Surface Crack Detection Research Articles

Performance analysis of various types of surface crack detection based on image processing

Major cracks on a highway or bridge's concrete surface have a massive risk of damages, accompanied by less maintenance, slow detection, and handling; the worst case of the damage is the structure's total collapse, which can produce fatalities. Moreover, Indonesia's climate and geographical location contribute to a higher level of potential damage to the structure. In order to reduce the potential damage, the need for a surface crack detection system arises. This research analysed three different databases (Database A, B, and C) with different surface concrete crack types, such as early thermal contraction, plastic shrinkage, corrosion reinforcement, and non-crack images. The total images from each Database vary from 14 images for Database A, 80 images for Database B, and 4000 images for Database C. The Otsu thresholding and mathematical morphology operations such as opening, closing, dilation, and erosion with pre-processing methods were combined and produced results for each Database with classification using Euclidean distance calculation. The best results for Database A and B were 100% using combination Otsu thresholding with Laplacian operator and Laplacian of Gaussian filter and the same result for a combination of mathematical morphological operations. The best result using Database C, which had more images than Database A and B, was 80,2% using a combination of mathematical morphological operations.

Image Processing with Deep Learning: Surface Defect Detection of Metal Gears through Deep Learning

Intelligent production requires improved data analytics and better technological possibilities to improve system performance and decision making. With the widespread use of the machine learning process, a growing need has arisen for processing extensive production data, equipped with high volumes, high speed, and high diversity. At this point, deep learning provides advanced analysis tools for processing and analyzing extensive production data. The deep convolutional neural network (DCNN) displays state-of-the-art performance on many grounds, including metal manufacturing surface defect detection. However, there is still space for improving the defect detection performance over generic DCNN models. The proposed approach performed better than the associated methods in the particular area of surface crack detection. The defect zones of disjointed results are classified into their unique classes by a DCNN. The experimental outcomes prove that this method meets the durability and efficiency requirements for metallic object defect detection. In time, it can also be extended to other detection methods. At the same time, the study will increase the accuracy quality of the features that can make a difference in the deep learning method for the detection of surface defects.

Detection of Metal Surface Cracks Based on Liquid Switch Controlled Spoof Surface Plasmon Polaritons

A spoof surface plasmon polaritons (SSPPs) sensor based on liquid switch control is proposed to detect metal surface cracks. The sensor consists of liquid switch controllers and power division structure. The liquid switch consists of a plastic pipe and syringe with water, which is fixed on the branch structure of SSPPs sensor. By filling the plastic pipe with water, the branch cannot transmit signal, which is called “ OFF” state. The branch retransmits signal by pumping water out of plastic pipe, which is called “ ON” state. By switching the “ ON” or “ OFF” state of sensor’s branches, it can control the transmission path of SSPPs sensor, that is, control the sensing area of SSPPs sensor. The detection method is based on the change of reflection coefficient of SSPPs sensor caused by cracks. By switching the states of branches in turn, it can detect whether there is a crack at each branch. The proposed sensor uses liquid switch instead of mechanical movement, which can realize rapid detection in the metallic materials with large surface area. The simulation and experimental results show that the SSPPs sensor can effectively detect metal surface crack with a width of 0.1mm.

A multi‐scale learning method with dilated convolutional network for concrete surface cracks detection

Concrete surface cracks detection is an important task to ensure the safety of infrastructure. Because of the complexity of background and low contrast of concrete surface, it is difficult to detect the cracks on the concrete surface accurately. To tackle this problem, a multi-scale dilated convolutional method for concrete surface crack detection is proposed to improve the accuracy of detection. The proposed network is based on the encoder-decoder structure of U-Net. Cascade multi-scale dilated convolutions in the centre of the network is used to get larger receptive field without additional parameters. In the decoder stage, the feature fusion module is used to integrated the multi-scale and multi-level side network feature for the final prediction. A large crack dataset is collected as a training set and other three smaller datasets are used for evaluation. Extensive experiments have been conducted on these three crack datasets, which achieves optimal dataset scale (ODS) F-score over 0.84, optimal image scale (OIS) F-score over 0.85 and average precision (AP) over 0.86. This algorithm performs better than the current crack detection, edge detection and semantic segmentation methods.

Effect of speckle size on surface crack detection via digital image correlation

Digital image correlation (DIC) is an experimental stress analysis technique used in nondestructive tests. The accuracy of DIC in crack detection depends on various factors such as the sizes of speckles and pixels. In the current study, a speckle pattern based on the spreading of nanoparticles with small speckles is compared with a conventional sprayed pattern to understand whether crack detection via DIC is improved by reducing the sizes of speckles and pixels. Owing to the small size of nanoparticles, an optical microscope is used for magnification. The spreading method for crack detection is first investigated experimentally. Results show that cracks can be detected easily when a 250 nm opening appears in the crack edges. Subsequently, the spreading method is compared with the conventional DIC, in which the spraying method is used for patterning, in terms of crack detection. Results show that by reducing the speckle size and closely analyzing the speckle pattern, the DIC technique is considerably better than the conventional technique in detecting small cracks. Moreover, the conventional method is more suitable for detecting large cracks.

Infrared Thermal Imaging Detection of Surface Cracks in External Insulation Layer of Building Exterior Wall

Infrared Thermal Imaging Detection of Surface Cracks in External Insulation Layer of Building Exterior Wall

Earthquake Crack Detection From Aerial Images Using a Deformable Convolutional Neural Network

Detecting the terrain surface cracks caused by earthquakes, which are termed coseismic ruptures, has important significance for discovering concealed faults, monitoring their movements, and forecasting possible follow-on earthquakes. On May 22, 2021, Maduo County in Qinghai province, China, suffered an earthquake with a magnitude of 7.4, which created densely distributed cracks. In this study, we designed an automatic crack detection framework based on remote sensing technology. With the use of low-altitude unmanned aerial vehicles (UAVs), we obtained very high-resolution aerial images of the area affected by the earthquake, which were further processed by photogrammetric software to produce digital orthophoto maps (DOMs). We then designed a novel terrain surface crack detection neural network, which differs from the previous methods that focus on detecting cracks in man-made object surfaces such as flat roads. We investigated the spatial property of the sinuous linear cracks and handled this by introducing adaptive deformable convolutions with a context-channel-space boosted mechanism. The feature extraction stage, feature optimization stage, and upsampling stage were embedded with the deformable convolutions to form a compact and powerful crack detector, named Crack-CADNet (the Context-chAnnel-space boosted Deformable convolutional neural network for crack detection). The postprocessing included filtering out the nontectonic cracks, aided by annotations from experts, and grouping and vectorizing the generated binary segmentation map as crack polygons, which were evaluated at the instance level. In addition to the first in-depth investigation of detecting earthquake cracks with aerial remote sensing and a deep learning based process, the crack detection network we propose outperformed the recent convolutional neural network (CNN)-based methods designed for general semantic segmentation and crack detection. Source code and the Maduo earthquake crack dataset will be available at http://gpcv.whu.edu.cn/data/.

Tunnel Crack Detection With Linear Seam Based on Mixed Attention and Multiscale Feature Fusion

Crack detection techniques have been rapidly developed in recent years due to the rise of deep learning. However, existing methods struggle to produce accurate crack segmentation results because cracks and linear seams on the tunnel lining surface have significant similarities in terms of intensity value and texture features. At the same time, due to the scarcity of data, the existing tunnel lining surface crack detection methods still use multi-step traditional image processing methods for detection, which is inefficient. In this paper, we collect and label a dataset of 200 tunnel lining surface crack images named Tunnel200. For the first time, a deep learning-based method is used to detect cracks in the tunnel lining surface. To deal with the characteristics of crack and linear seam, which mostly present long strip or curved shapes, we propose a Mixed Attention (MA) module by efficient embedding channel and positional information. Unlike common spatial attention that aggregates information throughout space, mixed attention aggregates feature directly along with two directions, height, and width, in the spatial dimension. In this way, the long-range dependence of the crack features can be effectively captured. The proposed MA is simple to incorporate into the network. Meanwhile, we embed it in the traditional U-shape network while employing an efficient multi-scale feature fusion technique to build the Tunnel Crack Detection Network (TCDNet). TCDNet outperforms other crack detection and semantic segmentation methods on the Tunnel200 dataset. Additionally, we evaluate our method on two publicly available crack datasets, Crack500 and DeepCrack, and our method gets superior performance.

Lightweight Neural Network for Real-Time Crack Detection on Concrete Surface in Fog

Cracks are one of the most common factors that affect the quality of concrete surfaces, so it is necessary to detect concrete surface cracks. However, the current method of manual crack detection is labor-intensive and time-consuming. This study implements a novel lightweight neural network based on the YOLOv4 algorithm to detect cracks on a concrete surface in fog. Using the computer vision algorithm and the GhostNet Module concept for reference, the backbone network architecture of YOLOv4 is improved. The feature redundancy between networks is reduced and the entire network is compressed. The multi-scale fusion method is adopted to effectively detect cracks on concrete surfaces. In addition, the detection of concrete surface cracks is seriously affected by the frequent occurrence of fog. In view of a series of degradation phenomena in image acquisition in fog and the low accuracy of crack detection, the network model is integrated with the dark channel prior concept and the Inception module. The image crack features are extracted at multiple scales, and BReLU bilateral constraints are adopted to maintain local linearity. The improved model for crack detection in fog achieved an mAP of 96.50% with 132 M and 2.24 GMacs. The experimental results show that the detection performance of the proposed model has been improved in both subjective vision and objective evaluation metrics. This performs better in terms of detecting concrete surface cracks in fog.

Road Surface Crack Detection Method Based on Conditional Generative Adversarial Networks.

Constant monitoring of road surfaces helps to show the urgency of deterioration or problems in the road construction and to improve the safety level of the road surface. Conditional generative adversarial networks (cGAN) are a powerful tool to generate or transform the images used for crack detection. The advantage of this method is the highly accurate results in vector-based images, which are convenient for mathematical analysis of the detected cracks at a later time. However, images taken under established parameters are different from images in real-world contexts. Another potential problem of cGAN is that it is difficult to detect the shape of an object when the resulting accuracy is low, which can seriously affect any further mathematical analysis of the detected crack. To tackle this issue, this paper proposes a method called improved cGAN with attention gate (ICGA) for roadway surface crack detection. To obtain a more accurate shape of the detected target object, ICGA establishes a multi-level model with independent stages. In the first stage, everything except the road is treated as noise and removed from the image. These images are stored in a new dataset. In the second stage, ICGA determines the cracks. Therefore, ICGA focuses on the redistribution of cracks, not the auxiliary elements in the image. ICGA adds two attention gates to a U-net architecture and improves the segmentation capacities of the generator in pix2pix. Extensive experimental results on dashboard camera images of the Unsupervised Llamas dataset show that our method has better performance than other state-of-the-art methods.

A surface cracks detection method for nuclear fuel pellets using an improved fully convolutional network

ABSTRACT Surface cracks, one of the main quality defects in nuclear fuel pellets, are usually subtle and hidden in complex backgrounds, making them difficult to detect accurately and threatening reactor safety. In order to achieve accurate cracks detection, a fuel pellet crack detection (FPCD) network is proposed based on the fully convolutional network with the VGG16 network structure as the encoder’s backbone (FCN-VGG16). The new architecture modifies the combination mode of convolution layers and pooling layers, and optimizes the size and number of convolution kernels. And a large weight is added for crack samples in the loss function to solve the class imbalance problem. Subsequently, a self-acquired dataset containing 2240 pellet surface crack images is used for training and performance evaluation of networks. Experiments show that the proposed method achieved an F1-score of 88.89% and an intersection over union (IoU) of 80.30%, obtaining better detection results for complex background crack images.

A new image processing strategy for surface crack identification in building structures under non‐uniform illumination

Crack detection in building structures is an important approach to evaluating the safety of these structures. However, in some cases under special circumstances, it is very difficult to identify all features of cracks by visual inspection. To tackle this problem and offer a higher potential for practical implementations, developing an automatic crack detection method seems essential. This paper establishes a novel algorithm for surface crack detection in building walls based on image processing techniques to enhance crack visibility. One of the most prevalent issues in surface crack detection from digital crack images is non-uniform illumination of background which makes narrow and tiny cracks indistinguishable. In this regard, an integrated approach, combining adaptive image threshold using local first-order statistics, contrast-limited adaptive histogram equalization, and noise filtering using non-linear diffusion filtering, is proposed to extract the whole skeleton of the crack from digital crack images with non-uniform illumination. Several realistic crack images, suffering from noise and uneven lighting of background, are processed through the proposed algorithm to demonstrate the robustness and high accuracy of the method.

Signal Enhancement in Surface Crack Detection with Gas-Coupled Laser Acoustic Detection

Ultrasonic signal enhancement resulting from constructive interference between direct Rayleigh waves and same waves reflected by a surface defect is exploited to increase crack identification capabilities of the Gas-Coupled Laser Acoustic Detection (GCLAD) non-contact detection technology. Highlights from simulations are provided regarding the interference phenomenon in the solid and its propagation in air, where GCLAD detection occurs. Experimental campaigns are preliminarily performed on a bar to evidence the effect of cracks on the GCLAD acquired signals. Then, a signal enhancement of +30% is reached on a plate, implying that defects are efficiently scanned by moving the GCLAD in proximity of the discontinuity. Since the GCLAD allows monitoring points of a piece belonging to the same line at once, its translation in one direction is sufficient to perform a two-dimensional scan, entailing reduction of inspection time and simple automation of the interrogation layout compared to other traditional or signal enhancement-based techniques.

Early damage detection of fatigue failure for RC deck slabs under wheel load moving test using image analysis with artificial intelligence

Reinforced concrete (RC) bridge decks suffer from cracking damages caused by traffic loading and environmental-related defects, such as the alkali-silica reaction (ASR). These require inspections involving measuring crack width and density followed by essential maintenance and repair works, however, there are no signs for fatigue failure. In this study, the out-of-plane shear deformations which cause small delaminations (pits) along surface cracks are proposed as an early indicator for fatigue failure. Thus, un-damaged and ASR-damaged RC deck slabs are tested under moving wheel loading and, using image-recognition for surface cracks detection, the pits along surface cracks are captured using an artificial intelligence (AI) model. The results show that, while both crack and pit density increase over the fatigue life of un-damaged slabs, there is an earlier sudden increase in pit density. In the case of the ASR-damaged slab, surface cracking is almost constant over the fatigue life until a sudden increase just prior to failure. Pit density, however, increases over the fatigue life with an earlier rapid increase before failure. The density of pits along cracks would be, therefore, a significantly earlier indicator of fatigue failure than crack density, offering the potential for more efficient and automatable bridge inspections.

A theoretical approach towards the modeling of vibrothermography using finite element methods

Vibrothermography offers the capability to detect surface cracks mainly through capturing vibration-induced frictional heat generation of contacting crack surfaces in metallic alloys. The multidisciplinary nature of this inspection technology makes it rather difficult to theoretically capture the system response without the proper application of finite element methods (FEM). It requires relating the material mechanical and thermal properties in a coupled finite element model to properly address wave propagation, contact mechanics, fracture mechanics, friction heat generation and heat diffusion in simulated vibrothermography inspections. In this study, a theoretical model is developed to estimate the dynamic strain response at any location in commonly used cantilever beams oscillating at a given frequency. This model is utilized to assess the convergence of FEM dynamic system response while determining the element size requirements to realistically model elastic wave propagation throughout the sample. Three different meshing criteria are assessed to properly capture the singularity at the crack tip. It is demonstrated that the use of solid element circular meshing criteria around the crack tip converges to singular quarter-node solid element solution commonly used in fracture mechanics and allows coupling it with heat generation and diffusion around the vicinity of a crack to deliver a practical approach for modelling vibrothermography. This effort brings FEM a step closer towards realistic simulation of vibrothermography in the future.

Research on Surface Crack Detection Based on Computer Image Recognition

In the process of large-scale industrial production, in order to control the surface accuracy and quality of components, it is necessary to carry out efficient detection of surface cracks. However, traditional manual visual inspection and other means are not conducive to large-scale industrial utilization due to the subjective human factors such as personnel experience and level. Based on this, this paper first analyses the basic principle of computer image recognition, and then studies the utilization of computer image recognition in surface crack detection, and gives the specific utilization steps, utilization methods and detection results.

Non-contact Crack Detection in Metals Using a Cutoff-Cavity Probe

This paper presents a non-contact method for the detection of surface cracks in metal materials through a forced-resonance microwave method (FRMM) using a cutoff cavity-backed narrow slot as a crack detection probe without using a vector network analyzer (VNA) at microwave frequencies. The FRMM uses the deviations in the ammeter or voltmeter readings of the forcefully obtained resonance of a cutoff-cavity probe for a metal material with or without cracks. The cutoff cavity-backed narrow slot on metal with no cracks produces a series resonance (maximum current) or a parallel resonance through an external control element located on a post inside the cutoff cavity. Cracks were detected by a change in this forced-resonance state (maximum current) when the cutoff-cavity probe was scanned over a crack. The characteristic crack signal was derived from the resonance current deviation on the ammeter located on a post inside the cavity probe. Galerkin’s method of moments was used to obtain a forced-resonance state from which the crack signal of the FRMM was calculated. The experimental measurements for non-contact (remote or lift-off) crack detection are also presented.

Detection and characterisation of short fatigue cracks by inductive thermography

ABSTRACT Inductive thermography can be excellently used to detect surface cracks in metals. A short induction heating pulse (0.1-1s) induces eddy currents in the sample and an infrared camera records the surface temperature distribution. As cracks disturb the eddy current distribution and the heat diffusion, they become visible in the infrared images. In this paper it is investigated, how different parameters influence the surface pattern around short cracks (0.5-12mm length). The main emphasis is on finite element simulations, but some experimental results are presented, too. The influence of crack geometry, as crack depth, length, inclination angle and crack shape below the surface are investigated for ferro-magnetic and austenitic steel. Around the crack tips high temperature ‘hot spots’ can be observed, which intensity increases with the crack depth. But this intensity is strongly affected by the crack shape, whether it is rectangular, trapezoid or half-penny shape. For longer cracks (6-8mm length) simulation results show, that in the middle of the crack the phase distribution can be used to estimate the crack depth. Furthermore, the effect of experimental parameters, as excitation frequency, heating pulse duration and the angle between crack line and induction coil are investigated in order to optimize an experimental setup.

Multi-sensor data collection and fusion using autoencoders in condition evaluation of concrete bridge decks

This paper presents a multi-sensor data collection and data fusion procedure for nondestructive evaluation/testing (NDE/NDT) of a concrete bridge deck. Three NDE technologies, vertical electrical impedance (VEI), ground-penetrating radar (GPR), and high-definition imaging (HDI) for surface crack detection, were deployed on the bridge deck. A neural network autoencoder was trained to quantify the relationship between VEI and GPR results using the data collected at common positions. This relationship was then used for fusion of VEI and GPR data to increase the reliability and spatial resolution of the NDE measurements and to generate a data-fused condition map that showed novel characteristics. Threshold values for VEI and GPR tests were obtained and used to determine the color scale in the fused map. Surface cracks identified from HDI show reasonable agreement with the deterioration areas on the data-fused condition map. Chloride concentration measurements on sound and deteriorated areas of the deck were consistent with the NDE results.

Semi-supervised semantic segmentation network for surface crack detection

The detection of surface crack is essential to ensure the safety and the serviceability of civil infrastructure. The automatic method is highly efficient and the test results are objective, which makes it gradually replace conventical manual inspection. Recently, semantic segmentation algorithms based on deep learning have shown excellent performance in crack detection tasks. However, the commonly used fully supervised segmentation method requires manual annotation of large amounts of data, which is time-consuming. In order to solve this problem, we propose a semi-supervised semantic segmentation network for crack detection. The proposed method consists of student model and teacher model. The two models have the same network structure and use the EfficientUNet to extract multi-scale crack feature information, reducing the loss of image information. The student model updates weights through the gradient descent of loss function, and the teacher model uses the exponential moving average weights of the student model. During training, the robustness of the model is improved by adding noise to the input data. When using only 60% of the annotated data, our method achieves an F1 score of 0.6540 on the concrete crack dataset and 0.8321 on the Crack500 dataset. The results show that our method can greatly reduce the workload of annotation while maintaining high accuracy.