Detection Of Rail Surface Defects Research Articles

Hybrid deep learning architecture for rail surface segmentation and surface defect detection

AbstractRail surface defects (RSDs) are a major problem that reduces operation safety. Unfortunately, the existing RSD detection systems have very limited accuracy. Current image processing methods are not tailored for the railway track and many fully convolutional networks (FCN)‐based methods suffer from the blurry rail edges (RE). This paper proposes a new rail boundary guidance network (RBGNet) for salient RS detection. First, a novel architecture is proposed to fully utilize the complementarity between the RS and the RE to accurately identify the RS with well‐defined boundaries. The newly developed RBGNet injects high‐level RS object information into shallow RS edge features by a progressive fused way for obtaining fine edge features. Then, the system integrates the refined edge features with RS features at different high‐level layers to predict the RS precisely. Second, an innovative hybrid loss consisting of binary cross entropy (BCE), structural similarity index measure (SSIM), and intersection‐over‐union (IoU) is proposed and equipped into the RBGNet to supervise the network and learn the transformation between the input and ground truth. The input and ground truth then further refine the RS location and edges. Conveniently, an image‐based model for RSD detection and quantification is also developed and integrated for an automatic inspection purpose. Finally, experiments conducted on the complex unmanned aerial vehicle (UAV) rail dataset indicate the system can achieve a high detection rate with good adaptation capability in complicated environments.

Visual quantitative detection of rail surface crack based on laser ultrasonic technology

The surface rolling contact fatigue crack is a common crack on the rail surface. When it extends to a certain length, it will lead to rail fracture and affect the safety of railway operation. Therefore, it is necessary to carry out quantitative detection of RCF cracks to ensure the safety of railway operation. The state-of-the-art shows that the visualization method has advantages for quantitative crack detection. In this study, a quantitative detection method combining non-contact laser ultrasonic testing technology and variational mode decomposition (VMD) is proposed to realize the visual detection of the oblique crack on the railhead surface. Wigner time-frequency distribution and fir1 filtering are used to preliminarily filter all scanning signals. Further, VMD is utilized to decompose the signal into multiple intrinsic mode function (IMF) component signals and improve signal-to-noise ratio (SNR). According to the correlation coefficient (C) and SNR characteristics between different IMF components and the original signal, the optimal IMF component is selected. Finally, visual crack-induced surface wave energy from the ultrasonic propagation images is realized according to the time-domain and time characteristics of signals. The visual quantitative detection of RCF cracks in different lengths is realized clearly and intuitively, and the quantitative error is less than 5%. This method lays a foundation for the visual detection of laser ultrasonic rail surface defects.

Detection of Rail Surface Defects by Full Spectrum Image Processing with the Help of Peridynamic Based Fuzzy Logic Algorithm

Bu çalışmada tam spektrum görüntü üzerinden ray kusurlarının belirlenmesine yönelik peridinamik tabanlı bulanık mantık algoritması geliştirilmiştir. Örnek uygulama için tam spektrum ray kusuru görüntüsünün, gri skala tanım kümesinde Gaussian maskelemesi gerçekleştirilmiş, görüntünün farklı boyut ve mertebelerden türev alma işlemleri peridinamik sayısal türev alma metodu kullanılarak elde edilmitşir. Alınan türevler bulanık mantık sisteminde değerlendiriltikten sonra görüntünün RGB ve HSV skalaları üzerinde üç temel dalga boyu dönüşümü yapılmıştır. Bulanık mantık çıktısı üzerine tatbik edilen en etkin üç dalga boyu ile ortak noktaların tespiti gerçekleştirilerek ray kusurlarının daha belirgin bir şekilde ortaya çıkması sağlanmıştır. Ayrıca üç farklı tam spektrum örnek ray kusur görüntüsü ile geliştirilen algoritma test edilip, literatürdeki mevcut kenar bulma algoritmalarına göre daha iyi sonuç verdiği tespit edilmiştir.

A Defect Detection Method for Rail Surface and Fasteners Based on Deep Convolutional Neural Network.

As a result of long-term pressure from train operations and direct exposure to the natural environment, rails, fasteners, and other components of railway track lines inevitably produce defects, which have a direct impact on the safety of train operations. In this study, a multiobject detection method based on deep convolutional neural network that can achieve nondestructive detection of rail surface and fastener defects is proposed. First, rails and fasteners on the railway track image are localized by the improved YOLOv5 framework. Then, the defect detection model based on Mask R-CNN is utilized to detect the surface defects of the rail and segment the defect area. Finally, the model based on ResNet framework is used to classify the state of the fasteners. To verify the robustness and effectiveness of our proposed method, we conduct experimental tests using the ballast and ballastless railway track images collected from Shijiazhuang-Taiyuan high-speed railway line. Through a variety of evaluation indexes to compare with other methods using deep learning algorithms, experimental results show that our method outperforms others in all stages and enables effective detection of rail surface and fasteners.

Rail Surface Defect Detection Based on Image Enhancement and Improved Cascade R-CNN

Rail Surface Defect Detection Based on Image Enhancement and Improved Cascade R-CNN

Research on deep learning method for rail surface defect detection

Rail surface defect detection plays a critical role in the maintenance of the rail transportation system. Video analysis technology is a promising method to detect defects due to its low cost and effectiveness. Several attempts with hand-craft features have been made to obtain the detection results by using traditional machine vision algorithms. However, these methods suffer from imprecise results due to challenging conditions, such as deteriorated and changeable lighting environment and various types of complex rail surface defects. Recently, classification methods with complex deep convolutional networks have become popular. Despite their high accuracy, these methods cannot meet the requirements of defects localisation and real-time processing in practice. To solve these problems, this study proposes a novel object detection algorithm to detect rail defects. The net architecture of the proposed algorithm includes a backbone network using MobileNet and several novel detection layers with multi-scale feature maps inspired by you only look once (YOLO) and feature pyramid networks. Two different architectures of MobileNet are used to estimate the performance of defects detection. The experimental results demonstrate the great potential of the proposed algorithm with fast inference speed and high accuracy in the industry.

A surface defect detection system for railway track based on machine vision

In order to detect rail surface defects more quickly and effectively, a rail surface defect detection system based on machine vision was constructed. A perfect lighting system is built to obtain high quality image information. At the same time, the linear CCD camera is used to collect the rail information, which is uploaded by FPGA, and the image information is processed and analyzed by MATLAB software in the upper computer. Through the methods of image denoising and image enhancement, the image quality is improved, and the image information suitable for operation is obtained; then, the rail positioning is completed by threshold segmentation, gray compensation and image binarization to simplify the image operation; finally, the final positioning of defects is realized through edge detection and feature extraction. By building a test platform on the rail maintenance vehicle, the rail surface defect information collected by the rail maintenance vehicle is analyzed and processed.

Pulsed Air-Flow Thermography for Natural Crack Detection and Evaluation

Thermography is one of advanced non-destructive test and evaluation for imaging defects. This paper proposes pulsed air-flow thermography for sample surface crack detection and evaluation. The proposed approach combines the advantages of both convection heat transfer of air-flow and solid heat conduction. By controlling the pulse width, the heat waves of different frequencies can be modulated to penetrate materials with different thermal conductivity. And the cleaning degree of specimen surface can be improved by high-speed air-flow convection. The theoretical model and experimental facilities of air-flow thermography are presented. Meanwhile, the offline detection of rail surface defects and the dynamic test of the defects on the outer cylindrical surface of the turn-plate are completed. The offline experiments are conducted to identify the characteristics of the proposed method for the detection of artificial and natural cracks. And the dynamic experimental results show that the defects on the outer cylindrical surface of the turn-plate can be identified when the linear velocity at the edge of the turn-plate is 1.356km/h and 6.782km/h. The results demonstrate the proposed method on testing and evaluation for rail surface cracks is effective at the low speed.

Morphological Detection and Extraction of Rail Surface Defects

Rail inspection by means of a visual system has been a subject of a number of publications in recent years. The main requirements with regard to such a system are that it has to be fast, nondestructive, and accurate. This article presents a system for rail defect detection and shape extraction utilizing morphological operations. The proposed system is evaluated using publicly available rail surface discrete defects (RSDDs) data sets of Type-I and Type-II. Several versions of the system are compared, and the best one is reduced to just 17 basic morphological operations implementing defect detection, defect extraction, and false-positive (FP) reduction scheme. Two evaluation methods are considered, which is a pixel-level method and a defect-level method. 70% precision and 79% recall were obtained at the defect level for Type-I RSDD. Similarly, 82% precision and 89% recall were obtained at the defect level for Type-II RSDD. Experiments showed that the FP reduction scheme is effective for type II RSDD when considering defect-level results, and it increases precision by 7%–9% while maintaining the recall almost intact. The execution time of the morphological operations of the detection algorithm together with FP reduction is in the order of 50 ms on a usual laptop.

Unsupervised Saliency Detection of Rail Surface Defects using Stereoscopic Images

Visual information is increasingly recognized as a useful method to detect rail surface defects due to its high efficiency and stability. However, it cannot sufficiently detect a complete defect in the complex background information. The addition of surface profiles can effectively improve this by including a 3-D information of defects. However, in high-speed detection, the traditional 3-D profile acquisition is difficult and separate from the image acquisition, which cannot satisfy the above-mentioned requirements effectively. Therefore, an unsupervised stereoscopic saliency detection method based on a binocular line-scanning system is proposed in this article. This method can simultaneously obtain a highly precise image as well as profile information while also avoids the decoding distortion of the structured light reconstruction method. In our method, a global low-rank nonnegative reconstruction algorithm with a background constraint is proposed. Unlike the low-rank recovery model, the algorithm has a more comprehensive low rank and background clustering properties. Furthermore, outlier detection based on the geometric properties of the rail surface is also proposed in this method. Finally, the image saliency results and depth outlier detection results are associated with the collaborative fusion, and a dataset (RSDDS-113) containing the rail surface defects is established for the experimental verification. The experimental results demonstrate that our method can obtain a mean absolute error of 0.09 and area under the ROC curve of 0.94, better than 15 state-of-the-art algorithms.

A Coarse-to-Fine Model for Rail Surface Defect Detection

Computer vision systems have attracted much attention in recent years for use in detecting surface defects on rails; however, accurate and efficient recognition of possible defects remains challenging due to the variations shown by defects and also noise. This paper proposes a coarse-to-fine model (CTFM) to identify defects at different scales. The model works on three scales from coarse to fine: subimage level, region level, and pixel level. At the subimage level, the background subtraction model exploits row consistency in the longitudinal direction, and strongly filters the defect-free range, leaving roughly identified subimages within which defects may exist. At the next level, the region extraction model, inspired by visual saliency models, locates definite defect regions using phase-only Fourier transforms. At the finest level, the pixel subtraction model uses pixel consistency to refine the shape of each defect. The proposed method is evaluated using Type-I and Type-II rail surface defect detection data sets and an actual rail line. The experimental results show that CTFM outperforms state-of-the-art methods according to both the pixel-level index and the defect-level index.

Algorithms and Instrument for Rapid Detection of Rail Surface Defects and Vertical Short-Wave Irregularities Based on FOG and Odometer

Defects and vertical short-wave irregularities significantly influence the maintenance of the railway tracks and the train safety, and hence, they need to be detected and eliminated in the early stages of formation. We studied the principles and algorithms to detect the defects and to measure the vertical short-wave irregularities by using a fiber optical gyroscope and an odometer, and the associated software was also developed. The principles of deformation index, namely, the ratio of angular rate to velocity and the method to determine the threshold of defects detection, are expatiated. The influence of the transfer function in measuring the vertical irregularities has long been a tough issue. The relation between the characteristics of deformation index and the wavelength of irregularities is studied by converting the deformation index from the time domain to the spatial domain. The post-processing module utilizes the results of the spectrum analysis of the deformation index to determine the wavelength of the track irregularities and eliminate the influence of the transfer function. To improve the accuracy of the calculation results, we analyzed the characteristics of the output of the gyroscope and the odometer and designed a series of targeted error-elimination methods. The forward and reverse dead-reckoning method is utilized to reduce the error accumulation, especially a high frequency noise when calculating the magnitude of the vertical short-wave irregularities. The experimental results show that the repeatability is 0.038 mm when the chord length is 1 m and 0.65 mm when the chord length is 10 m. The accuracy is 0.05 mm when compared with the results obtained using a ruler of 1 m. The moving speed of the instrument was approximately 0.6–1.2 m/s during the experiment. Both the accuracy and the efficiency can fulfill the requirements for the measurement of the short-wave irregularities.

A UAV-Based Visual Inspection Method for Rail Surface Defects

Rail surface defects seriously affect the safety of railway systems. At present, human inspection and rail vehicle inspection are the main approaches for the detection of rail surface defects. However, there are many shortcomings to these approaches, such as low efficiency, high cost, and so on. This paper presents a novel visual inspection approach based on unmanned aerial vehicle (UAV) images, and focuses on two key issues of UAV-based rail images: image enhancement and defects segmentation. With regards to the first aspect, a novel image enhancement algorithm named Local Weber-like Contrast (LWLC) is proposed to enhance rail images. The rail surface defects and backgrounds can be highlighted and homogenized under various sunlight intensity by LWLC, due to its illuminance independent, local nonlinear and other advantages. With regards to the second, a new threshold segmentation method named gray stretch maximum entropy (GSME) is presented in this paper. The proposed GSME method emphasizes gray stretch and de-noising on UAV-based rail images, and selects an optimal segmentation threshold for defects detection. Two visual comparison experiments were carried out to demonstrate the efficiency of the proposed methods. Finally, a quantitative comparison experiment shows the LWLC-GSME model achieves a recall of 93.75% for T-I defects and of 94.26% for T-II defects. Therefore, LWLC for image enhancement, in conjunction with GSME for defects segmentation, is efficient and feasible for the detection of rail surface defects based on UAV Images.

Real time detection system for rail surface defects based on machine vision

The detection of rail surface defects is an important part of railway daily inspection, according to the requirements of modern railway automatic detection technology on real-time detection and adaptability. This paper presents a method for real-time detection of rail surface defects based on machine vision. According to the basic principle of machine vision, an image acquisition device equipped with LED auxiliary light source and shading box has been designed and the portable testing model is designed to carry on the field experiment. In view of the real-time requirement, the method of extracting the target area from the original image is carried out without image pre-processing. The surface defects of the rail are optimized based on morphological process and the characteristics of the defects are obtained by tracking the direction chain code. It is demonstrated that the maximum positioning time of this proposed method is 4.65 ms and its maximum positioning failure rate is 5%. The real-time detection speed of this proposed method can reach 2 m/s, which can carry out real-time detection of artificial hand walking. The time of processing each picture is up to 245.61 ms, which ensures the real-time performance of the portable track defect vision inspection system. To a certain extent, the system can replace manual inspection and carry out the digital management of track defects.

A 3D Laser Profiling System for Rail Surface Defect Detection.

Rail surface defects such as the abrasion, scratch and peeling often cause damages to the train wheels and rail bearings. An efficient and accurate detection of rail defects is of vital importance for the safety of railway transportation. In the past few decades, automatic rail defect detection has been studied; however, most developed methods use optic-imaging techniques to collect the rail surface data and are still suffering from a high false recognition rate. In this paper, a novel 3D laser profiling system (3D-LPS) is proposed, which integrates a laser scanner, odometer, inertial measurement unit (IMU) and global position system (GPS) to capture the rail surface profile data. For automatic defect detection, first, the deviation between the measured profile and a standard rail model profile is computed for each laser-imaging profile, and the points with large deviations are marked as candidate defect points. Specifically, an adaptive iterative closest point (AICP) algorithm is proposed to register the point sets of the measured profile with the standard rail model profile, and the registration precision is improved to the sub-millimeter level. Second, all of the measured profiles are combined together to form the rail surface through a high-precision positioning process with the IMU, odometer and GPS data. Third, the candidate defect points are merged into candidate defect regions using the K-means clustering. At last, the candidate defect regions are classified by a decision tree classifier. Experimental results demonstrate the effectiveness of the proposed laser-profiling system in rail surface defect detection and classification.

차세대 고속열차의 레일표면 결함 검출 시스템

차세대 고속열차의 레일표면 결함 검출 시스템

Surface defect detection for high-speed rails using an inverse P-M diffusion model

Purpose – When using a machine vision inspection system for rail surface defect detection, many complex factors such as illumination changes, reflection inequality, shadows, stains and rust might inevitably deform the scanned rail surface image. This paper aims to reduce the influence of these factors, a pipeline of image processing algorithms for robust defect detection is developed. Design/methodology/approach – First, a new inverse Perona-Malik (P-M) diffusion model is presented for image enhancement, which takes the reciprocal of gradient as feature to adjust the diffusion coefficients, and a distinct nearest-neighbor difference scheme is introduced to select proper defect boundaries during discretized implementation. As a result, the defect regions are sufficiently smoothened, whereas the faultless background remains unchanged. Then, by subtracting the diffused image from the original image, the defect features will be highlighted in the difference image. Subsequently, an adaptive threshold binarization, followed by an attribute opening like filter, can easily eliminate the noisy interferences and find out the desired defects. Findings – Using data from our developed inspection apparatus, the experiments show that the proposed method can attain a detection and measurement precisions as high as 93.6 and 85.9 per cent, respectively, while the recovery accuracy remains 93 per cent. Additionally, the proposed method is computationally efficient and can perform robustly even under complex environments. Originality/value – A pipeline of algorithms for rail surface detection is proposed. Particularly, an inverse P-M diffusion model with a distinct discretization scheme is introduced to enhance the defect boundaries and suppress noises. The performance of the proposed method has been verified with real images from our own developed system.

Adaptive noise cancelling and time–frequency techniques for rail surface defect detection

Adaptive noise cancelling (ANC) is a technique which is very effective to remove additive noises from the contaminated signals. It has been widely used in the fields of telecommunication, radar and sonar signal processing. However it was seldom used for the surveillance and diagnosis of mechanical systems before late of 1990s. As a promising technique it has gradually been exploited for the purpose of condition monitoring and fault diagnosis. Time–frequency analysis is another useful tool for condition monitoring and fault diagnosis purpose as time–frequency analysis can keep both time and frequency information simultaneously. This paper presents an ANC and time–frequency application for railway wheel flat and rail surface defect detection. The experimental results from a scaled roller test rig show that this approach can significantly reduce unwanted interferences and extract the weak signals from strong background noises. The combination of ANC and time–frequency analysis may provide us one of useful tools for condition monitoring and fault diagnosis of railway vehicles.

Railway wheel-flat and rail surface defect modelling and analysis by time–frequency techniques

Damage to the surface of railway wheels and rails commonly occurs in most railways. If not detected, it can result in the rapid deterioration and possible failure of rolling stock and infrastructure components causing higher maintenance costs. This paper presents an investigation into the modelling and simulation of wheel-flat and rail surface defects. A simplified mathematical model was developed and a series of experiments were carried out on a roller rig. The time–frequency analysis is a useful tool for identifying the content of a signal in the frequency domain without losing information about its time domain characteristics. Because of this, it is widely used for dynamic system analysis and condition monitoring and has been used in this paper for the detection of wheel flats and rail surface defects. Three commonly used time–frequency analysis techniques: Short-Time Fourier Transform, Wigner–Ville transform and wavelet transform were investigated in this work.

Railway Wheel Flat and Rail Surface Defect Detection by Time-frequency Analysis

Damage to the surface of railway wheels and rails commonly occurs in most railways and, if not detected at an early stage, can result in rapid deterioration and possible failure incurring high maintenance costs. If detected at an early stage these maintenance costs can be minimised. This paper presents an investigation into the use of time-frequency analysis of vibrations in railway vehicles for early detection of wheel flats and rail surface defects. Time-frequency analysis is a useful tool for identifying the content of a signal in the frequency domain without losing information about its time domain characteristics. Three commonly used time-frequency analysis techniques: Short-Time-Fourier-Transform (STFT); Wigner-Ville Transform (WVT); and Wavelet Transform (WT) are investigated in this work.