10th Conference on Industrial Computed Tomography (iCT) 2020, 4-7 Feb, Wels, Austria

In this paper we present a method to compensate for scattered radiation artifacts. Using a newly developed workflow to determine the scattered radiation, scatter artifacts can be dramatically reduced to a level, where further evaluation of the data becomes possible with state-of-the-art algorithms. The method intrinsically corrects for all kinds of scatter problems that may occur in a CT system including scattering inside the object. We will demonstrate the approach for the inspection of large aluminum castings. We will show that using this method an image quality can be achieved, which is otherwise only available using fan-beam tomography with line detectors. In comparison to line-detector based fan-beam CT, scans can be achieved in significantly reduced measurement times in the order of few minutes instead of many hours.

    

   

3D X-ray Computed Tomography (CT) is increasingly being used for non-destructive inspection of objects. Conventional CT inspection requires many projections, typically spanning 360 to reconstruct a 3D image of the object, which is then segmented and subsequently compared with the reference computer-aided design (CAD) model. Such an inspection flowchart, however, is a time inefficient procedure, not suitable for inline inspection. To overcome this problem, we directly compare the measured projections with simulated ones from the CAD model. To do so, the simulated projections need to be created with the same acquisition geometry as the measured ones. When an object is inserted on a scanning system, its orientation may vary with respect to the default CAD model orientation. For this reason, 2D/3D registration between the CAD model and the measured projections of the real object is necessary. In this paper, we present a deep learning based method to accurately estimate the 3D orientation of an object from one projection image.

    

   

X-ray computed tomography (CT) allows for visualization of the interior of solid objects in a non-destructive and non-invasive manner. However, producing high-precision measurements takes a long time because thousands of sharp transmission images are required to reconstruct CT volumes. To address this problem, we propose a CT measurement method based on Convolutional Neural Networks (CNNs) that yields sharp transmission images by deblurring blurry ones. In this method, first, blurry images are obtained in a short measurement time, then they are deblurred by CNNs with fine-tuning and integrated by linear interpolation. This method shortens the measurement time indirectly because the process related to the CNNs is fast with GPUs and the blurry images with low levels of noise intensity do not require a long time. Besides, the fine-tuning may improve the output images’ sharpness. According to our experimental results, the proposed method is fast and can maintain the quality of data to a certain extent.

    

   

Industrial tasks that produce hundreds of terabytes of data per year require efficient evaluation tools for the purpose of saving valuable resources such as time or highly trained personnel. Therefore, processes with high automation potential need to be identified and put into practice. Before deploying a new tool, however, thorough investigation is required. It needs to be tested whether the achievable degree of automation of such a new tool yields results which are on one hand reliable and on the other hand of sufficient quality. This work evaluates the applicability of artificial neural networks (ANNs) to detecting micrometer-cracks and delaminations in reconstructed computed tomography (CT) volumes of previously stressed carbon fibre reinforced polymer (CFRP) pressure rods. Common network architectures, varying network parameters and different training sets have been investigated and compared in order to determine the combination that performs best. The constellation (network, hyperparameters, training data) that performed best reached an average precision (AP) of 0.87. Based on the rather small data set of approx. 1.6E3 images and the unstructured nature and great diversity of the investigated features, this result can be regarded as very good. The detected feature sizes ranged from approx. 100 micrometer to a centimeter in length and from tens of microns to a few hundred microns in width. The results suggest that artificial neural networks have the potential to be used reliably for the automatic detection of micrometer-cracks and delaminations in CT volumes of CFRP pressure rods, provided the data set used for training is large and diverse enough and the network is being updated when new data is available.

    

   

When using X-ray computed tomography in coordinate metrology, methods for the determination of surface points at subvoxel accuracy (hereinafter subvoxeling) are essential. In recent years, the demand to measure objects consisting of several materials has increased. While subvoxeling edge detection operators proven for mono-material applications are rarely suitable for multi-material specimens, a number of algorithms enable segmentation at voxel resolution. To overcome the gap, a new method has been developed. In a first step, volumes of material probabilities are calculated using a (pre-)segmentation while obtaining the implicitly given subvoxel-accurate information. The surface points can then be calculated using common edge detection operators. Initial results of the validation based on selected examples are presented.

    

   

We propose a new segmentation method for separating parts of a mechanical assembly in a CT volumetric image. Part segmentation is generally based on edge detection with gradient norms of the CT volumetric image, which often fails when parts made of the same material come into close contact. This is because an image is blurry between the parts with low gradient norms and indistinct edges. In order to enhance weak edges at such faded part boundaries, our idea is to enhance the gradient norm in a CT volumetric image. To find such faded boundaries, auxiliary information needs to be introduced. In this paper, two practical cases of the faded boundaries in plastic and welded assemblies of sheet metal parts are studied. For the former case, CAD mesh models of target parts are used. The faded boundaries are found in blurred regions on the CAD mesh boundaries. For the latter case, the thickness value of the sheet metal parts used, and it is assumed to be constant. The faded boundary can be located at the distance of the thickness from the outer surface of the part. By enhancing the gradient norms in those faded boundaries, the assembly of parts made of the same material can be segmented. The algorithms are implemented, and experimental results are demonstrated.

    

   

Scan time is a crucial factor in industrial computed tomography (CT). With modern computers performing reconstructions in seconds or minutes, the duration for the projection acquisition, which is often in the order of hours, becomes the limiting factor. Therefore, it is desirable to reduce the number of projections in order to cut costs or allow for new applications, such as inline CT. So far, many different approaches mostly targeting the reconstruction method have been performed to optimize the outcome for undersampled data, but only few work has been conducted concerning the scan planing. Nevertheless, the commonly used circular standard-trajectories for image acquisition are highly inefficient since they do not consider the geometry of the inspected objects. It has already been shown that using workpiece-tailored instead of conventional circular trajectories can shorten scan times significantly. In this work we present an observer model based algorithm that has previously been applied in medical and industrial studies and extend it in several aspects. Subsequently, we focus on comparing different observer models that have not been used for this optimization yet and evaluate their performance for identifying valuable projections that contribute most to the reconstructed image.

    

   

The Discrete Algebraic Reconstruction Technique (DART) [1] is a well-known method to reconstruct images from a set of X-ray projections acquired from objects that consist of only a small number of materials. For such materials, DART has been shown to lead to high quality images, even when the number of available projections is small or when the projections are acquired in a limited angular range. The core idea of DART is to reduce the size of the reconstruction problem by iteratively updating only those pixels that are likely to be misclassified. However, DART as proposed in [1] updates the image pixels independent of the material. This paper presents an improved pixel update strategy by introducing a probability map that measures the classification accuracy of each pixel based on its grey value evolution throughout the iterations. Through simulation experiments, we show that, compared to DART, our proposed method either improves upon convergence speed or on quality of the reconstructed image.

    

   

A 950 keV X-ray source with small focal spot using a linear accelerator has been developed. By installing a focusing mechanism of electron beam, smaller focal spot (0.3 mm in FWHM of the line spread function) is successfully achieved at 950 keV. 9.3 GHz (X-band) radio frequency is used for accelerating electrons in order to make the size of X-ray source smaller, compared to lower radio frequency such as 3.0 GHz (S-band). Size and weight of the developed X-ray source are W 600 mm × D 800 mm × H 500 mm and 230 kg, respectively. Clearer transmission images using a rectangular groove sample are obtained with small spot setup, compared to conventional one. CT images are also obtained for an engine block, and we demonstrated that advantage of 950 keV X-ray source for CT imaging such as improvement of interior visibility and decrease of artefact, compared to a 450 kV X-ray source. Therefore, we are convinced that this 950 keV X-ray source is widely useful for industrial non-destructive inspection of large structures, because of compactness, high transmission ability and spatial resolution.

    

   

Significant efforts are currently ongoing in X-Ray imaging to provide multimodal imaging systems, targeting better sensitivity and specificity for both biomedical or non-destructive testing (NDT) applications. X-Ray Phase Contrast Imaging (X-PCI) shows great capability to differentiate elements with similar absorption. For example, in the medical field, knowing the chemical composition of breast microcalcifications would help to differentiate malign and benign tumors. The composition can be determined from the measurement of the phase as the optical index of materials is directly related to the composition. We propose a novel, high-sensitivity X-ray quantitative phase imaging system based on a Hartmann wavefront sensor. The system provides high resolution (20?m without magnification) and high sensitivity (~100 nrad), and is compatible with tomographic experiments using both synchrotron beamlines or laboratory sources. We present here our first X-PCI prototype as well as the first images obtained. We also present an alternative design based on the same approach, providing larger field-of-view at the cost of some trade-off regarding resolution and sensitivity and the first tomographic results obtained with this imaging system.

    

   

As bonded composite repairs are gaining importance for modern civil aircraft, it is necessary to investigate such repairs in detail under relevant environmental conditions. In this work, X-ray computed tomography (XCT) was performed on bonded repairs for carbon fiber (CF) reinforced epoxy matrix composites before and after cyclic conditioning between dry/cold and hot/wet conditions. In detail, high resolution XCT scans in absorption contrast (AC) mode as well as Talbot-Lau grating interferometer (TLGI)-XCT scans to obtain additional differential phase contrast (DPC) and dark-field contrast (DFC) modalities were performed. The repair bonds' constituents could be identified by the high resolution XCT scans. Additional information about fiber alignment, at least of the fiber bundles, could be extracted from the TLGI-XCT scans. In order to gain detailed information on certain specimen features, specimen dimensions were reduced for additional high resolution XCT scans.

    

   

MicroCT has become a powerful and widely used tool for non-destructive testing and can be used for mapping the complex 3D structures of cracks and their interactions. The materials of most oil and gas transport pipes, buried or submerged, such as high strength low alloy steels, for example, are likely to suffer corrosion degradation and fracture due to the corrosive environment. Similarly, cracks are one of the most severe types of discontinuity in a welded joint since they are strong stress concentrators. High tensile stresses develop in the weld region as a result of the localized thermal expansion and contraction, associated with the welding thermal cycle. To better characterize the cracking behavior, it is important to gain information about the evolution of the 3D crack network. However, the search for improving image quality in the inspection of steel samples using X-ray beams is still challenging because of spreading effects that can cause noise in the 3D image. For this purpose, we performed microCT tests to verify cracks due corrosion and loss of weld adhesion cracks applying mathematical filters to improve the final image quality. To enhance details of the grey scale microCT slices, the image was improved using anisotropic diffusion (AD) and unsharp mask (UM) filters, which have been found to be highly effective for enhancement of digital fractured media. With the results was possible to verify cracks around 0.66 mm for corrosion cracks and 4.27 mm for cracks due to loss of weld adhesion, as well as the cracks network in the 3D visualization of the inspected materials.

    

   

An increasing number of industrial sectors are considering the potential of additive manufacturing as an asset to improve their production. Indeed, additive manufacturing enables the fabrication of very complex geometries and inner cavities that cannot be manufactured with conventional techniques. However, in critical sectors such as aerospace, defence and medical, the parts need to be certified, which requires parts to be non-destructively characterised in terms of flaws, geometry and dimensional accuracy. X-ray computed tomography is the only current 3D volumetric technique, which is suited for the non-destructive analysis of internal flaws, geometry and measurements of internal dimensions and roughness. However, regardless of its huge potential, Xray computed tomography is not as mature a technology for dimensional metrology as compared to conventional tactile coordinate measuring machines. In most cases there is no traceability to SI units in the dimensional domain. Recently, numerous reference standards (i.e. physical artefacts) addressing X-ray computed tomography dimensional accuracy have been published, but they do not necessarily address the calibration of XCT system in connection with AM parts. In this work, a new and improved standard in three different materials has been designed with a dual purpose: Fully calibrating X-ray computed tomography for dimensional measurements while being representative of additively manufactured parts in terms of flaws and material, meeting the needs of the industry. These standards will be used to metrologically validate X-ray computed tomography for the inspection of additively manufactured parts.

    

   

Ammonium nitrate (AN) prills are commonly used as an ingredient in industrial explosives and in fertilisers. Conventional techniques (such as BET or mercury intrusion porosimetry) can measure the open porosity and specific surface area of AN prill, but the closed porosity is not obtainable. This work was focused on evaluating X-ray computed tomography (XCT) as a non-destructive technique for the assessment of porosity in AN prills. An advanced data processing workflow was developed so that the segmentation and quantification of the CT data could be performed on the entire 3D volume, yet allowing the measurements (e.g.; volume, area, shape factor…) to be extracted for each individual phase (prill, open porosity, closed porosity) of each individual prill, in order to obtain statistically relevant data. Clear morphological and structural differences were seen and quantified between fertiliser and explosive products. Overall, CT can provide a very wide range of parameters that are not accessible to other techniques, destructive or non-destructive, and thus offers new insights and complementary information.

    

   

In the field of industrial Computed Tomography (CT), the orientation of the workpiece influences the local quality of the reconstructed volume data and therefore dimensional measurements. In the literature, different a priori criteria for an optimal orientation are proposed, which are purely based on the part and setup geometry. However, it is not evident to which extent they correlate with the measurement accuracy or how different quality parameters should be weighted, especially if a wider range of materials and therefore different artefact contributions should be covered. In this work, existing methods and criteria are compared and extended. In contrast to most existing investigations, different quality parameters are evaluated not only globally, but also restricted to relevant regions of interest (ROI). A straightforward derivation for an analytical determination of surface areas affected by cone-beam artefacts is presented. Results show that an ROI-based evaluation is important to find an optimal orientation for single measurands. The investigated quality parameter based on the prediction of surface areas which are affected by cone-beam artefacts was found to be applicable as exclusion criterion. This only holds true if the evaluation is restricted to relevant ROI surface areas, which should also be considered when evaluating the maximum X-ray penetration length as quality parameter. The correlation of the latter with resulting measurement deviations was found to depend on the measurement task.

    

   

Metal parts with controlled lattice structures can effectively be produced via additive manufacturing (AM) technologies. However, one of the critical aspects of metal AM products is the dimensional and geometrical accuracy. X-ray computed tomography (CT) can be applied to enable advanced control methods that are fundamental for improving the geometrical characteristics and the quality of complex metal AM parts. In this work, Ti6Al4V lattice structures produced by laser powder bed fusion were analysed using a metrological X-ray CT system. Two different approaches for determining the uncertainty of dimensional measurements based on the CT reconstructed volumes were implemented and compared: the "substitution method" and the "multiple measurements" approach. Advantages and limitations of both approaches are identified and discussed.

    

   

In order to perform measurements in industrial computed tomography (CT), the specimen in question has to be mounted on the kinematic system of the CT device. In principle this applies to all measuring machines, however, it is an especially critical step of a CT measurement as there are no companies providing mounting systems in particular for CT applications that have been tried and tested by many years of usage. As a mounting system for CT applications should not interfere with x-ray beams to ensure an uncorrupted measurement, many users rely on special mountings cut from foam, 3D printed mounts or custom- made parts. However, there has not been a systematic survey on which mounting strategy leads to minimal random error in measurement results. Therefore, this paper aims at determining the influence of different mounting strategies on the random measurement error and image quality of the resulting scans. To this end, measurements of a defined test-object using different mountings are performed and compared with respect to these quality parameters. Finally, a recommendation for the optimal choice of a mounting system based on the results is given.

    

   

Computed tomography (CT) is the only dimensional measurement technology that captures holistic geometric information of the complete object. The availability of equally and densely distributed measurement data gathered by CT enables the consideration of local form deviations of the object's surface for the definition of a datum system. A datum system describes a coordinate system that is used for referencing geometrical tolerances. Therefore, the datum definition by an approximation of datum features is replaced by a fitting method called virtual assembly (VA), where the datum surfaces are registered in order to simulate the real, physical workpiece contact. Besides describing the theory, in this paper the method is evaluated using a linear guide assembly as an example that is compared to the real assembly of the object.

    

   

In dimensional X-ray CT (dXCT), the definition of a resolution describing parameter (metrological structural resolution) is not standardized conclusively yet. The instrument transfer function (ITF) is currently being discussed in national and international standardization bodies as one candidate for such a parameter. The ITF describes the wavelength-dependent transfer of a surface modulation - which is the main focus of dXCT - in contrast to the modulation transfer function (MTF), a well-known parameter which describes the grey-value amplitude transfer function after volume reconstruction without including a surface determination. In this article, a practical implementation of two differently sized new ITF reference standards is presented. Their idea is to use a height profile - with nearly constant spectral amplitude and bandwidth-limited - which is extended periodically. This profile is diamond-turned on the outer side of a tube and can be segmented to obtain identical pieces. One piece is calibrated pars pro toto with a high-resolution tactile scanner (HRTS) to obtain the reference ITF. Comparisons, CT parameter dependency tests and simulations highlight the benefit of this new type of standards for the determination of the ITF as a measure for resolution.

    

   

X-ray computed tomography (CT) enables the determination of numerous dimensional measurands with a single scan. However, measurements are often affected by artefacts, which are mainly caused by the effects of the complex physical interactions between the used radiation, the measurement object and the detector as well as the algorithms used for measurement data processing. Surface regions affected by artefacts lead to an inaccurate surface determination and therefore to increased measurement deviations and uncertainties of geometric measurements. This contribution aims to demonstrate the possibility of detecting negatively affected surface regions by a qualitative examination of the underlying volume data in the region of each determined surface point. This classification is used to improve different kinds of sphere measurements, which are evaluated by performing regression analysis onto the measured point clouds. Because of the available qualitative classification for each surface point, it is possible to apply a suitable weighting metric before applying the regression analysis in order to reduce the influence of lowly classified surface areas onto the measurement result. The workflow is demonstrated with a calibrated multi-sphere specimen. The results show that a significant reduction of the measurement errors associated with the evaluation of sphere centre distances, sphere form and radius deviations, respectively, can be achieved, while using the regression analysis tools of the commercial software VGStudio Max (Volume Graphics GmbH).