Industrial Computed Tomography Images Research Articles

Metal Artifact Correction in Industrial CT Images Based on a Dual-Domain Joint Deep Learning Framework

Industrial computed tomography (CT) images reconstructed directly from projection data using the filtered back projection (FBP) method exhibit strong metal artifacts due to factors such as beam hardening, scatter, statistical noise, and deficiencies in the reconstruction algorithms. Traditional correction approaches, confined to either the projection domain or the image domain, fail to fully utilize the rich information embedded in the data. To leverage information from both domains, we propose a joint deep learning framework that integrates UNet and ResNet architectures for the correction of metal artifacts in CT images. Initially, the UNet network is employed to correct the imperfect projection data (sinograms), the output of which serves as the input for the CT image reconstruction unit. Subsequently, the reconstructed CT images are fed into the ResNet, with both networks undergoing a joint training process to optimize image quality. We take the projection data obtained by analytical simulation as the data set. The resulting optimized industrial CT images show a significant reduction in metal artifacts, with the average Peak Signal-to-Noise Ratio (PSNR) reaching 36.13 and the average Structural Similarity Index (SSIM) achieving 0.953. By conducting simultaneous correction in both the projection and image domains, our method effectively harnesses the complementary information from both, exhibiting a marked improvement in correction results over the deep learning-based single-domain corrections. The generalization capability of our proposed method is further verified in ablation experiments and multi-material phantom CT artifact correction.

Manufacturing errors detection and visualization system for parts with complex internal cavity structural through industrial CT image

To address the challenge of detection and visualization manufacturing errors in part with complex internal cavity structures, this paper present a method based on industrial computed tomography (CT) image. First, a two-dimensional (2D) CT sequence slice images are obtained using fan-beam high energy industrial CT scanning, which is then reconstructed into a three-dimensional (3D) measurement model. The measurement 3D model is registered with the original design model, and the surface information of the part is obtained by segmenting the surfaces. The error between reference surface of original design model and test surface of measurement model is calculated. By comparing the error value with the tolerance, the detection result is obtained and analyzed. The developed software system enables areas that do not meet the tolerance requirements to be visualized. This method can detect manufacturing errors for part with complex internal cavity structures, such as planeness error, cylindricity error, verticality error, and parallelism error. This method provides a basis for assessing the manufacturing and processing quality of part with complex internal cavity structures.

NeXtResUNet: A neural network for industrial CT image denoising

Image denoising is a critical issue in industrial computed tomography (CT) inspection. Most existing noise reduction algorithms are based on synthetic data, resulting in the loss of fine details, local region distortions, and other problems that arise during application in practical scenarios. As such, a fusion network, combining ConvNeXt, ResNet, and UNet is proposed in this study to address these issues. This algorithm was applied to a self-constructed industrial CT image denoising dataset, achieving a peak signal-to-noise ratio (PSNR) of 47.20 dB, which is 6% (44.48 dB) higher than that of the benchmark DnCnn. This also represents a 0.5% improvement over the current state-of-the-art (SOTA) SCUNet (46.94 dB). More importantly, the proposed network is superior to other existing models in terms of the overall visual effects and degree of image detail preservation. The network structure is also concise yet simple to expand. As such, it can be applied not only to image denoising, but also to CT image segmentation, super resolution, and related tasks.

A Dilated Residual Network for Turbine Blade ICT Image Artifact Removal

Artifacts are divergent strip artifacts or dark stripe artifacts in Industrial Computed Tomography (ICT) images due to large differences in density among the components of scanned objects, which can significantly distort the actual structure of scanned objects in ICT images. The presence of artifacts can seriously affect the practical application effectiveness of ICT in defect detection and dimensional measurement. In this paper, a series of convolution neural network models are designed and implemented based on preparing the ICT image artifact removal datasets. Our findings indicate that the RF (receptive field) and the spatial resolution of network can significantly impact the effectiveness of artifact removal. Therefore, we propose a dilated residual network for turbine blade ICT image artifact removal (DRAR), which enhances the RF of the network while maintaining spatial resolution with only a slight increase in computational load. Extensive experiments demonstrate that the DRAR achieves exceptional performance in artifact removal.

An Enhanced U-Net Approach for Segmentation of Aeroengine Hollow Turbine Blade

The hollow turbine blade plays an important role in the propulsion of the aeroengine. However, due to its complex hollow structure and nickel-based superalloys material property, only industrial computed tomography (ICT) could realize its nondestructive detection with sufficient intuitiveness. The ICT detection precision mainly depends on the segmentation accuracy of target ICT images. However, because the hollow turbine blade is made of special superalloys and contains many small unique structures such as film cooling holes, exhaust edges, etc., the ICT image quality of the hollow turbine blades is often deficient, with artifacts, low contrast, and inhomogeneity scattered around the blade contour, making it hard for traditional mathematical model-based methods to acquire satisfying segmentation precision. Therefore, this paper presents a deep learning-based approach, i.e., the enhanced U-net with multiscale inputs, dense blocks, focal loss function, and residual path in the skip connection to realize the high-precision segmentation of the hollow turbine blade. The experimental results show that our proposed enhanced U-net can achieve better segmentation accuracy for practical turbine blades than conventional U-net and traditional mathematical model-based methods.

CT image crack segmentation method based on linear feature enhancement.

Industrial computed tomography (CT) crack segmentation is a key technology in industrial CT image processing. Unfortunately, the interference of artifact and noise in CT image often bring great trouble to the crack segmentation. In order to improve the segmentation accuracy of cracks in CT images, we propose to develop and test a new crack segmentation algorithm based on linear feature enhancement by analyzing the features of cracks in CT images. Firstly, the total variational model is used to denoise the input image. Next, a Frangi multiscale filter is used to extract linear structures in the image, and then the extracted linear structures are used to enhance the contrast of the image. Finally, the cracks in the image are detected and segmented by Otsu algorithm. By comparing with the manual segmentation results, the average intersection-over-union (IOU) reaches 86.10% and the average F1 score reaches 92.44%, which verifies the effectiveness and correctness of the algorithm developed in this study. Overall, experiments demonstrate that the new algorithm improves the accuracy of crack segmentation and it is effective applying to industry CT images.

Sub-pixel high precision dimensional measurement method for aero-engine hollow turbine blade based on industrial computed tomography image

Aero-engine hollow turbine blades are work under prolonged high temperature, requiring high dimensional accuracy. Blade profile and wall thickness are important parameters to ensure the comprehensive performance of blades, which need to be measured accurately during manufacturing process. In this study, a high accuracy industrial computed tomography (ICT) measuring method was developed based on standard cylindrical pin and ring workpieces of different sizes. Combining ICT with cubic spline interpolation, a sub-pixel accuracy was achieved in measuring the dimension of component. Compared with the traditional and whole-pixel level image measurement method, the cubic spline interpolation algorithm has the advantages of high accuracy in image edge detection and thus high accuracy of dimensional measurement. Further, the technique was employed to measure the profile and wall thickness of a typical aerospace engine turbine blade, and an accuracy higher than 0.015 mm was obtained.

Slight overcharging cycling failure of commercial lithium-ion battery induced by the jelly roll destruction

Lithium-ion battery is the main power source of electric vehicles and hybrid electric vehicles due to its excellent properties. However, battery suffers overcharging during its use because of inconsistency among batteries, malfunction of charge control and inappropriate battery management. The failure behaviour, mechanism and diagnosis of battery under slight overcharging cycling are studied in this work. The experimental results indicate that battery capacity is approximately zero after failure. The industrial computed tomography images indicate that the jelly roll is destroyed by internal short circuit. This is the reason of battery failure. Impedances of battery increase much. The results of energy dispersive spectrometer indicate the occurrence of Al3+ ions. It means that the aluminium current collector is corroded. Finally, incremental capacity analysis and resistance estimation can be used to diagnose jelly roll destruction and detect failure battery respectively.

A lightweight solution of industrial computed tomography with convolutional neural network

As an advanced non-destructive testing and quality control technique, industrial computed tomography (ICT) has found many applications in smart manufacturing. The existing ICT devices are usually bulky and involve mass data processing and transmission. It results in a low efficiency and cannot keep pace with smart manufacturing. In this paper, with the support from Internet of things (IoT) and convolutional neural network (CNN), we proposed a lightweight solution of ICT devices for smart manufacturing. It consists of efforts from two aspects: distributed hardware allocation and data reduction. At the first aspect, ICT devices are separated into four functional units: data acquisition, cloud storage, computing center and control terminals. They are distributed and interconnected by IoT. Only the data acquisition unit still remains in the production lines. This distribution not only slims the ICT device, but also permits the share of the same functional units. At the second aspect, in the data acquisition unit, sparse sampling strategy is adopted to reduce the raw data and singular value decomposition (SVD) is used to compress these data. They are then transmitted to the cloud storage. At the computing center, an ICT image reconstruction algorithm and a CNN are applied to these compressed sparse sampling data to obtain high quality CT images. The experiments with practical ICT data have been executed to demonstrate the validity of the proposed solution. The results indicate that this solution can achieve a drastic data reduction, a storage space save and an efficiency improvement without significant image degradation. The presented work has been helpful to push the applications of ICT in smart manufacturing.

Toward Comprehensive Industrial Computed Tomography Image Quality Assessment: I. Phantom Design

Abstract When comparing the performance of different industrial X-ray computed tomography (CT) systems, reconstruction algorithms, or scan protocols, it is important to assess how well the required inspection and measurement tasks can be performed. Furthermore, it can be very informative to quantify image quality (IQ) metrics that can provide insight into the IQ characteristics that lead to the resulting inspection or measurement task performance. Inspection and measurement task performance is determined by basic characteristics such as spatial resolution; feature contrast, size, and shape; random noise (noise due to statistical uncertainty in measurements); and image artifacts. In this report, we describe a modular phantom set that enables robustly quantifying these characteristics and also enables assessing the performance of the inspection or measurement tasks themselves. The phantom set includes two phantom bodies and several insert types that can be optionally installed in the bodies. Phantom body extensions can be optionally included to increase scatter. The phantom bodies combined with the available insert types can comprehensively evaluate all important IQ metrics and inspection or measurement tasks. The precisely-known phantom body geometry and insert location, geometry, and orientation supports automatic analysis of large, complex experiments of multiple variables. This phantom set, with the associated image analysis software, could potentially serve as a general evaluation method for non-destructive testing (NDT) CT.

Nondestructive and fined characterization of injection molded ceramic parts by industrial computed tomography

Abstract Nondestructive testing technology of industrial computed tomography(ICT) is introduced to investigate the injection molded bodies (green body, debinded body and, sintered part) in this study. The results show that ICT can characterize the defects qualitatively and quantitatively by comparing and analyzing the gray distribution of ICT images. The size, numbers and location of pores can be characterized precisely and the density distribution of the green body is quantitatively evaluated. The structure difference and evolution related with processing parameters is also analyzed by this technique. In addition, an image processing method is introduced to three-dimensional reconstruct the two dimensional images after threshold segmentation, and observe the internal structure and spatial morphology of the green body.

Defect features recognition in 3D industrial CT images

Due to the limitations of production conditions, there is a certain probability that workpiece product has internal defects, which will have a certain impact on the performance of workpiece. Therefore, the internal defects detection of workpiece is essential. This study proposed a defect recognition method based on industrial computed tomography (CT) image to identify the internal defects of workpiece. The block fractal algorithm was used to locate the defect parts of the image, then the improved k-means clustering algorithm was used to segment the defect parts, and feature vector was extracted by Hu invariant moments. Finally, the firefly algorithm and radial basis function (RBF) neural network were combined to identify the defect. It was found from the experiments that the algorithm in this study had the accuracy of 97.89%, which proved the reliability of the algorithm and provided some suggestions for the defects recognition.

Compressing industrial computed tomography images by means of contour coding

An improved method for compressing industrial computed tomography (CT) images is presented. To have higher resolution and precision, the amount of industrial CT data has become larger and larger. Considering that industrial CT images are approximately piece-wise constant, we develop a compression method based on contour coding. The traditional contour-based method for compressing gray images usually needs two steps. The first is contour extraction and then compression, which is negative for compression efficiency. So we merge the Freeman encoding idea into an improved method for two-dimensional contours extraction (2-D-IMCE) to improve the compression efficiency. By exploiting the continuity and logical linking, preliminary contour codes are directly obtained simultaneously with the contour extraction. By that, the two steps of the traditional contour-based compression method are simplified into only one. Finally, Huffman coding is employed to further losslessly compress preliminary contour codes. Experimental results show that this method can obtain a good compression ratio as well as keeping satisfactory quality of compressed images.

Feature points extraction of different structure for industrial computed tomography image contour

In order to accurately obtain the physical structure and dimension, the feature points extraction must be carried out from the ICT (industrial computed tomography) image contour. Therefore, the discrete data could be transformed into CAD models, and different methods have been developed for the feature points extraction, according to geometric features, the function and shape of mechanical parts. We discussed the feature points extraction methods, and one method is not enough to all the structures, so we must first distinguish the structure as regular and irregular. For the regular structures, by using a positive–negative factor, the curvature estimation method was improved. This method was designed and verified for the actual part, and has an improved accuracy in the feature points extraction. While for the irregular structures we used the polygon approximation method in the feature points extraction. For a complex part composing of several regular structures and irregular structures in the same ICT contour, we divided the whole contour into the regular and irregular structures, then used the corresponding methods for each structure. Last all feature points were extracted. The actual examples showed that the used methods are practical and effective.

工业CT图像边缘伪影校正

工业CT图像边缘伪影校正

面向快速原型的工业CT图像内外轮廓自适应判别方法

面向快速原型的工业CT图像内外轮廓自适应判别方法

A comprehensive method of contour extraction for industrial computed tomography images

A novel contour extraction method is presented in this paper. In reverse engineering based on industrial computed tomography, quality of extracted contours greatly influences the reverse results. As a widely used method of edge extraction, the wavelet transform modulus maximum algorithm is also employed to extract contours. It often gets results through thresholding wavelet transform coefficients and then follows it up by edge tracking. Generally, threshold affects the accuracy of edge localization, and additional edge tracking increases complexity. To overcome these shortcomings, in this paper, we present a method of contour extraction synchronizing with edge tracking but omitting the thresholding process. Firstly, wavelet transform is performed on an original gray image to get modulus matrix. Secondly, by manually clicking inside and outside of the desired contour on the original image, a starting contour point can be obtained through determination of local modulus maximum value. Thirdly, on the mask centered on the starting contour point, a next contour point can be automatically detected by searching local modulus maximum value. Then, starting from that new contour point and recursively seeking the local modulus maximum value, an occlusive contour can be correctly found. Experiments show that this method can directly extract continuous and single-pixel wide contours from gray image.

工业CT图像轮廓提取与优化系统设计

工业CT图像轮廓提取与优化系统设计

基于神经网络边缘提取的工业CT图像与CAD模型的比对检测

A method to analyze the manufacture error of a workpiece based on the comparison inspection between Industrial Computed Tomography(ICT) images and Computer Aided Design(CAD) model was discussed.Firstly,the edged surfaces of ICT images were extracted by the Cellular Neural Network(CNN) with adaptive templates and the data were fused in three directions to obtaine the complete 3D edge surfaces.Then,the Principal Component Analysis(PCA) with the method of minimum bounding box were combined to perform a rough registration,and Singular Value Decomposition and Iterative Closest Point(SVD-ICP) algorithm were used to realize the refined registration for the edged surface data and the CAD model.In experiment,the k-d tree was used to improve the calculation speed of searching for the closest point.The experimental results validate that the comparison inspection method is automatic,visualized and high-accuracy.By the improved comparison inspection method for ICT images and CAD model,the ICT technology can be used to analyze and improve the manufacturing process.

Industrial computed tomography image size measurement

As one of the most useful modern detection technologies, Industrial Computed Tomography (ICT) image size measurement can correctly non-destructively measure the size of workpieces’ inner construction, and it is considered as the standard for quality assurance and reverse engineering. In view of the advantages and disadvantages compared to conventional methods, this paper improves the precision of image size measurement with a new algorithm that uses an approximate function to describe edge degradation. First, this algorithm constructs the approximate function and determines the optimal point of edge detection, based on image intensity and inflexions. Then, in order to accurately extract the image edge, this algorithm is used to revise the primary image, completing construction of the CT image. Excellent results are obtained from simulations and experiments. The experimental results indicate that the relative error is 2% for the CT image when the step evolution of the image edge is pooled. The relative error of this method is decreased by as much as 1.5% compared to wavelet transformation and ridgelet transformation. Therefore, this new algorithm demonstrates increased effectiveness in extracting an accurate measurement of the CT image edge.