Self-movable subway tunnel structure apparent disease detection equipment and method

Abstract

The present invention provides a self-moving subway tunnel structure apparent disease detection equipment and method. The present invention is based on a wide-area shooting area array camera, splices, processes and interprets the acquired images, and combines mobile measurement and precise dead reckoning to achieve high-precision and rapid detection of multiple diseases in the entire subway tunnel. It aims to solve the problems of slow manual detection, low accuracy of disease identification relying on traditional experience, and low intelligence level of disease detection equipment.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel detection, and specifically relates to a self-moving subway tunnel structure apparent disease detection equipment and method, and in particular to a self-moving subway tunnel structure apparent disease detection equipment and method based on a wide-area shooting array camera. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the rapid economic growth and rapid urbanization, urban rail transit has developed rapidly. Due to the extremely complex tunnel environment, complex geological environment, and the frequent impact of operating vibration, construction disturbance, extreme weather and other factors, underground engineering structural diseases are easily caused.

[0004] As the service time is extended, the operating mileage increases year by year, and the number of defects increases year by year, operating tunnels are facing massive detection needs. However, traditional detection methods are mainly based on manual detection, which has low efficiency, difficult data backtracking, and difficult to guarantee accuracy. Therefore, it is urgent to develop intelligent tunnel disease detection equipment to meet the growing demand for disease detection. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a self-moving subway tunnel structure surface disease detection equipment and method. The present invention is based on a wide-area shooting area array camera to stitch, process and interpret the acquired images, and combines mobile measurement and precise dead reckoning to achieve high-precision and rapid detection of multiple diseases in the entire subway tunnel, aiming to solve the problems of slow manual detection, low accuracy of disease identification relying on traditional experience, and low intelligence level of disease detection equipment.

[0006] According to some embodiments, the present invention adopts the following technical solutions:

[0007] A self-moving subway tunnel structure surface disease detection equipment, including a self-moving trolley, a wide-area shooting array camera, a laser scanning system, a control system and an equipment platform, wherein:

[0008] A walking mechanism is provided at the bottom of the self-propelled trolley, and the walking mechanism is driven by a driving mechanism. The walking mechanism is used to move bidirectionally / unidirectionally along the tunnel track and stop at a set measuring point; a displacement sensor and a mileage encoder are provided on the walking mechanism;

[0009] The self-moving vehicle is provided with an equipment platform, and the wide-area shooting area array camera can be rotatably arranged on the equipment platform;

[0010] The wide-area shooting area array camera has a bracket in the form of a semicircular arc, and a plurality of area array cameras are sequentially arranged on the outer arc of the bracket;

[0011] The laser scanning system is used to obtain point cloud data in the tunnel;

[0012] The control system is used to perform dead reckoning based on information obtained by the displacement sensor and the mileage encoder, and receive images and point cloud data of each measuring point position, perform panoramic stitching on each image based on each image obtained by the wide-area shooting area array camera, analyze the stitched images, and identify defect cracks; combine the dead reckoning and point cloud data to fuse and generate a three-dimensional point cloud scene, map and align the point cloud data and image information, and realize the positioning of the defect cracks.

[0013] As an optional implementation, both ends of the self-moving vehicle are provided with a vehicle head, the vehicle head is provided with a searchlight system, and the vehicle head is provided with a camera system and a radar system.

[0014] As an optional implementation, the equipment platform is arranged in the middle position of the self-moving vehicle, and a lifting mechanism is arranged at the lower end of the platform, and there is a loading space on the platform.

[0015] As an optional embodiment, a fixed guide rail is provided on the equipment platform, a rotating device is movably provided on the fixed guide rail, the rotating device has a certain degree of rotational freedom, the position of the rotating device on the fixed guide rail is changeable, a carrying device is provided on the rotating device, and the bracket is provided on the carrying device.

[0016] As an optional embodiment, the walking mechanism is a wheel hub, which is arranged on both sides of the self-moving trolley, the wheel hub moves along the tunnel track, and liquid-gas hybrid shock absorbers are arranged on both sides of the wheel hub, and the wheel hub is driven by a wheel hub motor;

[0017] The walking mechanism is also provided with an inclination sensor to detect the inclination angle of the self-moving vehicle.

[0018] As an optional implementation, each area array camera measurement surface is provided with a surface state measuring device and a fill light source, and the area array camera is equipped with a lens automatic focusing module.

[0019] The working method based on the above self-moving subway tunnel structure apparent disease detection equipment includes the following steps:

[0020] When the self-moving vehicle moves to the predetermined measuring point, the wide-area shooting array camera and the laser scanning system are started simultaneously to obtain the image data and point cloud data of the corresponding scene at the same time;

[0021] Based on the information obtained by the displacement sensor and the mileage encoder, dead reckoning is performed, and the dead reckoning and point cloud data are combined to generate a 3D point cloud scene.

[0022] Based on the images acquired by the wide-area array camera, the images are panoramically stitched, and the stitched images are analyzed to identify defective cracks;

[0023] The point cloud data and image information are mapped and registered to locate the defective cracks.

[0024] As an optional implementation, the specific process of performing dead reckoning includes:

[0025] Get the coordinate information of pre-configured measuring points;

[0026] Based on the coordinate information of the measuring point and the current position obtained by the displacement sensor and the mileage encoder, forward and reverse dead reckoning are performed to determine the initial aerial survey position;

[0027] Combine the initial aerial survey positions with the 3D point cloud scene of the tunnel;

[0028] The actual sensor data is used to correct the initial aerial survey position, and the corrected aerial survey position information is integrated with the 3D point cloud scene again to form the final true 3D point cloud scene.

[0029] As an optional implementation, the specific process of mapping and registering the point cloud data and the image information includes:

[0030] Establish the spatial projection relationship between the laser scanning system and the area array camera, take multiple images and perform multiple scans in an environment with known geometric features, and use the acquired data to calculate the external and internal parameters;

[0031] Using the spatial projection relationship obtained by calibration, the point cloud and image are converted from their respective three-dimensional coordinate systems to a unified two-dimensional plane. The point cloud is mapped to the two-dimensional plane through the orthographic projection algorithm to form a point cloud map; the image maintains its two-dimensional characteristics to form an image map;

[0032] On a two-dimensional plane, the point cloud image is converted into a grayscale image, and the point cloud grayscale image and the image image are accurately registered using image processing technology;

[0033] Based on the point cloud grayscale image and the registered image, subtle defects on the tunnel surface are identified.

[0034] As an optional implementation, the process of acquiring an image with a wide-area shooting area array camera includes:

[0035] Correct the focal length and viewing angle information of each camera lens to ensure that the positional relationship of each camera in space is consistent. By knowing the size and distance of the object, adjust the parameters of each camera lens so that the imaging results of each camera lens can correspond to each other.

[0036] Use a distributed autofocus algorithm to unify the focus results of each camera;

[0037] The imaging results of different camera lenses are combined to form a complete image based on the information weight of each pixel.

[0038] As an optional implementation method, the specific process of intelligent interpretation of disease images includes:

[0039] Access existing images covering different types of diseases;

[0040] Preprocess the image.

[0041] Extract features including crack size, texture, shape, or edge information;

[0042] Use the extracted features to train a machine learning model;

[0043] Use the trained machine learning model to process the new images acquired.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention uses a self-propelled vehicle equipped with a multifunctional fixed equipment platform to achieve automatic and rapid conversion of different equipment, and the cruising range is further improved. The self-propelled vehicle is equipped with a multifunctional self-propelled carrying platform, which has the characteristics of platform universalization, functional modularization, and standardized interface. It has a high degree of modularization and automation, and can carry and quickly switch different equipment.

[0046] In order to solve the problem of limited detection range, the present invention utilizes a wide-area shooting area array camera module in the form of a semicircular arc and is equipped with a lens automatic focusing module to realize high-speed real-time synchronous sampling and scan defects in tunnels above a certain angle, thereby achieving optimal space utilization and lightweight design.

[0047] The area array camera of the present invention is equipped with a surface state measurer and a fill light source, which achieves the best resolution and imaging angle of view in different motion scenes of a single camera, can ensure that the collected image data is clear and without frame loss, and realizes high-definition image collection under high-speed moving conditions.

[0048] The wide-area shooting area array camera of the present invention is arranged on a platform through a rotating mechanism, and the position of the rotating mechanism is movable. The shooting position and shooting angle can be flexibly adjusted according to the measurement point situation, and a tilt sensor is configured to detect the tilt angle of the self-moving trolley to ensure the stability of the trolley when shooting images, or the tilt angle is compensated by adjusting the posture of the wide-area shooting area array camera.

[0049] The present invention utilizes a control system to achieve one-button automatic matching focus of multiple lenses, and utilizes fill light to ensure that image acquisition is almost not affected by light interference, thereby effectively improving the accuracy of later image stitching and image recognition.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0052] Figure 1 It is a structural schematic diagram of a self-movable subway tunnel structure apparent disease detection equipment;

[0053] Figure 2 It is a schematic diagram of the structure of a wide-area imaging area array camera.

[0054] Among them, 1. Self-moving vehicle, 2. Searchlight system, 3. Self-moving vehicle travel system, 4. Camera system and radar system, 5. Embedded control panel, 6. Multifunctional fixed equipment platform, 7. Front and rear moving guide rails, 8. Carrying device, 9. Area array camera, 10. Surface state measuring device, 11. Camera module mounting bracket, 12. Rotating device, 13. Rotating device fixing guide rail, and 14 is laser scanning system. DETAILED DESCRIPTION

[0055] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0056] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0057] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0058] In the absence of conflict, the embodiments in this application and the features in the embodiments may be combined with each other.

[0059] Embodiment 1

[0060] like Figure 1 As shown, a self-moving subway tunnel structure surface disease detection equipment based on a wide-area shooting array camera includes:

[0061] The self-propelled vehicle 1 can move in two directions and move along the tunnel track through the moving wheels on both sides of the vehicle body. The inner side of the wheel hub is provided with a displacement sensor, a tilt sensor and a mileage encoder. Liquid-gas hybrid shock absorbers are suspended on both sides of the wheel hub to better absorb impact, improve driving stability, and ensure fast and high-precision image acquisition.

[0062] Both front ends of the self-propelled vehicle are equipped with a searchlight system 2, equipped with an advanced attitude stabilization system, and can be used for observation without being affected by light under various adverse conditions such as foggy days, greatly broadening the scope of operation.

[0063] The camera system and radar system 4 are located on the front side above the front of the self-propelled vehicle 1, and are used to detect the road conditions of the traveling section and transmit signals to the hub motors of the traveling wheel hubs.

[0064] The built-in control panel 5 is located at the front side of the unmanned driving cabin of the self-propelled vehicle, and is equipped with a high-precision inertial navigation system, a wireless transmission and a remote control system.

[0065] The multifunctional fixed equipment platform 6 is located between the two front ends of the self-moving vehicle. The platform can be raised and lowered with one button and can be used to carry different equipment.

[0066] The wide-area shooting area array camera device is installed and fixed on the carrying device 8 in a semicircular arc form. A rotating device 12 and a front and rear movable guide rail 7 are provided below the carrying device 8. The rotating device 12 is movably arranged on the front and rear movable guide rail 7, so that a disease scanning of more than 240 degrees of the tunnel can be realized.

[0067] Each area array camera 9 is equipped with an electric focusing module for adjusting the focal length, focus and aperture, thereby improving the adaptability of the equipment to complex environments.

[0068] Each area array camera 9 is also equipped with a surface state measuring device 10 to cope with the surface state measurement in complex environments and detect defects as early as possible; and an LED fill light source is provided to evenly and effectively illuminate the tunnel wall without affecting driving safety, so that the camera can still capture sufficiently clear images during high-speed driving, ensuring the shooting quality in dim environments.

[0069] The camera module mounting bracket 11 is fixed on the mounting device 8 via a U-shaped clamp.

[0070] In this embodiment, the wide-area shooting area array camera adopts a high-density stroboscopic light source and a high-precision electronically controlled lens, which can realize high-definition image acquisition under high-speed moving conditions. Through one-click automatic matching focus of multiple lenses, it ensures that image acquisition is almost not affected by light interference.

[0071] The system also includes a processor or a control system, which is used to perform dead reckoning based on information obtained by the displacement sensor and the mileage encoder, and receive images and point cloud data of each measuring point position, perform panoramic stitching on each image based on each image obtained by the wide-area shooting area array camera, analyze the stitched images, and identify defective cracks; combine dead reckoning and point cloud data to fuse and generate a three-dimensional point cloud scene, map and align the point cloud data and image information, and realize the positioning of the defective cracks.

[0072] In this embodiment, the laser scanning system can be set on one side of the front of the vehicle, and an installation port is also reserved on the other side of the front of the vehicle, which can be adjusted to be installed on the other side according to actual needs; the accuracy and resolution of the laser scanner can be set through the dedicated control software embedded in the embedded control panel, and the scanning settings of the laser scanning system can be performed in the embedded control panel, and the settings can be completed after selecting the required resolution and accuracy.

[0073] The laser scanning system consists of a long-range laser scanner and a short-range laser scanner. The long-range laser scanner is suitable for large-sized tunnels, and the short-range laser scanner is suitable for smaller tunnels or when higher details are required.

[0074] The laser scanning system used in this embodiment has a 360° panoramic scanning function and a high scanning speed, with a maximum speed of 1.2 million points per second, which can quickly and accurately capture the tunnel contour while the vehicle is moving. The laser scanning system has an accuracy of up to 0.020 mm and a maximum resolution of 0.010 mm, and can accurately obtain detailed information and complete three-dimensional data on the surface of complex objects.

[0075] Embodiment 2

[0076] The working method of the self-movable subway tunnel structure apparent disease detection equipment as described in the first embodiment includes the following steps:

[0077] When the self-moving vehicle moves to the predetermined measuring point, the wide-area shooting array camera and the laser scanning system are started simultaneously to obtain the image data and point cloud data of the corresponding scene at the same time;

[0078] Based on the information obtained by the displacement sensor and the mileage encoder, dead reckoning is performed, and the dead reckoning and point cloud data are combined to generate a 3D point cloud scene.

[0079] Based on the images acquired by the wide-area array camera, the images are panoramically stitched, and the stitched images are analyzed to identify defective cracks;

[0080] The point cloud data and image information are mapped and registered to locate the defective cracks.

[0081] The specific process is described below.

[0082] First, the multiple lenses of the wide-area array camera are focused and the images collected are stitched. In some embodiments, panoramic stitching software is configured to stitch the photos taken by each camera into a panoramic image. The stitched images are analyzed in real time through the intelligent video analysis software installed, and wall cracks are identified through algorithms.

[0083] This embodiment uses a sensor on a camera or other optical device to detect the distance information of the photographed object, and then calculates the best focus point through a built-in algorithm to achieve the purpose of clear imaging. In a system with multiple lenses, each lens can have its own sensor and focus module for independently performing the autofocus function.

[0084] The specific process includes:

[0085] (1) Lens calibration and synchronization: In a system with multiple lenses, the focal length, viewing angle and other information of each lens are first calibrated to ensure that their positions in space are consistent. By knowing the size and distance of the object, the parameters of each lens are adjusted so that their imaging results can correspond to each other.

[0086] (2) Autofocus algorithm: A distributed autofocus algorithm is used. Each lens performs independent autofocus based on its own sensor readings, and then the autofocus mechanism unifies their results.

[0087] (3) Image fusion and matching: After multiple lenses are focused, the imaging results of different lenses are fused together to form a complete, high-quality image based on the information weight of each pixel.

[0088] (4) Real-time feedback and adjustment: In the automatic matching focus technology, a real-time feedback mechanism for image quality assessment is set up. The image clarity evaluation function is used to determine which focus point has the highest imaging quality, so that the focus strategy can be adjusted according to the actual imaging effect.

[0089] In the process of identifying diseases based on the spliced ​​images, the existing detection frame or image recognition network can be used for disease identification. The specific steps are as follows:

[0090] (1) Image preprocessing: including image resizing, cropping, denoising, contrast enhancement, etc., to facilitate subsequent processing and analysis. The image brightness and contrast can be adjusted to highlight the diseased area, or a filter can be used to remove noise from the image.

[0091] (2) Feature extraction: After image preprocessing, the next stage is feature extraction. The purpose of this stage is to extract disease-related features from the image to facilitate subsequent identification and classification.

[0092] The feature extraction in this embodiment uses deep learning technology to automatically extract features, which mainly includes the steps of data collection and preprocessing, data annotation, selection of appropriate deep learning model, model training, verification and evaluation, feature extraction, post-processing and analysis.

[0093] a. Data collection and preprocessing: First, collect a large amount of tunnel disease image data, and then preprocess the images to improve the accuracy and efficiency of model training and testing. The preprocessing steps may include image enhancement (such as rotation, scaling, shearing), denoising, contrast adjustment, brightness adjustment, and image size normalization.

[0094] b. Annotated data: Annotations can be bounding boxes, polygons, or pixel-level labels.

[0095] c. Choose a suitable deep learning model: Choose a suitable deep learning model according to the specific requirements of the task. Commonly used models include convolutional neural networks (CNNs) such as AlexNet, VGGNet, ResNet, etc. For image segmentation problems, you can choose models such as U-Net, FCN, Mask R-CNN, etc.

[0096] d. Model training: Use the labeled dataset to train the selected deep learning model. This process involves adjusting the model parameters to minimize the loss function and thus improve the model's predictive power. Hyperparameter tuning may also be required during the training process, such as learning rate, batch size, and optimizer selection.

[0097] e. Verification and evaluation: Evaluate the performance of the model on an independent test set to verify its generalization ability. Common evaluation indicators include accuracy, recall, F1 score, IoU (intersection over union), etc. Depending on the evaluation results, you may need to go back and adjust the model structure or training strategy.

[0098] f. Feature extraction: Once the model has been trained and validated, it can be used to extract features from new tunnel disease images. This is usually achieved through forward propagation, that is, inputting an image to a trained model and outputting the feature representation learned by the model.

[0099] g. Post-processing and analysis: The extracted features usually need further processing and analysis to obtain information about the type, location, severity, etc. of tunnel defects. This may involve machine learning methods such as cluster analysis, anomaly detection, and regression analysis.

[0100] (3) Defect recognition: Use machine learning or deep learning models to identify defects in images. Convolutional neural networks (CNNs) can be used to identify tunnel cracks or other types of defects. Model training usually requires a large amount of labeled data to learn how to distinguish normal and abnormal image features.

[0101] (4) Result verification and evaluation: The recognition results are verified and evaluated to ensure the accuracy and reliability of the recognition, and an independent test dataset is used to evaluate the performance of the model.

[0102] This embodiment can also combine inertial navigation system (INS), laser scanning and other sensor data to achieve dynamic precision measurement without GNSS signal. Through real-time correction and data fusion technology, the cumulative error of high-frequency position and attitude information provided by the inertial navigation system is effectively reduced, and the overall measurement accuracy is improved. The absolute accuracy of tunnel dynamic measurement is increased to 3 mm.

[0103] The specific dead reckoning process includes the following steps:

[0104] (1) Obtain the coordinate information of predefined control points (or measuring points). Control points can be specific locations in the tunnel or key points such as tunnel entrances and exits.

[0105] (2) Forward / reverse dead reckoning initial POS: The dead reckoning positioning method predicts the future position based on the known position information and the current sensor data. Forward dead reckoning means predicting the position of the self-moving vehicle in the tunnel from a known starting point; reverse dead reckoning means reversing the position of the self-moving vehicle from a known end point. These two methods can be used to obtain the initial train position information, i.e., the initial POS.

[0106] (3) Fusion of the initial 3D point cloud scene: The initial POS is combined with the 3D point cloud scene of the tunnel. A point cloud is a data set containing a large number of 3D coordinate points representing the surface of an object. By fusing the initial POS and the point cloud scene, a more realistic 3D tunnel model can be created.

[0107] (4) Track correction to obtain accurate POS information: Use actual sensor data to correct the initial POS to obtain more accurate position information.

[0108] (5) Forming a fused true 3D point cloud scene: The corrected precise POS information is integrated with the 3D point cloud scene again to form a real fused true 3D point cloud scene that reflects the current position of the self-moving vehicle.

[0109] In order to solve the problem that too much noise on the tunnel structure surface affects the reliability of tunnel structure analysis, this embodiment uses a tunnel point cloud denoising method based on wavelet analysis. In terms of denoising, wavelet transform can decompose a signal or image into a series of sub-signals or sub-images in scale and orientation. These sub-signals or sub-images contain different frequency components of the original signal or image.

[0110] By properly processing these sub-signals or sub-images, noise can be effectively removed and useful signal or image information can be retained. Wavelet transform is used to achieve tunnel cross-section point cloud data scale segmentation, noise extraction and filtering, which solves the problem of interference of tunnel ancillary facilities and surface defects on measurement accuracy and realizes the automatic extraction of tunnel convergence, deformation, misalignment, and intrusion parameters.

[0111] Of course, other denoising methods can also be used.

[0112] In order to improve the accuracy and precision of disease detection, this embodiment also provides a method for converting point clouds and images into two-dimensional tunnel point cloud maps and image maps through spatial orthographic projection, and then achieving pixel-level fusion and registration of point cloud grayscale maps and images through calibration.

[0113] Specifically include:

[0114] (1) Time synchronization: Ensure that the timestamps of the data collected by the laser scanning system and the area array camera are the same to ensure that the point cloud and image data correspond to the scene at the same time.

[0115] (2) Spatial calibration: Establish the spatial projection relationship between the laser scanning system and the area array camera. Take multiple images and perform multiple scans in an environment with known geometric features, and use these data to calculate external parameters (such as rotation and translation matrices) and internal parameters (such as the focal length and distortion coefficient of the camera).

[0116] (3) Spatial orthographic projection: Using the spatial projection relationship obtained by calibration, the point cloud and image are transformed from their respective three-dimensional coordinate systems to a unified two-dimensional plane. The point cloud is mapped to the two-dimensional plane through the orthographic projection algorithm to form a point cloud map; the image retains its original two-dimensional characteristics to form an image map.

[0117] (4) Pixel-level registration: On a two-dimensional plane, the pixels of the point cloud and the image can be directly matched. Since point cloud data usually contains grayscale information, the point cloud can be converted into a grayscale image. Then, through image processing techniques such as feature matching and transformation model fitting, accurate registration of the point cloud grayscale image and the image can be achieved.

[0118] (5) Post-processing: After the registration is completed, the fused data can be filtered, denoised, and enhanced to improve the quality of the final 3D model.

[0119] In this way, the high precision of point cloud data and the rich color information of image data can be fully utilized to provide strong data support for tunnel inspection, evaluation and maintenance.

[0120] By converting point cloud data into grayscale or depth maps, computer vision and machine learning technologies can be better used to analyze and identify subtle defects on the tunnel surface, efficiently solving problems such as subtle defects on the tunnel surface (0.3mm cracks, 5mm chipping, water seepage), and solving the problem of rapid positioning, analysis and extraction of railway tunnel defects.

[0121] This embodiment can achieve high-precision and rapid detection of multiple defects in the entire subway tunnel, aiming to solve the problems of slow manual detection, low accuracy of defect identification relying on traditional experience, and low intelligence level of defect detection equipment.

[0122] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0124] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0125] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principle of the present invention without creative labor shall be included in the protection scope of the present invention.

Claims

1. A self-moving subway tunnel structure apparent disease detection equipment, characterized in that: It includes a self-propelled vehicle, a wide-area array camera, a laser scanning system, a control system and an equipment platform, including: A walking mechanism is provided at the bottom of the self-propelled trolley, and the walking mechanism is driven by a driving mechanism. The walking mechanism is used to move bidirectionally / unidirectionally along the tunnel track and stop at a set measuring point; a displacement sensor and a mileage encoder are provided on the walking mechanism; The self-moving vehicle is provided with an equipment platform, and the wide-area shooting area array camera can be rotatably arranged on the equipment platform; The wide-area shooting area array camera has a bracket in the form of a semicircular arc, and a plurality of area array cameras are sequentially arranged on the outer arc of the bracket; Each area array camera measurement surface is equipped with a surface state measuring device and a fill light source, and the area array camera is equipped with a lens autofocus module; the autofocus module adopts a distributed autofocus algorithm; The laser scanning system is used to obtain point cloud data in the tunnel; The control system is used to perform dead reckoning based on information obtained by the displacement sensor and the mileage encoder, and receive images and point cloud data of each measuring point position, perform panoramic stitching on each image according to each image obtained by the wide-area shooting area array camera, analyze the stitched image, and identify defect cracks; combine the dead reckoning and point cloud data to fuse and generate a three-dimensional point cloud scene, map and align the point cloud data and image information, and use the calibrated spatial projection relationship to convert the point cloud and image from their respective three-dimensional coordinate systems to a unified two-dimensional plane. The point cloud is mapped to the two-dimensional plane through an orthographic projection algorithm to form a point cloud map, and the image maintains its two-dimensional characteristics to form an image map. On the two-dimensional plane, the point cloud map is converted into a grayscale map, and the point cloud grayscale map and the image map are accurately aligned by using image processing technology. Based on the point cloud grayscale map and the aligned image, subtle defects on the tunnel surface are identified to achieve the positioning of the defect cracks.

2. The self-moving subway tunnel structure apparent disease detection equipment as claimed in claim 1, characterized in that: Both ends of the self-propelled vehicle are provided with a vehicle head, the vehicle head is provided with a searchlight system, and the vehicle head is provided with a camera system and a radar system.

3. The self-moving subway tunnel structure apparent disease detection equipment as claimed in claim 1, characterized in that: The equipment platform is arranged at the middle position of the self-moving trolley, and a lifting mechanism is arranged at the lower end of the platform, and a loading space is provided on the platform.

4. The self-moving subway tunnel structure apparent disease detection equipment as claimed in claim 1, characterized in that: The equipment platform is provided with a fixed guide rail, on which a rotating device is movably provided. The rotating device has a certain degree of rotational freedom, and the position of the rotating device on the fixed guide rail is changeable. The rotating device is provided with a carrying device, and the bracket is provided on the carrying device.

5. The self-moving subway tunnel structure apparent disease detection equipment as claimed in claim 1, characterized in that: The walking mechanism is a wheel hub, which is arranged on both sides of the self-moving trolley. The wheel hub moves along the tunnel track direction, and liquid-gas hybrid shock absorbers are arranged on both sides of the wheel hub. The wheel hub is driven by a wheel hub motor; The walking mechanism is also provided with an inclination sensor to detect the inclination angle of the self-moving vehicle.

6. A working method of the self-movable subway tunnel structure apparent disease detection equipment according to any one of claims 1 to 5, characterized in that: The following steps are involved: When the self-moving vehicle moves to the predetermined measuring point, the wide-area shooting array camera and the laser scanning system are started simultaneously to obtain the image data and point cloud data of the corresponding scene at the same time; Based on the information obtained by the displacement sensor and the mileage encoder, dead reckoning is performed, and the dead reckoning and point cloud data are combined to generate a 3D point cloud scene. Based on the images acquired by the wide-area array camera, the images are panoramically stitched, and the stitched images are analyzed to identify defective cracks; The point cloud data and image information are mapped and registered to locate the defective cracks.

7. The working method according to claim 6, characterized in that: The specific process of dead reckoning includes: Get the coordinate information of pre-configured measuring points; Based on the coordinate information of the measuring point and the current position obtained by the displacement sensor and the mileage encoder, forward and reverse dead reckoning are performed to determine the initial aerial survey position; Combine the initial aerial survey positions with the 3D point cloud scene of the tunnel; The actual sensor data is used to correct the initial aerial survey position, and the corrected aerial survey position information is integrated with the 3D point cloud scene again to form the final true 3D point cloud scene.

8. The working method according to claim 6, characterized in that: The specific process of mapping and registering point cloud data and image information includes: Establish the spatial projection relationship between the laser scanning system and the area array camera, take multiple images and perform multiple scans in an environment with known geometric features, and use the acquired data to calculate the external and internal parameters; Using the spatial projection relationship obtained by calibration, the point cloud and image are converted from their respective three-dimensional coordinate systems to a unified two-dimensional plane. The point cloud is mapped to the two-dimensional plane through the orthographic projection algorithm to form a point cloud map; the image maintains its two-dimensional characteristics to form an image map; On a two-dimensional plane, the point cloud image is converted into a grayscale image, and the point cloud grayscale image and the image image are accurately registered using image processing technology; Based on the point cloud grayscale image and the registered image, subtle defects on the tunnel surface are identified.

9. The working method according to claim 6, characterized in that: The process of acquiring images with a wide-area array camera includes: Correct the focal length and viewing angle information of each camera lens to ensure that the positional relationship of each camera in space is consistent. By knowing the size and distance of the object, adjust the parameters of each camera lens so that the imaging results of each camera lens can correspond to each other. Use a distributed autofocus algorithm to unify the focus results of each camera; The imaging results of different camera lenses are combined to form a complete image based on the information weight of each pixel.

Images