Search

As one of the main pollution sources in China, haze can blur the railway monitoring and threaten railway safety. In this paper, we propose an end-to-end multi-scale residual network (MSRN) which can achieve remarkable dehazing effect on railway monitoring images. The method optimizes the image dehazing algorithm in three aspects: network structure, loss function and hazy dataset. Firstly, inspired by the residual network, the paper presents a method of fusing multi-scale feature information based on the residual network, which can extract more effective information at different scales. Secondly, a combined loss function is designed to achieve better convergent results by balancing training time, training calculations, and precision. Thirdly, the paper synthesizes an outdoor dataset specifically for railway scenarios, which relies on real depth maps and various outdoor images. Extensive experimental results on both full reference image quality assessment and no reference image quality assessment of image restoration demonstrate that the proposed algorithm shows higher performance than the state-of-the-art algorithms. Moreover, the haze of railway monitoring images is removed under hazy weather, and the detection algorithm achieved higher detection accuracy on the images after dehazing. The proposed network structure, loss function, and hazy dataset are discussed and analyzed in detail to verify the effectiveness of the proposed method.

On purpose of aiding detection and recognition for railway infrastructure and dramatic changes in the environment around railways, visual inspection based on unmanned aerial vehicle (UAV) images is a highlight. However, UAV images often suffer from degradation for fog or haze, which limits the inspection efficiency. Most existing methods depend on a suboptimal two-step network with much more redundant procedures where transmission map and atmospheric light are estimated at first, and then haze-free images can be acquired using a dehazing model. This paper presents a novel end-to-end network for UAV-based railway images dehazing, and focuses on two key issues: network architecture and loss function. With regards to the first aspect, based on a pyramidal network structure, densely pyramidal residual network (DPRnet) consists of dense residual block and enhanced residual blocks, which heavily exploits the feature maps of all preceding layers and considerably increased depth at different scale, respectively. With regards to the second, a new loss function introducing structural similarity index is proposed to preserve more structural information, thereby restore the appealing perceptual quality of the hazy images. Finally, quantitative and qualitative evaluations illustrate that the DPRnet achieves better performance over the classic methods, yet remains efficient and convenient.

Trains shuttle in semiopen environments, and the surrounding environment plays an important role in the safety of train operation. The weather is one of the factors that affect the surrounding environment of railways. Under haze conditions, railway monitoring and staff vision could be blurred, threatening railway safety. This paper tackles image dehazing for railways. The contributions of this paper for railway video image dehazing are as follows: (1) this paper proposes an end-to-end residual block-based haze removal method that consists of two subnetworks, namely fine-grained and coarse-grained network can directly generate the clean image from input hazy image, called RID-Net (Railway Image Dehazing Network). (2) The combined loss function (per-pixel loss and perceptual loss functions) is proposed to achieve both low-level features and high-level features so to generate the high-quality restored images. (3) We take the full-reference criterion (PSNR&amp, SSIM), object detection, running time, and sensory vision to evaluate the proposed dehazing method. Experimental results on railway synthesized dataset, benchmark indoor dataset, and real-world dataset demonstrate our method has superior performance compared to the state-of-the-art methods.

Total knee arthroplasty (TKA) is an effective treatment method for severe knee osteoarthritis and other knee-related diseases. Accurate match of prostheses is a crucial factor to improve the clinical efficacy and patients’ postoperative satisfaction in TKA, to which no enough attention is paid currently. In this paper, we introduce deep learning algorithm to analyze the patients’ multimodal data, such as preoperative radiograph of knees and relevant physical features (e.g. sex, height, weight), and design a software system for preoperative prediction of prosthetic type in TKA. The main processing steps include the pre-processing of X-ray images and the prediction of prosthetic type based on convolutional neural network. Research on loss function and model structure is implemented to fit the dataset better for further improvement of prediction accuracy. Transfer learning method is employed to address the problem of inadequate data. The experimental results shows that our prediction system can achieve the same accuracy level compared with that of traditional methods manipulated by experienced doctors, while it can complete the preoperative prediction automatically with lower cost and better stability.

To eliminate the shortcomings in the grouting sleeve and other connection forms in precast concrete shear wall structure induced by the poor grouting, difficult quality assurance, low construction efficiency, and high cost, a new type of monolithic assembled concrete shear wall structure with non-connected vertical distributed reinforcement was proposed in this paper. The new shear wall structure is composed of middle precast wall panels and cast-in-place boundary members. The vertical distributed reinforcement in the middle precast wall panels is not connected between the upper and lower floors. This technique was based on the normal section bearing capacity of equal strength and inclined section bearing capacity of equal strength compared with the cast-in-place shear wall structure. Three full-scale models of precast shear walls and one cast-in-place contrast specimen were tested for their seismic performance under pseudo-static loading. All precast specimens are cast based on the manufacture specification of the factory. By analysis of test results, including the failure modes, load-lateral drift hysteresis curves , bearing capacity, hysteretic characteristics, deformation components, ductility, rigidity degeneration, energy dissipation , and strain distribution, all the specimens were obtained and compared. Comparisons between test results indicate that the proposed shear walls have a similar bearing capacity as that of the cast-in-place shear wall and improved energy dissipation and ductility. Moreover, the specimen with inclined reinforcement performs better in drift capacity compared to companion specimens. The contribution of flexural deformation to the total displacement of each specimen is the largest, follow by the contribution of shear deformation , and the contribution of slip is the smallest. The strain of non-connected vertical distributed reinforcement of the precast panel did not yield during the loading process. The cross-section of the precast shear wall panel conforms to the assumption of plane cross-section according to the strain distribution before the reinforcement yields.

In order to further improve the accuracy of propellant gauging during the geosynchronous satellite transfer orbit, combined with the existing propellant gauging method, a time series analysis method for satellite propellant consumption fusion algorithm is proposed. The error characteristics of the prediction data are analyzed according to the principle of different gauging methods. On the basis of this, the variance of each gauging method is estimated by time series analysis and the optimal weighted fusion estimation based on least squares method is given. The practical application shows that the fusion algorithm can make full use of the advantages of different gauging methods and provide a more reasonable estimation of the real-time propellant consumption.

In this paper, the stochastic distribution of steel fiber in concrete beam is realized by using MATLAB modeling software. Firstly, MATLAB software is used to establish random distribution of steel fibers of different number, length and position, and the steel fiber model is assembled in an embedded way with the concrete model. Then the ABAQUS software is used to carry out the three point bending monotonic loading test on the model, and the stress curves and deflection curves of different steel fiber concrete are obtained. The effect of steel fiber on tensile and crack resistance in concrete is analyzed. The effect of volume, length and location of steel fiber on stress, strain and deflection of concrete beam is studied. The results show that the increase of the volume fraction can decrease the deflection and increase the tensile strength of the material, while the length has no significant effect. The distribution can significantly affect the efficiency of the fiber.

Compared to virtual navigation methods like joystick-based teleportation in Virtual Reality (VR), real walking enables more natural and realistic physical behaviors and better overall user experience. This paper presents a smartphone-based hybrid interface that combines a smartphone and a handheld controller, which enables real walking by viewing the physical environment and asymmetric 2D-3D input at the same time. The phone is virtualized in VR and streams a view of the real world to a collocated virtual screen, enabling users to avoid or remove physical obstacles. The touchscreen and the controller provide an asymmetric input choice for users to improve interaction efficiency in VR. We implemented a prototype system and conducted a pilot study to evaluate its usability. Refereed/Peer-reviewed