurement data was solved using a Bayesian approach that assumed Gaussian priors for the bond line thickness and thermal diffusivity of the adherends. Finally, the outputs of the thermography based method were compared to measurements that were collected using a micrometer and ultrasound testing.

Composite materials have extensively been used in the aerospace industry. There are several inspection methods to ensure the safety of these composites; pulse thermography (PT) is one of the most promising ones. Both reflection and transmission modes of PT could be applied. However, few studies reveal the advantages and disadvantages of these two modes. This paper presents a quantitative comparison of pulsed thermal imaging in two modes of reflection and gradual heat transfer in carbon fibres. Experimental work was conducted on carbon fibre-reinforced plastic (CFRP) samples with different thicknesses and thermal images were recorded in both modes. Thermal images were further processed using statistical analysis and machine learning algorithms. Comparing the results from both modes, there is a marked improvement in the accuracy when the reflection mode is employed.

Lynntech is seeking to develop real-time realistic nondestructive evaluation (NDE) and structural health monitoring (SHM) physics-based simulations, and automated data reduction/analysis methods, for large datasets. Recently, computational efficient Neural Network based simulations have demonstrated the possibility to synthesize data with an orders-of-magnitude increase in speed compared to standard computational techniques [1,2]. In this contribution, we report our initial experimental results for our Generative Adversarial Network for Realistic Physics Simulations, or GAN4RPS.

This work presents a convolutional neural network (CNN), trained on simulated data and used for the detection of cracks resulted by inductive thermography measurements. In inductive thermography the sample under study is heated with a short heating pulse and an infrared (IR) camera records the emitted surface radiation during both heating and cooling. The recorded IR sequence is then evaluated to a phase image using Fourier transform. In phase images, short surface cracks become visible due to the hot spots around the defect tips and due to the low phase value along the crack line. For the training of a deep neural network many images are necessary, which should be different from the images to be evaluated. This is why FEM simulations have been carried out varying crack length, depth and inclination angle. Additional Gaussian noise and augmentation have been added to these simulated images before using them to train a CNN. Samples with real cracks along a weld have been created in Inconel 718, and the CNN, trained on the simulation results, has been used for semantic segmentation of these real samples’ phase images, in order to identify the defects. Additionally, the samples were investigated by computer tomography, and this 3D information of the cracks is compared to the phase image results.

We address the characterization of defects that behave as heat sources in nondestructive thermographic techniques. First, we consider tilted heat sources of rectangular shape. We calculate the evolution of the surface temperature distribution generated in a short excitation. For the characterization, we make use of the thermogram obtained at the end of the excitation and the temperature evolution at the center of the early heated region. A sensitivity analysis indicates that the optimum excitation duration corresponds to a thermal diffusion length similar to the depth of the deepest end of the heat source. By fitting synthetic data with added noise, we analyze the influence of the signal to noise ratio and the inclination of the heat source on the fitted parameters. Inductive thermography experiments carried out on insulating samples with embedded Cu slabs confirm the ability of the method to characterize tilted heat sources and indicate that the penetration is the most elusive parameter. In the second part, we present a methodology to deal with horizontal heat sources of unknown geometry. We average the thermogram obtained at the end of the excitation in circumferences concentric with the center of the heated region. This averaged radial profile, together with the temperature evolution at the center of the heated region is fitted to a circular heat source model. Fittings of experimental data taken on samples with horizontal rectangular Cu slabs allow determining the area with accuracy better than 20% and the depth with 10%.

Non-destructive testing applications are one of the most crucial steps in maintaining aviation activities in a profitable and timely manner. Infrared thermography (IRT) is a functional technique that uses the thermal radiation/temperature relationship on the inspected structure to ensure efficient detection, in particular when the defect is on a surface or near the surface. Ultrasonic (UT) inspection is an alternative technique that uses the propagation of ultrasound waves into the inspected material for defect detection. While IRT suffers from detectability problems with the increasing structure thickness, UT has inspection limitations on the surface or near-surface area according to applied frequency. Overcoming these limitations of individual methods with the synergistic effect of the fusion approach might provide more precise and apparent marks for defect detection. In this study, decision-level fusion has been applied using the maximum fusion rule to combine unimodal inspection data and compare. Impact-defected Carbon Fiber Reinforced Polymer (CFRP) composite structures have been chosen to represent aerospace structures. The results show the proposed fusion approach is promising in terms of identifying defect location, size and depth to inform further stages such as repair.

Recently, we have successfully tackled the challenge of measuring the 3D shape of uncooperative materials, i.e., materials with optical properties such as being glossy, transparent, absorbent, or translucent. By projecting sequential thermal fringes in the long-wave infrared (LWIR) combined with a stereo camera setup in the midwave infrared (MWIR), we were able to three-dimensionally record object shapes within one second. However, in many applications, e.g., for 100 % quality assurance, even shorter measurement times are required. To achieve camera frame rates higher than 125 fps at room temperature, Max Planck’s law of thermal emission teaches us a change in the camera spectral range from MWIR to LWIR. If irradiation and image acquisition have to run in parallel, the camera chips must therefore be protected against the radiation projected by the CO₂ laser at a wavelength of 10.6 µm. Appropriate filters have been available only recently. In this contribution, we present our high-speed LWIR 3D sensor. The work includes a characterization of our setup regarding its measurement accuracy and speed. The results are compared to the performance of previous thermal 3D sensors. We show 3D measurement results of static objects as well as of a dynamic measurement situation of a transparent object. Furthermore, we demonstrate that our setup enables us to extend the measurability of material classes towards objects with high thermal conductivities.

The current context is for the development of renewable energies. These energies are often intermittent and require storage systems in particular for heat. Peritectic materials are excellent candidates for heat storage. They are suitable for a reversible chemical reaction near the time of a reversible solid-liquid phase transformation. Classic differential scanning calorimetry methods turn out being insufficient to characterize such materials. One of the reasons is related to the ignorance of the local heterogeneities and to the non-uniform distribution of the source terms of phase change when the material does not behave as a pure substance. It is therefore proposed here to develop methods consisting in measuring the temperature field evolution of these kinds of materials in the vicinity of the phase change phenomenon, by using the IR thermography. Some preliminary results are here presented.

Ultrathin absorber can improve specific detectivity and response time of microbolometers. We experimentally demonstrate an average absorptance of 48% +/- 2.5% in the 8-13 microns (769-1250 1/cm) spectral range in 10nm thick titanium nitride (TiN), a value bordering on the 50% fundamental absorptance limit for a suspended thin film. Such absorptance is close to the fundamental limit of 50% for free-standing ultrathin films.

Mobile robots performing aircraft visual inspection play a vital role in the future automated aircraft maintenance, repair and overhaul (MRO) operations. Autonomous navigation requires understanding the surroundings to automate and enhance the visual inspection process. The current state of neural network (NN) based obstacle detection and collision avoidance techniques are suitable for well-structured objects. However, their ability to distinguish between solid obstacles and low-density moving objects is limited, and their performance degrades in low-light scenarios. Thermal images can be used to complement the low-light visual image limitations in many applications, including inspections. This work proposes a Convolutional Neural Network (CNN) fusion architecture that enables the adaptive fusion of visual and thermographic images. The aim is to enhance autonomous robotic systems’ perception and collision avoidance in dynamic environments. The model has been tested with RGB and thermographic images acquired in Cranfield’s University hangar, which hosts a Boeing 737-400 and TUI hangar. The experimental results prove that the fusion-based CNN framework increases object detection accuracy compared to conventional models.

A drone-based inspection system that can move “freely” around an aircraft to perform the inspection of all the areas of interest in a fast and effective manner can have significant impact in reducing inspection time and cost. However, active thermography inspection using drone is challenging because the drone carrying the optical and thermal cameras is subjected to vibration and undesired motion. Since active thermography relies on the pixel temperature evolution over time, an unstable thermal video from a flying drone can cause error in the output results as any movement between the acquired images will affect the pixel position in the successive frames and thus disrupt the monitoring of the temperature evolution. This paper presents the outcome of experimental runs, where a commercially available drone equipped with both thermal and optical cameras was used to inspect a helicopter Main Rotor Blade (MRB) in a laboratory environment.

The need for energy efficiency in building components is constantly increasing, as the current legislation is constantly pushing to decrease the overall energy demand and increase the share of renewable energies in buildings. In this framework, the use of Phase Change Materials (PCM) in construction elements is a powerful tool both to reduce the energy consumption and to improve the integration of renewable energy sources. PCM could be applied both to the building envelope and to the building energy system. In the latter, PCM are typically applied in energy storage elements, exploiting their ability to accumulate and release heat according to the energy need of the building. In this work, the thermal behavior of PCM under dynamic conditions is investigated using three methods based on infrared thermography. The accurate knowledge of the thermal behavior of PCM, obtained through the experimental measurements, is crucial for the design of energy storage elements, as the thermal behavior of PCM-based storage components impacts the overall design of the building energy system.

Active thermography was used for characterization of multi-layered paintings panel structures and analysis of defects caused by aging and environmental effects. Pulsed Thermography, setup was applied to provide and inspect a dynamic thermal response, which was recorded by mid- and long wavelength infrared TELOPS cameras. Control, synchronization and data analyses were provided by an automated Professional software. Active thermography was demonstrated as being appropriate for characterization of various defects on painting layers and detection of under−drawings, pentimenti and canvas. Such multispectral approach provided simultaneous complementary information on the specimen under inspection.

Thermal imaging and sensing market include devices relying on microbolometer arrays, thermopiles, or pyroelectric sensors. Between 2019 and 2022, demand peaked pulled by fever detection systems used to fight Covid-19 pandemic and remained significant in years after despite lower sales in 2022 due to global chip shortage situation. After ramping-up their production in 2020, Chinese players eventually doubled there share of thermal imager global shipments and managed to stabilize it by adapting their offer. Next wave of demand for thermal imagers might come from automotive for AEB (Autonomous Emergency Braking), with the potential help of safety regulations. Meanwhile thermopiles have been recently integrated in a new model of consumer smartwatch.

This study developed an end-to-end procedure to overcome common issues faced during the analysis of passive infrared thermography (IRT) sequences from outdoor concrete infrastructures. The processing pipeline includes the automatic pre-processing of raw thermograms, data cleaning and organization, image adjustment, and sequential image registration. One image registration method was implemented, and the results were evaluated using the Euclidean distance metric. Furthermore, the resulting sequences were processed using signal processing techniques to increase the detectability of the defects. The results from outdoor IRT surveys over two academic samples are presented, where one image per minute was taken for 24 hours on slabs and columns representative structures. By addressing the difficulties encountered during the analysis of passive IRT sequences, our contribution can broaden the spectrum of the application of IRT for the condition assessment of concrete infrastructure.

Subsurface delamination is one of the main damage mechanisms that affect the integrity of structural components. Its detection effectively and reliably is a crucial step for in-service assurance and avoiding further accidents. In that scenario, it is necessary to explore alternatives to current inspection practices so that effective non-destructive methods can be implemented in the field. This paper investigates the application of infrared thermography and ground penetrating radar to detect and evaluate subsurface delamination in concrete components. To this aim, laboratory specimens made of reinforced concrete with Teflon inserts to simulate internal delamination are inspected with the step-heating approach. The IR data collected during the heating and cooling process is then evaluated and processed with pulsed-phase thermography and principal component regression. The extension or severity of the delamination is then evaluated with ground penetrating radar. The results will be evaluated to determine the applicability of the methods at larger scales.

We present a model for laser-spot lock-in thermography that takes into account heat conduction between the sample and the air. We show that the effect of the coupling is only significant in low diffusivity materials. In this case, the dependence of the surface temperature distribution with both the conductivity and the diffusivity of the material provides a method to measure both properties simultaneously. We present experiments performed on several materials confirming that it is possible to determine both the thermal conductivity and diffusivity of thermal insulators with good precision and accuracy by laser-spot lock-in thermography.

Cannabis pests such as spider mites, thrips and aphids cause enormous damage to crops. The lack of regulations regarding pesticides that can be used on cannabis, is a source of concerns for growers and the federal government since no regulations are yet in place. In order to favor prevention, as opposed to curing the problem, we developed a system based on convolutional neural networks and transfer learning techniques trained on multispectral images that is capable of detecting the early state of parasitic stress on cannabis plants instead of giving an operator the difficult task of visually detect whether the plant is already infected or not. This gives the grower time to remove pest-infected plants before they spread throughout the whole crop.

Dynamic thermography is a well-established diagnostic tool for breast cancer screening that can be used in conjunction with mammography and clinical breast examination (CBE). Thermographic imaging biomarkers, known as thermomics, have been shown to detect vasodilation in breast tissue, indicating abnormalities and lesions. Heterogeneous thermal patterns also reveal angiogenesis or the formation of new blood vessels. This study applied thermal imaging biomarkers, and thermographic imaging, for breast cancer screening. We applied two low rank embedding approaches, Gaussian and Bell embedding, to obtain the optimal thermomics with the help of elbow method, which resulted in finding breast thermal heterogeneity. Non-negative Matrix Factorization (NMF) was used to create a low-ranked representation of thermal images. High dimensional radiomics and thermomics were then extracted, and feature abundance was reduced using spectral clustering. The best results of the Deep semiNMF with Bell embedding method combining clinical information and demographics yield 81.6% (±3.9%). The model was trained with constant hyperparameters setting across the comparison to predict abnormality, and the results demonstrated promising preliminary performance. Optimal biomarkers have the potential to preserve thermal heterogeneity, leading to early detection of breast cancer, and can serve as a non-invasive tool to aid CBE. Codes corresponded with this proceeding can be found at the following GitHub repository: https://github.com/BardiaYo/SPIEThermosense2023.git

Infrared thermography (IRT) technology has evolved during the last decade extending its capabilities to the industrial and infrastructures level. Because of the necessity to perform regular inspections of in-service assets such as bridges, it becomes necessary to investigate and develop efficient inspection technologies that can adapt to the needs of the industry. So, IRT is considered an effective technology to perform NDE. However, its integration with other sensing technologies such as visible cameras still needs to be further investigated so the inspection and maintenance strategy can be more effective when inspecting large structures and assets. Hence, this project investigates fusion strategies and proposes a multi-modal processing pipeline using a deep learning-based panorama stitching method for infrared and visible images. Then, an image registration method to fuse infrared and visible images so identification of defects in visible and thermal spectra becomes more efficient.

In recent years, many studies have focused on using deep-learning approaches for automatic defect detection in the thermographic inspection of industrial and construction components. Deep Convolutional Neural Networks have proven to perform remarkably on thermal defect detection. However, their convergence and accuracy are heavily associated with having a large amount of training data to avoid overfitting and ensure reliable detection. Unfortunately, the number of available labeled thermal datasets for inspection-related applications is very limited. One of the practical approaches to address this issue is data augmentation. This paper proposes a novel approach for augmenting simulated thermal defects on regions of interest using coupled thermal and visible images. The visible images are employed to extract regions of interest in both modalities using a texture segmentation method. Later, the introduced method is used to augment thermal defects on thermal images.

With new thermal imaging technologies emerging each day, thermal imaging sensors are showing of a trend of decreasing pitch lengths. In this paper, the statistical advantages of using different pixel pitch sizes for LWIR uncooled thermal imaging sensors are investigated. Comparisons are made between 12um and 17um pitch sizes and the results are evaluated using the Johnson criteria, PSNR, MSE and observatory methods. It is found that 12um pitch sizes give better results in terms of both the statistical methods and observatory methods. Edge sharpness and PSNR values are also increased.