Image Defects Research Articles

Restoration method of ancient mural image defect information based on neighborhood filtering

The local data of ancient murals is seriously damaged, and image noise exists in the process of restoration, which affects the quality of restoration of ancient murals. Therefore, this paper studies the restoration method of ancient mural image defect information based on neighborhood filtering. On the premise of obtaining the causes of ancient mural defects, this method enhances image data based on spatial domain enhancement method, extracts pixel similar information based on neighborhood filtering, searches in the whole image, and removes image noise used to repair local areas; By extracting the line drawing features of mural, the defect part of ancient mural image can be repaired. The experimental results show that the peak signal-to-noise ratio of the repaired image is the highest and the quality of the image is better under the application of the repair method.

Intelligent restoration of ancient murals based on discrete differential algorithm

Many defects now appear in ancient murals due to both natural and man-made factors. To better repair the damaged ancient murals, this paper improves the existing intelligent restoration method. It identifies and marks the crack-producing area, categorizes the mural image into two types – texture area and flat area according to the local gradient features, decides the initial sample block and calculates the weight, analyzes the extracted pixel data and applies discrete differential algorithm to supplement image defects. Through experiments, the method is validated in meeting the needs of image continuity law and human vision. It can shorten the repair time and restore mural cracks in an intelligent way.

Research on Detection Method for Welding Seam Defects in Ultrasonic TOFD Image Based on Mask R-CNN

In this paper, the technology of ultrasonic TOFD was used in the welding seam inspection of construction steel structure. 162 ultrasonic TOFD scanning images including five kinds of common defects were obtained through field test. The five kinds of common welding defects include stoma, slag inclusion, lack of fusion, incomplete penetration and crackle. Combined with artificial intelligence image recognition technology, an automatic detection algorithm of welding seam defects in ultrasonic TOFD scanning images based on Mask R-CNN was developed. By comparing the effects of automatic location and classification of welding seam defects based on Mask R-CNN after the same number of iterative training under different parameter configurations, the results show that Mask R-CNN can automatically locate and classify welding seam defects and the recognition effect is the best when epoch is 667, steps_per_epoch is 15 and learning rate is 0.001. Under this parameter configuration, the recognition and classification of five kinds of common defects such as stoma, slag inclusion, lack of fusion, incomplete penetration and crackle are correct. The correct recognition confidence of all five kinds of defects can reach more than 0.9, and the location is precise.

Automatic detection of crystallographic defects in STEM images by unsupervised learning with translational invariance

Automatic detection of crystallographic defects in STEM images by unsupervised learning with translational invariance

Optimal Algorithm of Full Focus Array Based on Sparse Matrix

Ultrasonic phased array all-focus imaging algorithm has the advantages of high accuracy and full range dynamic focusing, but the existing imaging time-consuming problem limits its practical industrial application. In order to meet the requirements of real-time imaging and clear and distinguishable image defects in all-focus array ultrasound scanning. In this paper, a sparse optimization model of TFM phased array is established. Based on the Fermat principle, the delay time of each element under two layers of media is calculated, and a two-layer media phased array all-focus imaging algorithm is established, using minimal redundancy. The sparse design of the MRLA is carried out, and the all-focus imaging and sparse all-focus imaging experiments of the two-layer medium are carried out, and the influence of the sparse emission array on the defect quantitative accuracy and the calculation efficiency of the all-focus algorithm is discussed. The final optimized sparse array is compared with the sound field beam pattern of the full array, the smallest redundant array, and the genetic algorithm optimized array, and an all-focus scanning environment is established based on the array. Scanning images can not only obtain the maximum amount of information under minimum redundancy, but also have good sidelobe characteristics. At the same time, the array is sparse, which greatly improves the efficiency of all-focus scanning, and provides a certain reference value for the research of all-focus imaging technology.

Deep Learning for the Detection and Recognition of Rail Defects in Ultrasound B-Scan Images

Rail defect detection is crucial to rail operations safety. Addressing the problem of high false alarm rates and missed detection rates in rail defect detection, this paper proposes a deep learning method using B-scan image recognition of rail defects with an improved YOLO (you only look once) V3 algorithm. Specifically, the developed model can automatically position a box in B-scan images and recognize EFBWs (electric flash butt welds), normal bolt holes, BHBs (bolt hole breaks), and SSCs (shells, spalling, or corrugation). First, the network structure of the YOLO V3 model is modified to enlarge the receptive field of the model, thus improving the detection accuracy of the model for small-scale objects. Second, B-scan image data are analyzed and standardized. Third, the initial training parameters of the improved YOLO V3 model are adjusted. Finally, the experiments are performed on 453 B-scan images as the test data set. Results show that the B-scan image recognition model based on the improved YOLO V3 algorithm reached high performance in its precision. Additionally, the detection accuracy and efficiency are improved compared with the original model and the final mean average precision can reach 87.41%.

Generative Adversarial Networks to Improve the Robustness of Visual Defect Segmentation by Semantic Networks in Manufacturing Components

This paper describes the application of Semantic Networks for the detection of defects in images of metallic manufactured components in a situation where the number of available samples of defects is small, which is rather common in real practical environments. In order to overcome this shortage of data, the common approach is to use conventional data augmentation techniques. We resort to Generative Adversarial Networks (GANs) that have shown the capability to generate highly convincing samples of a specific class as a result of a game between a discriminator and a generator module. Here, we apply the GANs to generate samples of images of metallic manufactured components with specific defects, in order to improve training of Semantic Networks (specifically DeepLabV3+ and Pyramid Attention Network (PAN) networks) carrying out the defect detection and segmentation. Our process carries out the generation of defect images using the StyleGAN2 with the DiffAugment method, followed by a conventional data augmentation over the entire enriched dataset, achieving a large balanced dataset that allows robust training of the Semantic Network. We demonstrate the approach on a private dataset generated for an industrial client, where images are captured by an ad-hoc photometric-stereo image acquisition system, and a public dataset, the Northeastern University surface defect database (NEU). The proposed approach achieves an improvement of 7% and 6% in an intersection over union (IoU) measure of detection performance on each dataset over the conventional data augmentation.

Dysmorphic and illness anxiety-related unwanted intrusive thoughts in individuals with obsessive-compulsive disorder.

Unwanted intrusive thoughts (UITs) are considered normal variants of the obsessions found in obsessive-compulsive disorder (OCD). Similarly, intrusive and persistent preoccupations about appearance defects in body dysmorphic disorder (BDD) and images and thoughts about illness in illness anxiety disorder (IAD) are abnormal variants of the thoughts and concerns about appearance and health found in non-clinical individuals. This study examines whether patients with OCD have frequent and distressing UITs with contents related to BDD and IAD, in addition to OCD-related UITs. Thirty-nine participants with OCD (Mage = 32.45, standard deviation [SD] = 11.57; 63% men) completed the Questionnaire of Unpleasant Intrusive Thoughts (QUIT), designed to assess obsessional, dysmorphic and illness anxiety UITs and their associated functional consequences (emotions, appraisals, and neutralizing or control strategies). Up to 71% of the individuals with OCD experienced the three UIT modalities. As expected, OCD-related UITs were the most frequent, although were no differences among the three UIT contents in terms of the disturbance they caused. The OCD intrusions were the most interfering, egodystonic and dysfunctionally appraised, and they instigated more neutralizing behaviours. Nonetheless, all UITs instigate the need to do something to alleviate the discomfort caused by their intrusion and attempts to suppress them. UITs with different contents not always related to typical obsessive themes are a common experience in patients with OCD. These UITs could have detrimental consequences for the course of the disorder itself, and they should be adequately addressed in both the assessment and treatment of these patients.

Using Machine Learning Techniques to Detect Defects in Images of Metal Structures

Using Machine Learning Techniques to Detect Defects in Images of Metal Structures

Restoration and Enhancement of Historical Stereo Photos.

Restoration of digital visual media acquired from repositories of historical photographic and cinematographic material is of key importance for the preservation, study and transmission of the legacy of past cultures to the coming generations. In this paper, a fully automatic approach to the digital restoration of historical stereo photographs is proposed, referred to as Stacked Median Restoration plus (SMR+). The approach exploits the content redundancy in stereo pairs for detecting and fixing scratches, dust, dirt spots and many other defects in the original images, as well as improving contrast and illumination. This is done by estimating the optical flow between the images, and using it to register one view onto the other both geometrically and photometrically. Restoration is then accomplished in three steps: (1) image fusion according to the stacked median operator, (2) low-resolution detail enhancement by guided supersampling, and (3) iterative visual consistency checking and refinement. Each step implements an original algorithm specifically designed for this work. The restored image is fully consistent with the original content, thus improving over the methods based on image hallucination. Comparative results on three different datasets of historical stereograms show the effectiveness of the proposed approach, and its superiority over single-image denoising and super-resolution methods. Results also show that the performance of the state-of-the-art single-image deep restoration network Bringing Old Photo Back to Life (BOPBtL) can be strongly improved when the input image is pre-processed by SMR+.

Selection and evaluation of spherical acquisition trajectories for industrial computed tomography

In conventional industrial computed tomography, the source–detector system rotates in equiangular steps in-plane relative to the part of investigation. While being by far the most frequently used acquisition trajectory today, this method has several drawbacks like the formation of cone beam artefacts or limited usability in case of geometrical restrictions. In such cases, the usage of alternative spherical trajectories can be beneficial to improve image quality and defect visibility. While investigations have been performed to relate the influence of the trajectory choice in the typical metrological case of a high number of available projections, so far barely any work has been done for the case of few source–detector poses, which is more relevant in the field of non-destructive testing. In this work, we provide an overview of quantitative metrics that can be used to assess the image quality of reconstructed computed tomography volumes, discuss their advantages and drawbacks and propose a framework to investigate the performance of several non-standard trajectories with respect to previously defined regions of interest. Inspired by pseudorandom sampling methods for Monte–Carlo-algorithms, we also suggest an entirely new trajectory design, the low-discrepancy spherical trajectory, which extends the concept of equiangular planar trajectories into three dimensions and can be used for benchmarking and comparison with other spherical trajectories. Last, we use an optimization method to calculate task-specific acquisition trajectories and relate their performance to other spherical designs.

Research on Defect Classification and Detection Technology of Image Processing Collected by Computer

The paper uses median filter computer algorithms to detect and classify machine component image defects using image processing, pattern recognition and machine vision theories. In the experiment, automatic extraction of defect images and minimization of defect images are completed to reduce processing volume and storage space requirements, and automatically determine defect categories. The article processes the collected defect images, and the experimental results prove that the method can correctly realize the detection of track surface defects and has certain applicability.

Defect Detection in Atomic Resolution Transmission Electron Microscopy Images Using Machine Learning

Point defects play a fundamental role in the discovery of new materials due to their strong influence on material properties and behavior. At present, imaging techniques based on transmission electron microscopy (TEM) are widely employed for characterizing point defects in materials. However, current methods for defect detection predominantly involve visual inspection of TEM images, which is laborious and poses difficulties in materials where defect related contrast is weak or ambiguous. Recent efforts to develop machine learning methods for the detection of point defects in TEM images have focused on supervised methods that require labeled training data that is generated via simulation. Motivated by a desire for machine learning methods that can be trained on experimental data, we propose two self-supervised machine learning algorithms that are trained solely on images that are defect-free. Our proposed methods use principal components analysis (PCA) and convolutional neural networks (CNN) to analyze a TEM image and predict the location of a defect. Using simulated TEM images, we show that PCA can be used to accurately locate point defects in the case where there is no imaging noise. In the case where there is imaging noise, we show that incorporating a CNN dramatically improves model performance. Our models rely on a novel approach that uses the residual between a TEM image and its PCA reconstruction.

A Simple Subjective Evaluation of Enface OCT Reflectance Images Distinguishes Glaucoma From Healthy Eyes.

PurposeWe present a subjective approach to detecting glaucomatous defects in enface images and assess its diagnostic performance. We also test the hypothesis that if reflectivity changes precede thickness changes in glaucoma there should be reduced correlation between the modalities in glaucoma compared to controls.MethodsTwenty glaucoma participants and 20 age-matched controls underwent high-resolution OCT scans of one eye. 4 µm-thick enface slabs were constructed through the retina. Enface indices were depths of first gap in visible retinal nerve fiber bundles (RNFBs) and last visible bundle, subjectively evaluated in six sectors of a 3.5 mm circle around the optic disc. Retinal nerve fiber layer thickness (RNFLT) along the same circle was extracted at angles corresponding to enface indices. Between-group differences were tested by linear mixed models. Diagnostic performance was measured by partial receiver operating characteristic area (pAUC).Results First gap and last visible bundle were closer to the inner limiting membrane in glaucoma eyes (both P < 0.0001). Enface indices showed excellent diagnostic performance (pAUCs 0.63–1.00), similar to RNFLT (pAUCs 0.63–0.95). Correlation between enface and RNFLT parameters was strong in healthy (r = 0.81–0.92) and glaucoma eyes (r = 0.73–0.80).ConclusionsThis simple subjective method reliably identifies glaucomatous defects in enface images with diagnostic performance at least as good as existing thickness indices. Thickness and reflectivity were similarly related in healthy and glaucoma eyes, providing no strong evidence of reflectivity loss preceding thinning. Objective analyses may realize further potential of enface OCT images in glaucoma.Translational RelevanceNovel enface OCT indices may aid glaucoma diagnosis.

Attention U-Net with Feature Fusion Module for Robust Defect Detection

U-Net shows a remarkable performance and makes significant progress for segmentation task in medical images. Despite the outstanding achievements, the common case of defect detection in industrial scenes is still a challenging task, due to the noisy background, unpredictable environment, varying shapes and sizes of the defects. Traditional U-Net may not be suitable for low-quality images with low illumination and corruption, which are often presented in the practical collections in real-world scenes. In this paper, we propose an attention U-Net with feature fusion module for combining multi-scale features to detect the defects in noisy images automatically. Feature fusion module contains convolution kernels of different scales to capture shallow layer features and combine them with the high-dimensional features. Meanwhile, attention gates are used to enhance the robustness of skip connection between the feature maps. The proposed method is evaluated on two datasets. The best precision rate and MIoU of defect detection are 95.6% and 92.5%. The best F-score of concrete crack detection is 95.0%. Experimental results show that the proposed approach achieves promising results in both datasets. It demonstrates that our approach consistently outperforms other U-Net-based approaches for defect detection in low-quality images. Experimental results have shown the possibility of developing a mixture system that can be deployed in many applications, such as remote sensing image analysis, earthquake disaster situation assessment, and so on.

Floquet wave theory-based time-corrected ultrasonic total focusing method for fiber-reinforced composite laminate

Fiber reinforced polymers (FRPs) are increasingly used in thick primary load-bearing structures. Nevertheless, manufacturing and in-service defects occur with a higher chance as the FRP thickness increases and thus the potential structure defects should be detected and evaluated. To image defects in thick FRP over 10 mm thick, this study proposes a frequency-and-structure dependent time-corrected ultrasonic total focusing method (FS-TFM) based on Floquet wave theory, which differs from conventional TFM that the wave velocity correction along different propagation direction bases on not only the structural anisotropy and inhomogeneity of FRP, but also on the probing frequency. First an analytical Floquet-wave-based dynamic homogenization for a crossply FRP laminate is performed, to obtain the homogenization region and then the wave anisotropy and dispersion therein. Then numerical finite element analysis follows, to further interrogate the time domain feature of wave signals. With the understanding of wave propagation, by accurately correcting the wave velocity anisotropy related to both the inspected crossply FPR and wave frequency, the FS-TFM imaging technique is proposed to focus the defect-scattered wave energy on the defect location. In addition, the signal-to-noise ratio (SNR) analysis of side-drilled-hole (Ø2 mm) imaging via different imaging algorithms shows that the proposed FS-TFM achieves the highest SNR at the frequency approaching the upper limit of the homogenization region. Finally, the experimental validation further indicates the potential of the proposed FS-TFM for accurate defect imaging in thick FRP.

Comparing optimization methods for deep learning in image processing applications

Abstract Training a deep learning network requires choosing its weights such that the output minimizes a given loss function. In practice, stochastic gradient descent is frequently used for solving the optimization problem. Several variants of this approach have been suggested in the literature. We study the impact of the choice of the optimization method on the outcome of the learning process at the example of two image processing applications from quite different fields. The first one is artistic style transfer, where the content of one image is combined with the style of another one. The second application is a real world classification task from industry, namely detecting defects in images of air filters. In both cases, clear differences between the results of the individual optimization methods are observed.

An ameliorated deep dense convolutional neural network for accurate recognition of casting defects in X-ray images

Recognizing defects in X-ray images plays an important role in the detection of internal defects in titanium alloy castings. However, the existing manual defects recognition methods have common drawbacks such as unstable artificial recognition, misrecognition, huge workload, and low efficiency of recognition. To make up for the shortcomings, an ameliorated deep dense convolutional neural network (BX-Net) was presented to accurately recognize casting defects in X-ray images and effectively extract highly discriminative features of different categories. DenseNet121 was used as the backbone of BX-Net and the feature extracted by DenseNet121 was fully shared by the two inputs of a bilinear pooling layer. Transfer learning was applied to reduce the demand for data and hyperparameters’ tuning. The backbone of BX-Net was firstly trained on the ImageNet and then all layers of BX-Net was fine-tuned on nine-hundred X-ray images of TiAl aero casting components. Other six deep convolutional neural networks (DenseNet121, EfficientNetB4, EfficientNetB7, ResNet50, VGG16 and Xception) were also trained to be compared with the presented BX-Net. Two Support Vector Machines were trained on the LBP features data set and HOG features data set of nine-hundred X-ray images of TiAl aero casting components respectively Experiments comparing BX-Net with other six deep learning models and two machine learning models on one hundred X-ray images (test set) of TiAl aero casting components were carried out. The comparison results show that BX-Net has the least parameters except for the Densenet121. The recall and accuracy of BX-Net were 99% and 99% respectively. In addition, the comparison results also show that BX-Net was the only model that learned discriminative feature representation of the casting X-ray image data set. The BX-Net proposed in this paper is expected to overcome the shortcomings of manual defects recognition.

Novel Feature Fusion Module-Based Detector for Small Insulator Defect Detection

The failure of an insulator may compromise the safety of the entire power transmission system. Therefore, insulator defect detection is vital for the safe operation of power systems. However, insulator defects in an insulator image may have varying sizes, and several currently available methods do not have satisfactory detection accuracy for small defects. To address this issue, we propose an improved detection network for small insulator defects with a batch normalization convolutional block attention module (BN-CBAM) and a feature fusion module. The BN-CBAM is designed to better exploit channel information and enhance the effect of different channels on the feature map. In addition, we propose a feature fusion module that fuses multi-scale features from different layers to improve small object detection performance. Moreover, to address the scarcity of aerial images, a data augmentation method based on the fusion of the target segment and background is introduced. Experiments demonstrate that the proposed method achieves better small insulator defect detection performance than other state-of-the-art approaches. In addition, data augmentation methods enrich sample diversity and enhance the generalizability of the network.

An effective defect detection method based on improved Generative Adversarial Networks (iGAN) for machined surfaces

Abstract In actual industrial applications, the predictive performance of the deep learning model mainly depends on the size and quality of the training samples, while the collection period of defective samples is long or even difficult to obtain. In this article, an effective method based on improved Generative Adversarial Networks (iGAN) is proposed to detect defects of machined surfaces on the basis of positive sample training and image restoration. This model could help to detect surface defects through positive sample learning without the training for defect samples and traditional artificial labels. In this method, an improved reconstructed image network model is constructed based on the Generative Adversarial Networks. Through this method, a defective image could be repaired by determining the threshold of the residual image through the Otsu algorithm, and the difference between the input image and the repaired image will be compared to obtain the defect area. Finally, the experiment for the model validation is conducted on the complex surface image of the engine cylinder head. The result shows that the improved Generative Adversarial Networks is effective in both the image restoration and defect identification process.