Pipeline defect visual identification method for urban drainage intelligent operation and maintenance

Abstract

The invention provides a pipeline defect visual identification method for urban drainage intelligent operation and maintenance, and relates to the field of pipeline maintenance. The urban drainage intelligent operation and maintenance pipeline defect visual identification method is realized through an urban drainage intelligent operation and maintenance pipeline defect visual identification system, and the pipeline defect identification system comprises five function modules, namely an image acquisition module, an image preprocessing module, an image analysis module, a feature database module and an early warning maintenance module. The image acquisition module has a camera shooting acquisition function and an acquisition optimization function, and the image preprocessing module is used for carrying out optimization processing on an image acquired by the image acquisition device. A visual identification method is adopted to complete a detection task on the premise of guaranteeing safety, and the possibility of danger caused by contact is reduced.

Description

Technical Field

[0001] The present invention relates to the field of pipeline maintenance, and specifically to a pipeline defect visual recognition method for intelligent operation and maintenance of urban drainage. Background Art

[0002] In the modern industrial and urban infrastructure system, the urban drainage pipeline network undertakes the key task of transporting various types of water resources. Its safe operation is directly related to production continuity and the quality of public life. However, with the increase in the service life of the pipeline and the complexity of the environment, defects such as corrosion, cracks, and holes frequently appear. Traditional detection methods such as manual inspections and simple non-destructive testing technologies are not only inefficient, but also have many limitations in accuracy and comprehensiveness.

[0003] Therefore, the present invention proposes a visual recognition method for pipeline defects in intelligent operation and maintenance of urban drainage, which uses photography and intelligent detection to effectively solve the above-mentioned troubles and problems. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the shortcomings of the prior art, the present invention provides a visual recognition method for pipeline defects in urban drainage intelligent operation and maintenance, which can detect drainage pipeline defects to the greatest extent without damaging the internal and external environment of the drainage pipeline. The visual recognition method can complete the detection task under the premise of ensuring safety, reduce the possibility of danger caused by contact, and realize the automatic operation and remote monitoring of pipeline detection. Operators do not need to go to the site in person, and can command the pipeline robot to work deep into the narrow pipeline space in the control center, and receive the detection images and results in real time, which greatly reduces the workload of the detection and maintenance staff.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a pipeline defect visual recognition system for intelligent operation and maintenance of urban drainage, characterized in that: the system is controlled by a pipeline defect recognition system, the pipeline defect recognition system includes five functional modules: an image acquisition module, an image preprocessing module, an image analysis module, a feature database module and an early warning maintenance module, the image acquisition module includes a camera acquisition function and an acquisition optimization function, the image preprocessing module is used to optimize the image acquired by the image acquisition device, the image preprocessing module includes an image grayscale function, an image noise reduction function and a geometric correction function, the image analysis module includes a data extraction function and a feature classification and recognition function, the early warning maintenance module includes a summary of the data analyzed by the image analysis module, the generation and execution of the early warning plan and a maintenance scheduling function, and the feature database supports rapid recognition of image analysis feature classification by establishing a pipeline defect recognition database.

[0008] Preferably, the camera acquisition function of the image acquisition module is realized by an industrial camera, a high-definition camera and a three-dimensional sensor. The industrial camera is used for the detection requirements of different flow rates in the sewer pipe. The high-definition camera is used to capture the subtle texture and defect characteristics of the pipe surface. The three-dimensional sensor uses laser triangulation, structured light coding or stereo vision technology to obtain three-dimensional morphological data of the pipe surface. The acquisition optimization function ensures that the camera is perpendicular to the pipe surface or a specific optimal shooting environment by changing the lighting adjustment and adjusting the shooting angle.

[0009] Through the above technical solutions, industrial cameras, combined with light sources of specific wavelengths, can penetrate oil stains and clearly present subtle defects on the pipe wall. High-definition cameras can be equipped with wide-angle lenses to expand the range of a single shot and improve detection efficiency. Three-dimensional sensors can collect three-dimensional information on the surface of the pipeline, not only capturing appearance images but also obtaining depth data, which has obvious advantages in detecting structural defects such as deformation and dents.

[0010] Preferably, the image grayscale function of the image preprocessing module enhances the contrast of the image by remapping the grayscale value of the image, thereby highlighting the detail information in the image; the image noise reduction function uses a median filtering algorithm to remove noise from the image, improve the clarity and quality of the image, and avoid noise interference with subsequent analysis; the geometric correction function uses a Gaussian algorithm to correct image deformation caused by factors such as camera shooting angle and position, thereby ensuring the accuracy and consistency of the image;

[0011] Through the above technical solution, grayscale stretching is a basic image enhancement method, which belongs to linear point operation. Its core purpose is to improve the dynamic range of grayscale when processing images. Image denoising uses the median filtering algorithm to sort the grayscale values ​​in the pixel neighborhood and take the middle value as the pixel value after filtering. It has a significant effect on the removal of impulse noise such as shot noise and can better protect the edge. When processing noisy pipeline images, median filtering can maintain the sharpness of defect edges. After median filtering, the crack edge is clearly discernible.

[0012] Preferably, the data extraction of the image analysis module includes a traditional feature extraction function and a deep learning extraction function, the traditional feature extraction includes an edge detection function and a texture analysis function, the deep learning extraction function uses a convolutional neural network VGGNet to perform intelligent AI recognition after recognition training on a large amount of pipeline defect image data, and the classification feature recognition function includes a data set construction function, a deep learning function and a model optimization function;

[0013] Through the above technical solution, the edge detection function adopts the SobeL method edge detection;

[0014] SobeL edge detection method uses horizontal and vertical The convolution kernel takes partial derivatives of the image, calculates the gradient size and direction, and quickly detects edges. The local binary pattern (LBP) is used for texture analysis. The central pixel is used as the threshold, and the neighborhood pixels are binarized and combined into a binary number as the LBP value of the central pixel.

[0015] Preferably, the feature database module includes corrosion defect feature data, damage defect feature data and deformation defect feature data, the corrosion defect feature data includes appearance performance data, texture feature data and edge feature data, the damage defect feature data includes crack morphology data, hole morphology data and visual feature data, and the deformation defect feature data includes appearance change data, shape feature data and geometric parameter data;

[0016] Through the above technical solution, the surface of the corrosion defect presents irregular pits and mottled shapes, and the color is different from the surrounding normal pipe wall. The color of the corroded metal pipe becomes darker and rusty, and the corrosion of the plastic pipe may show signs of whitening and brittleness. The boundary between the corrosion area and the normal pipe wall is blurred. When the SobeL edge method is used for detection, the edge continuity is relatively poor, and the gradient amplitude changes relatively slowly. It is necessary to combine texture features for auxiliary judgment. The damage defects include crack damage and hole damage. The crack damage is slender, straight or winding, and narrow in width. It appears as a linear area with lower brightness on the image, which is in sharp contrast with the pipe wall background. The hole damage is mostly circular or elliptical, with relatively clear boundaries. Obvious depressions can be seen in the three-dimensional image, forming a depth difference with the pipe wall plane.

[0017] Preferably, the early warning solution generation function of the early warning maintenance module generates a pipeline defect maintenance solution through the data analysis results generated by the data center, and the early warning maintenance scheduling function shares the result data of pipeline defects with the maintenance department and the emergency department in real time through cross-departmental collaboration, thereby realizing information interconnection between departments and improving the efficiency of pipeline maintenance through multi-dimensional data information.

[0018] Working principle: The implementation steps of the pipeline defect visual recognition method for urban drainage intelligent operation and maintenance are as follows:

[0019] S1: Install industrial cameras, high-definition cameras, 3D sensors and searchlights on the pipeline detection machine, place the pipeline detection machine in the city drainage pipeline, and collect images inside the pipeline;

[0020] S2: After the device of the image acquisition module acquires the image data inside the pipeline, it transmits the image data to the image preprocessing module of the system, which quickly grayscales and reduces noise as well as performs geometric correction on the image;

[0021] S3: Pipeline defect model recognition and construction, through the pipeline defect model data in the feature database, using convolutional network neural deep learning to quickly extract image data and quickly analyze drainage pipeline defects:

[0022] S4: Obtain accurate pipeline defect detection results, share the detection results with relevant maintenance departments in real time through the early warning maintenance function, and initiate a rapid maintenance plan for emergency repairs.

[0023] (III) Beneficial effects

[0024] The present invention provides a pipeline defect visual recognition method for intelligent operation and maintenance of urban drainage, which has the following beneficial effects:

[0025] 1. The present invention provides a visual recognition method for pipeline defects in intelligent operation and maintenance of urban drainage. A pipeline robot is used to extend into the interior of the pipeline to take photos and videos to check for defects on the inner wall of the pipeline. There is no need to directly contact the pipeline surface, thus avoiding additional damage to the pipeline material. This is especially important for some old, fragile pipelines or pipelines with special media in the old urban areas of the city. It can detect drainage pipeline defects to the greatest extent without damaging the internal and external environment of the drainage pipeline. The visual recognition method can complete the detection task under the premise of ensuring safety, reducing the possibility of danger caused by contact.

[0026] 2. The present invention provides a visual recognition method for pipeline defects in urban drainage intelligent operation and maintenance. Compared with traditional manual inspections, it greatly shortens the detection time, can quickly obtain image information on the pipeline surface, and can be analyzed instantly through advanced image processing algorithms, greatly improving the detection efficiency, helping to timely discover potential problems and reduce the probability of failures, and can realize the automated operation and remote monitoring of pipeline detection. Operators do not need to be present on site, but can command pipeline robots to operate in narrow pipeline spaces in the control center, and receive detection images and results in real time, greatly reducing the workload of detection and maintenance staff.

[0027] 3. The present invention provides a visual identification method for pipeline defects in smart operation and maintenance of urban drainage. Visual camera detection is applicable to pipelines of various materials, whether they are metal pipelines such as water supply pipes made of steel and copper alloys, or drainage pipes made of plastic materials. The differences in optical properties on the surfaces of pipelines of different materials can be adapted by adjusting lighting conditions, camera parameters, etc., to ensure effective detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a general system diagram of visual identification of pipeline defects for intelligent operation and maintenance of urban drainage of the present invention;

[0029] Figure 2 A system diagram of a feature database module for visual identification of pipeline defects for intelligent urban drainage operation and maintenance of the present invention;

[0030] Figure 3 A system diagram of an image analysis module for visual identification of pipeline defects for intelligent urban drainage operation and maintenance of the present invention;

[0031] Figure 4 It is a schematic diagram of a pipeline defect feature hole damage image processed for pipeline defect visual recognition in the urban drainage intelligent operation and maintenance of the present invention;

[0032] Figure 5 It is a schematic diagram of pipeline defect feature corrosion image processing for pipeline defect visual recognition of urban drainage intelligent operation and maintenance of the present invention;

[0033] Figure 6 It is a schematic diagram of a pipeline defect feature deformed image processed for pipeline defect visual recognition in the urban drainage intelligent operation and maintenance of the present invention;

[0034] Figure 7 This is a schematic diagram of the pipeline defect feature crack damage image processed for the pipeline defect visual identification of the urban drainage intelligent operation and maintenance of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of this application, it should be noted that the terms used here are only for describing specific implementations, and are not intended to limit the exemplary implementations according to the present application. For ease of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship. The techniques, methods and devices known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the techniques, methods and devices should be regarded as part of the authorization specification. In all examples shown and discussed here, any specific value should be interpreted as merely exemplary, not as a limitation. Therefore, other examples of the exemplary embodiments may have different values. It should be noted that similar numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0036] Embodiment 1:

[0037] like Figure 1-7 As shown, an embodiment of the present invention provides a pipeline defect visual recognition system for intelligent operation and maintenance of urban drainage. The system is controlled by a pipeline defect recognition system. The pipeline defect recognition system includes five functional modules: an image acquisition module, an image preprocessing module, an image analysis module, a feature database module, and an early warning maintenance module. The image acquisition module includes a camera acquisition function and an acquisition optimization function. The image preprocessing module is used to optimize the image acquired by the image acquisition device. The image preprocessing module includes an image grayscale function, a picture noise reduction function, and a geometric correction function. The image analysis module includes a data extraction function and a feature classification and recognition function. The early warning maintenance module includes a function of summarizing the data analyzed by the image analysis module, generating and executing an early warning plan, and a maintenance scheduling function. The feature database supports rapid recognition of image analysis feature classification by establishing a pipeline defect recognition database.

[0038] The camera acquisition function of the image acquisition module is realized through industrial cameras, high-definition cameras and three-dimensional sensors. Industrial cameras are used to detect different flow rates in sewer pipes. High-definition cameras are used to capture subtle textures and defect features on the pipe surface. The three-dimensional sensor uses laser triangulation, structured light coding or stereo vision and other technologies to obtain three-dimensional morphology data on the pipe surface. The acquisition optimization function ensures that the camera and the pipe surface remain perpendicular or in a specific optimal shooting environment by changing the lighting adjustment and adjusting the shooting angle. The industrial camera, with a light source of a specific wavelength, can penetrate oil stains and clearly present subtle defects on the pipe wall. The high-definition camera can be equipped with a wide-angle lens to expand the single shooting range and improve detection efficiency. The three-dimensional sensor can collect three-dimensional information on the pipe surface, not only capturing the appearance image, but also obtaining depth data , it has obvious advantages in detecting structural defects such as deformation and dents. The image grayscale function of the image preprocessing module enhances the contrast of the image by remapping the grayscale value of the image, thereby highlighting the detailed information in the image. The image noise reduction function uses the median filter algorithm to remove noise in the image, improve the clarity and quality of the image, and avoid noise interference in subsequent analysis. The geometric correction function uses the Gaussian algorithm to correct image deformation caused by factors such as camera shooting angle and position to ensure the accuracy and consistency of the image. Grayscale stretching is a basic image enhancement method and belongs to linear point operation. Its core purpose is to improve the dynamic range of grayscale during image processing. Image noise reduction uses the median filter algorithm to sort the grayscale values ​​in the pixel neighborhood and take the middle value as the pixel value after filtering. It has a significant effect on the removal of impulse noise such as shot noise and can better protect the edge. When processing noisy pipeline images, the median filter can maintain the sharpness of the defect edge. After the median filter, the crack edge is clearly discernible;

[0039] The data extraction of the image analysis module includes traditional feature extraction and deep learning extraction functions. Traditional feature extraction includes edge detection and texture analysis. The deep learning extraction function uses the convolutional neural network VGGNet to perform intelligent AI recognition after training a large amount of pipeline defect image data. The classification feature recognition function includes data set construction, deep learning and model optimization. The edge detection function uses the SobeL method for edge detection.

[0040] SobeL edge detection method uses horizontal and vertical The convolution kernel calculates the partial derivative of the image, calculates the gradient size and direction, and quickly detects edges. The local binary pattern (LBP) is used for texture analysis. The central pixel is used as the threshold, and the neighborhood pixels are binarized and combined into a binary number as the LBP value of the central pixel. The feature database module contains corrosion defect feature data, damage defect feature data and deformation defect feature data. The corrosion defect feature data contains appearance performance data, texture feature data and edge feature data. The damage defect feature data contains crack morphology data, hole morphology data and visual feature data. The deformation defect feature data contains appearance change data, shape feature data and geometric parameter data. The corrosion defect surface presents different Regular pits and mottled shapes, with colors different from the surrounding normal pipe walls. Metal pipes become darker and rusty at the corroded areas. Plastic pipes may become white and brittle when corroded. The boundary between the corroded area and the normal pipe wall is blurred. When the SobeL edge method is used for detection, the edge continuity is relatively poor, and the gradient amplitude changes relatively slowly. Texture features need to be combined for auxiliary judgment. Damage defects include crack damage and hole damage. Crack damage is slender, straight or winding, and narrow in width. It appears as a linear area with lower brightness on the image, which forms a sharp contrast with the pipe wall background. Hole damage is mostly circular or elliptical, with relatively clear boundaries. Obvious depressions can be seen in the three-dimensional image, forming a depth difference with the pipe wall plane.

[0041] The early warning plan generation function of the early warning maintenance module generates a pipeline defect maintenance plan based on the data analysis results generated by the data center. The early warning maintenance scheduling function shares the result data of pipeline defects with the maintenance department and the emergency department in real time through cross-departmental collaboration, realizes information interconnection between departments, and improves the efficiency of pipeline maintenance through multi-dimensional data information.

[0042] Example 2: Figure 1-7 As shown, the implementation steps of the pipeline defect visual recognition method for urban drainage intelligent operation and maintenance provided by the embodiment of the present invention are:

[0043] S1: Install industrial cameras, high-definition cameras, 3D sensors and searchlights on the pipeline detection machine, place the pipeline detection machine in the city drainage pipeline, and collect images inside the pipeline;

[0044] S2: After the device of the image acquisition module acquires the image data inside the pipeline, it transmits the image data to the image preprocessing module of the system, which quickly grayscales and reduces noise as well as performs geometric correction on the image;

[0045] S3: Pipeline defect model recognition and construction, through the pipeline defect model data in the feature database, using convolutional network neural deep learning to quickly extract image data and quickly analyze drainage pipeline defects:

[0046] S4: Obtain accurate pipeline defect detection results, share the detection results with relevant maintenance departments in real time through the early warning maintenance function, and initiate a rapid maintenance plan for emergency repairs

[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The visual recognition system for pipeline defects in urban drainage intelligent operation and maintenance is characterized by: The system is controlled by a pipeline defect recognition system, which includes five functional modules: an image acquisition module, an image preprocessing module, an image analysis module, a feature database module and an early warning maintenance module. The image acquisition module includes a camera acquisition function and an acquisition optimization function. The image preprocessing module is used to optimize the image acquired by the image acquisition device. The image preprocessing module includes an image grayscale function, a picture noise reduction function and a geometric correction function. The image analysis module includes a data extraction function and a feature classification and recognition function. The early warning maintenance module includes a function of summarizing the data analyzed by the image analysis module, generating and executing an early warning plan, and a maintenance scheduling function. The feature database supports rapid recognition of image analysis feature classification by establishing a pipeline defect recognition database.

2. The pipeline defect visual recognition system for urban drainage intelligent operation and maintenance according to claim 1 is characterized by: The camera acquisition function of the image acquisition module is realized by an industrial camera, a high-definition camera and a three-dimensional sensor. The industrial camera is used to detect the different flow rates of the sewer pipe. The high-definition camera is used to capture the subtle texture and defect characteristics on the pipe surface. The three-dimensional sensor uses laser triangulation, structured light coding or stereo vision technology to obtain the three-dimensional morphology data of the pipe surface. The acquisition optimization function ensures that the camera remains perpendicular to the pipe surface or in a specific optimal shooting environment by changing the lighting adjustment and adjusting the shooting angle.

3. The pipeline defect visual recognition system for urban drainage intelligent operation and maintenance according to claim 1 is characterized by: The image grayscale function of the image preprocessing module enhances the contrast of the image by remapping the grayscale value of the image, thereby highlighting the detail information in the image. The image noise reduction function uses a median filtering algorithm to remove noise from the image, improve the clarity and quality of the image, and avoid noise interference with subsequent analysis. The geometric correction function uses a Gaussian algorithm to correct image deformation caused by factors such as the camera shooting angle and position, thereby ensuring the accuracy and consistency of the image.

4. The pipeline defect visual recognition system for urban drainage intelligent operation and maintenance according to claim 1 is characterized by: The data extraction of the image analysis module includes traditional feature extraction function and deep learning extraction function. The traditional feature extraction includes edge detection function and texture analysis function. The deep learning extraction function uses convolutional neural network VGGNet to perform intelligent AI recognition after recognition training on a large amount of pipeline defect image data. The classification feature recognition function includes data set construction function, deep learning function and model optimization function.

5. The pipeline defect visual recognition system for urban drainage intelligent operation and maintenance according to claim 1 is characterized by: The feature database module contains corrosion defect feature data, damage defect feature data and deformation defect feature data. The corrosion defect feature data contains appearance performance data, texture feature data and edge feature data. The damage defect feature data contains crack morphology data, hole morphology data and visual feature data. The deformation defect feature data contains appearance change data, shape feature data and geometric parameter data.

6. The pipeline defect visual recognition system for urban drainage intelligent operation and maintenance according to claim 1 is characterized by: The early warning scheme generation function of the early warning maintenance module generates a pipeline defect maintenance scheme through the data analysis results generated by the data center. The early warning maintenance scheduling function shares the result data of pipeline defects with the maintenance department and the emergency department in real time through cross-departmental collaboration, realizes information interconnection between departments, and improves the efficiency of pipeline maintenance through multi-dimensional data information.

7. A visual identification method for pipeline defects in urban drainage intelligent operation and maintenance, characterized by: The pipeline defect visual recognition method for intelligent operation and maintenance of urban drainage is implemented by a pipeline defect visual recognition system for intelligent operation and maintenance of urban drainage. The implementation steps of the pipeline defect visual recognition method for intelligent operation and maintenance of urban drainage are as follows: S1: Install industrial cameras, high-definition cameras, 3D sensors and searchlights on the pipeline detection machine, place the pipeline detection machine in the city drainage pipeline, and collect images inside the pipeline; S2: After the device of the image acquisition module acquires the image data inside the pipeline, it transmits the image data to the image preprocessing module of the system, which quickly grayscales and reduces noise as well as performs geometric correction on the image; S3: Pipeline defect model recognition and construction, through the pipeline defect model data in the feature database, using convolutional network neural deep learning to quickly extract image data and quickly analyze drainage pipeline defects: S4: Obtain accurate pipeline defect detection results, share the detection results with relevant maintenance departments in real time through the early warning maintenance function, and initiate a rapid maintenance plan for emergency repairs.

Images