Search

When using a three-dimensional point cloud scanner to collect data points, due to the influence of various uncertain factors, the data points obtained by the three-dimensional electronic scanning device are always mixed with some noise points. If the noise contained in the point cloud is not filtered through technical methods, these noises will definitely have an impact on the subsequent extraction and recognition of the target set characteristics. Therefore, based on the accurate acquisition of 3D data, we research and design a fast denoising preprocessing algorithm that can be used for high-density 3D data to provide more accurate and effective 3D measurement data for subsequent 3D reconstruction and recognition processing.

Fasteners that hold rails in a fixed position are essentially important infrastructure components, and abnormal fasteners may cause a train to derail. The periodical inspection of fasteners is a significant guarantee for the safety of railway operation, and machine vision systems are popularly applied for fastener inspection. In this paper, we present an end-to-end abnormal fastener detection method, which identifies abnormal fasteners from a track image that contains a rail, fasteners, sleepers, etc. The proposed method is inspired from the well-established Faster R-CNN, but customizes with two aspects according to the application of fastener inspection. On the one hand, a light-weight backbone network is employed instead of complicated network to quicken detecting speed, and a threshold pruning algorithm is designed to reduce false positive rate. On the other hand, a hybrid loss function, combining weighted softmax loss function with center loss function, is devised to handle both the problems of class imbalance and small inter-class differences. Furthermore, two data synthesis methods are brought forward to solve the small sample problem. The proposed method is verified on a real-world image set of railway inspection, and the experimental results demonstrate that our method outperforms four established baselines.

We propose an automatic inspection method for detecting the railway fence defects based on train-borne line scan imaging. Once we get the whole fence images of continuous guardrail, the guardrail location of railway fence can be extracted as foreground by image segmentation algorithm; then we recognize the fence defect by the analysis of distance between the adjacent guardrails, based on the truth that the distance between any two normal guardrails is more or less fixed, while broken fence often causes the guardrail missing, which will cause the fixed distance change several-fold. We propose a maximum entropy thresholding segmentation based on three dimensional histogram MVG to realize the guardrail location, and the corresponding algorithm is also designed to analyze the range of guardrail interval for realizing the detection of defective fence. The experimental results show that the proposed algorithms not only perform well at the metal fence of general-speed ballast railway but also at the concrete fence of high-speed ballastless railway.

In this work, a model mimicking cerebral rhythms based on the coupled Wilson-Cowan oscillator with double nodes is proposed to achieve image enhancement. The inputs of the model are images to be enhanced and the outputs are node responses of excited subpopulation. To explain the mechanisms of the method, parameters are selected to meet conditions of limit cycles. Thus, in those conditions, the model produce oscillation by means of nonlinear dynamic analysis. In our experiments, image patches with continuous gray values are employed as stimulus and the response curves are similar to classical Gamma correction curves. Is it implicit that the Gamma correction can be explained by some cerebral rhythms? After some comparisons with other methods, the model proposed in the work shows better results.

In combination with the mathematical model and finite element analysis, we carry the research on tower crane structural damage detection method in the load lifting process. The paper takes the laboratory mini tower crane as the study object. First we establish the FEM model of the tower crane under the normal and the damage condition. Then get the dynamic displacement of the normal and the damage status under the lifting dynamic load. Finally establish the AR model and select the residual variance of AR model as the damage sensitive factor. The results of the study show that the residual variance in good condition is close to zero while the injure states is greater than zero and the more serious the damage, the greater the residual variance. These different provide a method for tower crane diagnosis.

In this paper, we present a method of estimating the vanishing point from the railway environment images. Vanishing point plays a very important role in the machine vision based railway-environment surveillance methods, e.g. estimating the pose of the camera, video segmentation and panorama. In the application of railway-environment surveillance, we most care about the vanishing point corresponding to the lines parallel to the rails. Based on the prior, we design a fast and effective method-Selected Line Segment Detector (SLSD) to detect the line segments corresponding to the scene-lines parallel to the rails, and then use a cross-iteration method to find the vanishing point in the railway environment images. We give both theoretical analysis and algorithm implementation. Our experimental results show the validation of the proposed method.

With the rapid development of the usage of digital imaging and communication technologies, there appears a great demand for fast and practical approaches for image quality assessment algorithms which can match human judgments accurately. In this paper, we draw the human visual characteristics into no-reference image quality assessment field, and propose a no-reference image quality assessment method based on the PSO-ε-SVR for JPEG compressed images. Firstly, we analyse the 5×5 filter operator and propose a 3×3 filter operator for 5×5's shortcomings. Then we use the 3×3 filter operator to extract the texture features of the distorted image, and extract the contrast and edge structure features, After that, we get the MTSC value which comes from these characteristics fusion. Moreover, we use the MTSC as the input object to establish prediction model of no-reference image quality assessment based on PSO-ε-SVR. In order to achieve the best effect, we use particle swarm optimization algorithm to find the optimal parameters of ε-SVR. Finally, the image quality is predicted by this method. The experimental results show that this method has good predictive ability, and its predictive value is high consistent with the DMOS value. The results can better simulate the human visual characteristics. And the method proposed in this paper is superior to the traditional evaluation algorithm in some performances.

In this paper, the authors present a method of detecting the rail track bends and flaws. Different from the traditional machine-vision based methods which analyze the images of the rail surface, the authors estimate the shake of the camera - which is caused by the flaw or bend of the rail track - from the video used for railway environment surveillance. The authors provide both theoretical analysis - based on the theory of optical flow and passive navigation - and the corresponding fast algorithms, i.e. the brightness-pattern-matching code (BPMC) and shaking-type code (STC). The authors method uses the surveillance video of the railway environment directly, thus, the authors do not need extra expensive instruments.

In this paper, we present a method of estimating the attitude of the camera, i.e. the angle between the optical axis and the direction of the camera's motion, from the image. For the machine vision based railway-environment-surveillance methods, the alignment between the world coordinate system and the camera coordinate system is very important. The alignment error will influence the analyzed results severely, because the scene depth is very large. Thus, estimating the attitude of the camera and adjusting the machine vision algorithms by the attitude-estimation results are important preprocessing steps. Moreover, we give both theoretical analysis and algorithm implementation of detecting the vanishing points based on the selected line segment detector (SLSD) and least-square technique. The experimental results show the validation of our method.

The video captured from forward moving camera has been used for environment surveillance, endoscopy, pipeline inspection and mobile navigation, from which 3D information acquisition will further facilitate the applications. In this paper, we proposed a new method named bi-slit projections to achieve the stereo imaging from forward motion video captured by only single camera. The method only requires simple geometrical calculation without computation of any 3D structure. The epipolar geometry about parallax and baseline also be formulated in detail. The experimental results including indoor and outdoor scene demonstrate the effectiveness of our proposed method by generating the stereoscopic 3D image and calculating the lateral depth map.

The high demand from railway safety has promoted the sensing technology to avoid manual inspection that is error-prone, low efficiency, labor-intensive and high costs. However, few works have emphasized railway environmental monitoring so far which is equally important in railway safety maintenance. In this paper, we propose a railway scene rendering system for virtual maneuver and inspection. Based on the route panoramas extracted from a real train-borne video, we generate the compact and high fidelity scenes over a long distance route in cyberspace. We analyze the relation between train motion and the video sampling in order to obtain the route panoramas of the railways at the just-sampling rate. Then, we render scene tunnels based on the route panorama in a virtual space for interactive navigation, inspection, and scene indexing. The experimental results show remarkable advantages in the data storage and automatic inspection. The compact data format with sufficient visual information can be transmitted and released in cyberspace for fast access, maneuver, and automatic retrieval in railway environment monitoring.

With the rapid development of high-speed railway, the safety running of trains becomes extremely urgent. Video is a direct and effective manner for railway environment monitoring. In this paper, we propose a route panoramic representation to produce a virtual environment of the railway scene from a train-borne video, which provides an all-round display for the entire length of railway. We generate the route panorama from the forward motion video and then render 3D scene. The experimental results show many advantages of the representation in the large data storage, browsing, and examination. It will be used for railway safety checking, railway facility inspection and virtual sightseeing from the train in the future.

The method of intelligent fault diagnosis based on wavelet packet transform and support vector machine is presented. The key to fault diagnosis is feature extracting and fault classifying. According to the method, the energy of frequency bands after wavelet packet decomposition of the vibration signals is taken as the extracted features and input to support vector machine(SVM). Then, fault pattern is recognized using SVM multiple fault classifier. Finally, the proposed method is applied to the fault diagnosis of rolling bearings, and the results of the experiment demonstrate that the proposed method is efficient in intelligent fault diagnosis.

We propose a new approach based upon the adaptive chirplet decomposition to characterize the time-dependent behavior of machine vibration signals. This approach employs maximum projective decomposition algorithm combined fractional Fourier transform with quasi-Newton method to estimate the parameters. Then, the expectation maximization algorithm is used to refine the results. Compared with traditional time-frequency analysis methods, such as short-time Fourier spectrum and Wigner distribution, the simulation results show that this method can obtain more accurate estimation, finer time- frequency resolution and de-noising capability. Finally, the proposed method is applied to the fault diagnosis of bearing, and the results of the experiment demonstrate that the proposed method is efficient in signal feature extraction. the adaptive decomposition (ACD) algorithm. Our approach also incorporates fractional Fourier transform (FRFT) to estimate the chirp rate, and employs expectation maximization method to refine the parameters. We demonstrate the adaptive chirplet decomposition, and yield a better time-frequency distribution compared with the traditional, STFT and WVD. Finally, the proposed method is applied to the fault diagnosis of bearing, and the results of the experiment demonstrate that the proposed method is efficient in signal feature extraction. Following this introduction, Section II of this paper introduces the adaptive chirplet decomposition and shows the implementation of the algorithm. In Section III, we verify the method by simulation. In Section IV, we apply the method to fault diagnosis experiment and finally the conclusion is summarized in Section V.