Three-dimensional imaging reconstruction system and method based on X-ray nondestructive testing

Abstract

The invention discloses a three-dimensional imaging reconstruction system and method based on X-ray nondestructive testing. Relates to the technical field of power equipment detection. According to traditional electrical equipment detection, a three-dimensional model is constructed mainly according to a two-dimensional picture shot around a target area, complete sampling and shooting cannot be carried out in a scene where the installation position of the electrical equipment is limited and the space is narrow, and consequently, large errors exist in nondestructive detection; the X-ray digital imaging technology is adopted to carry out nondestructive testing on the electrical equipment, the historical database is utilized to carry out fusion reconstruction on the collected images and historical data, the image collection number is effectively reduced, calculation and storage burdens are reduced, the efficiency and precision of three-dimensional reconstruction are improved, and the method is particularly suitable for detection of the electrical equipment with limited space and has wide application prospects. The method has a good market application prospect, can significantly improve the detection efficiency of the power equipment, reduces the detection cost, and improves the accuracy of equipment fault recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment detection, and in particular to a three-dimensional imaging reconstruction system and method based on X-ray non-destructive testing. Background Art

[0002] In the process of ensuring the stable operation of the power system, it is crucial to conduct non-destructive testing of power equipment. Its purpose is to discover potential defects of the equipment in advance and prevent failures. In fact, due to the wide variety of power equipment and the complex installation environment, the traditional non-destructive testing method of taking a large number of two-dimensional photos around the target area to build a three-dimensional model is difficult to implement efficiently in many actual scenarios. For example, tension clamps are often installed at high altitudes and the surrounding space is narrow, with many obstacles. The internal structure of GIS equipment is compact and closed. When multi-directional X-ray digital imaging technology is used around these devices, it is inevitable that there will be problems such as limited shooting angles and incomplete image acquisition, which makes the traditional detection method have certain limitations. This limitation will seriously affect the overall performance of the system in situations such as narrow spaces and complex installation environments.

[0003] Nondestructive testing technology can cover a variety of scenarios and situations. Its main purpose is to quickly convey the health status of the equipment to the operator when testing power equipment to ensure the safe operation and power supply reliability of the power system. In the above-mentioned space-constrained testing scenarios, if traditional testing methods are still used, it will have a serious negative impact on the testing efficiency, thereby affecting the accuracy, timeliness and completeness of the test results, and may even affect the subsequent maintenance and repair of power equipment and the normal power supply of the power system. Therefore, it is very important to ensure that nondestructive testing technology can adapt to the complex and changing power equipment testing scenarios. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a three-dimensional imaging reconstruction system and method based on X-ray non-destructive testing. The technical problem to be solved is: traditional power equipment detection is mainly based on taking a large number of two-dimensional photos around the target area. Usually, hundreds of two-dimensional photos are required for each location to extract feature points to construct a three-dimensional model. When used in scenes such as tension clamps and GIS equipment with limited installation locations and narrow spaces, it is impossible to take photos around the target area, resulting in large errors in non-destructive testing. The present invention provides a three-dimensional imaging reconstruction system and method based on X-ray non-destructive testing to solve the above problems.

[0005] The purpose of the present invention is mainly achieved through the following technical solutions:

[0006] The three-dimensional imaging reconstruction system based on X-ray nondestructive testing includes: a historical database unit, an X-ray imaging unit, an image positioning unit, a three-dimensional reconstruction unit and a defect recognition unit; wherein:

[0007] The historical database unit includes a historical database storage module and a data calling module; the historical database storage module stores a basic information library of power equipment, a standard defect type library, a historical detection database and a defect database; the basic information library of power equipment stores basic information of power equipment including a three-dimensional model, model, installation location, service life and historical fault records of power equipment; the standard defect type library stores standardized three-dimensional defect type models including cracks, relaxation and corrosion.

[0008] The historical detection database stores two-dimensional images of the detected area collected by the X-ray imaging unit and basic information of the detected power equipment; the defect database stores three-dimensional defect images of the detected area identified as having defects by the defect recognition unit.

[0009] The data calling module quickly reads the contents written into the distributed storage architecture by the historical database storage module; when the X-ray imaging unit collects images, the basic information of the inspected power equipment is called according to the inspected area; when the defect recognition unit recognizes that the inspected area has defects, the data calling module retrieves the standardized three-dimensional defect type model.

[0010] The X-ray imaging unit includes an image acquisition module and an image preprocessing module; the image acquisition module acquires images of the detected power equipment, sets N angles for shooting according to geometric symmetry characteristics, and outputs a two-dimensional slice image, wherein N is less than or equal to 10; and three angles are perpendicular to each other; the image preprocessing module performs image denoising and image enhancement on the images acquired by the image acquisition module in sequence, and outputs a two-dimensional X-ray image to the image positioning unit; the image preprocessing module performs image denoising on the two-dimensional slice image by using a median filter to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise, improves the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image, enhances the contrast, and outputs a two-dimensional X-ray image.

[0011] The image positioning unit includes: an X-ray image feature extraction module, a historical data feature extraction module and a feature point matching module; the X-ray image feature extraction module extracts geometric features and texture features from the X-ray image to obtain shape information of the equipment components of the X-ray image, and then extracts feature points of the X-ray image, and uses the feature points as the basis for image positioning; the historical data feature extraction module slices the three-dimensional model of the power equipment at equal intervals along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment; the feature point matching module calculates the similarity value between the X-ray image and the two-dimensional image of the power equipment, sets a threshold, and those that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment, and the image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data.

[0012] The three-dimensional reconstruction unit uses an image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model based on the X-ray image collected from the X-ray imaging unit and the three-dimensional model of the power equipment obtained from the historical database unit, and uses the three-dimensional structure provided by the three-dimensional model of the power equipment as supplementary information.

[0013] The defect recognition unit uses the features extracted by the X-ray image feature extraction module and the historical data feature extraction module based on the three-dimensional reconstruction model and the standardized three-dimensional defect type model in the historical database, and uses an image matching algorithm to match the defect features of the three-dimensional reconstruction model with the features of the standardized three-dimensional defect type model to determine the defect type.

[0014] How this solution works:

[0015] Specifically for the detection scenarios of complex installation locations and limited space of power equipment, a systematic improvement is made on the basis of the existing non-destructive testing technology. Based on the current reality that the types of defects in power equipment are relatively simple, conventional defects are cracks, loose connectors or fasteners, and a large number of defective parts generally do not appear in a single inspection, and the influence of space limitations around the inspected object is considered in the process of building a three-dimensional model. The historical data is comprehensively considered in the three-dimensional reconstruction process to reduce the problem of low detection efficiency caused by factors such as shooting difficulties and large data volumes. From the perspective of facilitating the use of X-ray digital imaging technology to collect images, the traditional practice of taking a large number of photos around is abandoned, and only a preset number of X-ray images at key angles are taken. The X-ray images are positioned through feature point matching and geometric and texture feature matching, and other supplementary data required for three-dimensional construction are retrieved and used from the pre-built historical database. The three-dimensional model is reconstructed through data fusion for defect identification, providing an efficient non-destructive testing system suitable for the detection of various complex power equipment.

[0016] In order to further optimize the above technical solution, the image acquisition module includes an X-ray generator and a detector and a mechanical motion device.

[0017] It is equipped with an X-ray transmitter and a receiver to take X-ray images of the inspected area of ​​the inspected power equipment; the X-ray generator and detector are installed on a mechanical motion device, which adopts a high-precision mechanical motion device to drive the X-ray generator and detector to move according to the position of the inspected area. When they reach the position of the inspected area, the mechanical motion device drives the X-ray generator and detector to move to the set angle to start image acquisition. The X-ray generator emits high-energy electromagnetic wave X-rays, and the X-rays pass through the inspected area. Because there is no X-ray detector, it receives the X-ray penetration signal, records the X-ray intensity passing through the object at each angle and converts it into a digital image signal, and generates a two-dimensional slice image according to the X-ray attenuation coefficient of each point inside the object.

[0018] In order to further optimize the above technical solution, the image preprocessing module performs image denoising on the two-dimensional slice image by using median filtering to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise to improve the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image, enhance the contrast and output a two-dimensional X-ray image.

[0019] In order to further optimize the above technical solution, the X-ray image feature extraction module includes a geometric feature extraction submodule, a texture feature extraction submodule and a feature point extraction submodule.

[0020] The geometric feature extraction submodule processes the collected X-ray images using an edge detection algorithm, finds the maximum points of the intensity gradient in the X-ray images, connects the maximum points into edge lines, and thus extracts the internal contour of the detected device.

[0021] The texture feature extraction submodule uses a gray level co-occurrence matrix to analyze image texture and calculates the joint distribution of pixel gray levels at different directions and distances in the X-ray image to supplement the geometric features of the X-ray image.

[0022] The feature point extraction submodule uses a scale-invariant feature transformation algorithm to extract feature points with scale and rotation invariance from the X-ray image, and calculates extreme points in different scale spaces to represent image details as the basis for image positioning.

[0023] In order to further optimize the above technical solution, the historical data feature extraction module includes two-dimensional cross-sectional image generation geometric feature extraction.

[0024] The two-dimensional cross-sectional image generation and geometric feature extraction slices the three-dimensional model of the power equipment in the historical data according to the set direction and equal spacing, and converts the three-dimensional model into a series of two-dimensional cross-sectional images.

[0025] The two-dimensional cross-sectional image uses an edge detection algorithm to extract the geometric features corresponding to the X-ray image and restore the shape contours of the three-dimensional model in different cross-sections.

[0026] In order to further optimize the above technical solution, the feature point matching module includes a feature point-based initial matching submodule, an error matching elimination submodule and a space transformation relationship calculation submodule.

[0027] The initial matching based on feature points adopts the nearest neighbor matching algorithm to match the feature points extracted from the X-ray image with the two-dimensional cross-sectional view of the historical data or the voxelized feature points; the Euclidean distance of the nearest neighbor matching algorithm is used for measurement, and the Euclidean distance between the feature points is used as the similarity measurement to quickly search the historical data for the point closest to the feature point of the X-ray image as the matching point; the error matching elimination adopts the random sampling consistency algorithm to screen the initial matching results, and removes the error matching points through multiple random samplings.

[0028] The spatial transformation relationship calculation is based on the selected feature points, and the least square method is used to solve the linear equations to calculate the transformation matrix, which is used as the basis for the spatial transformation between the X-ray image and the historical data three-dimensional model to perform translation, rotation and scaling transformations.

[0029] In order to further optimize the above technical solution, the three-dimensional reconstruction unit image reconstruction algorithm adopts an iterative reconstruction algorithm to obtain a three-dimensional structure of the two-dimensional X-ray image through multiple iterations.

[0030] In order to further optimize the above technical scheme, the defect recognition unit extracts shape features and spatial positions from the three-dimensional reconstructed model and the standardized three-dimensional defect type model in the historical database; uses an image matching algorithm to match the defect features of the three-dimensional reconstructed model with the features of the standardized three-dimensional defect type model; the shape features use an edge detection algorithm to obtain the defect boundary; the spatial position uses the position of the defect in the three-dimensional coordinate system; the difference between the shape features and the spatial position is calculated using the Euclidean distance, and the similarity of the Euclidean distance is obtained as the same defect type within a set threshold; the shape features and spatial positions of the three-dimensional reconstructed model are used as defect parameters, the defect parameters include geometric parameters, position and direction, the defect parameters are combined with the operating parameters and material performance parameters of the power equipment, the operating parameters include load and pressure, and the material performance parameters include strength and toughness, and the severity is evaluated using a physical model.

[0031] The present invention also includes a three-dimensional imaging reconstruction method based on X-ray non-destructive testing, which is implemented based on the system and includes the following steps:

[0032] Step S1, establish a historical database to facilitate storage and calling of the historical database; the historical database storage includes a basic information library of power equipment, a standard defect type library, a historical detection database and a defect database; the basic information library of power equipment stores basic information of power equipment including a three-dimensional model, model, installation location, service life and historical fault records of power equipment; the standard defect type library stores standardized three-dimensional defect type models including cracks, relaxation and corrosion.

[0033] Step S2, using X-ray imaging equipment to perform X-ray imaging, image acquisition and image preprocessing; image acquisition performs image acquisition on the detected power equipment, and the image acquisition is shot according to the set N angles to output a two-dimensional slice image, wherein N is less than or equal to 10; among which 3 angles are perpendicular to each other; image preprocessing performs image denoising and image enhancement on the image acquisition image in sequence, and outputs a two-dimensional X-ray image to be transmitted to the image positioning unit; the image preprocessing performs image denoising on the two-dimensional slice image by using median filtering to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise, improves the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image, enhances the contrast and outputs a two-dimensional X-ray image.

[0034] Step S3, positioning the X-ray image on a specific device. Image positioning includes an X-ray image feature extraction module that extracts geometric features and texture features from the X-ray image to obtain shape information of device components in the X-ray image, and then extracts feature points from the X-ray image, using the feature points as the basis for image positioning.

[0035] The image positioning also includes a historical data feature extraction module which slices the three-dimensional model of the power equipment at equal intervals along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment.

[0036] Finally, feature point matching is performed, and the similarity value between the X-ray image and the two-dimensional image of the power equipment is calculated. A threshold is set, and those that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment. The image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data.

[0037] Step S4, perform three-dimensional reconstruction; based on the X-ray image collected from the X-ray imaging and the three-dimensional model of the power equipment obtained from the historical database, use the image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model, and use the three-dimensional structure provided by the three-dimensional model of the power equipment as supplementary information.

[0038] Step S5, performing defect identification; according to the features extracted from the 3D reconstructed model and the standardized 3D defect type model in the historical database, an image matching algorithm is used to match the defect features of the 3D reconstructed model with the features of the standardized 3D defect type model to determine the defect type.

[0039] In summary, the present invention has the following beneficial effects compared with the prior art:

[0040] 1. Significantly reduce the amount of shooting. Because the traditional cumbersome method of taking hundreds of photos around is abandoned, only a few X-ray photos of key angles are taken, which not only greatly shortens the on-site shooting time, reduces the radiation dose of X-rays to equipment and operators, but also reduces the risk of equipment failure caused by taking too many photos.

[0041] 2. Effectively reduce system storage and computing power requirements. Because a small amount of image acquisition reduces data storage pressure, and in the 3D reconstruction process, historical data is used to share part of the computing workload, so the computing power requirements for the data processing unit are also reduced accordingly, reducing the hardware procurement and maintenance costs of the system.

[0042] 3. Greatly accelerate the efficiency and accuracy of 3D reconstruction. Because the reconstruction method based on the fusion of collected images and historical data avoids the complex feature extraction and model building process of a large number of images from scratch, with the support of historical data, it can generate accurate 3D models in a short time, providing strong support for timely discovery of potential defects of power equipment and ensuring the operation of power systems.

[0043] 4. Improve defect detection accuracy. Because it is combined with the defect library in historical data, it can identify and locate common defects on the equipment, which directly increases the reliability and accuracy of detection.

[0044] The present invention adopts X-ray digital imaging technology to perform non-destructive testing on power equipment, and utilizes the historical database to reconstruct the fusion of collected images and historical data, which effectively reduces the number of image collections, reduces the computing and storage burden, and improves the efficiency and accuracy of three-dimensional reconstruction. It is particularly suitable for the detection of power equipment with limited space, has good market application prospects, can significantly improve the detection efficiency of power equipment, reduce detection costs, and improve the accuracy of equipment fault identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0046] Figure 1 This is a structural diagram of a three-dimensional imaging reconstruction system based on X-ray nondestructive testing;

[0047] Figure 2 A step diagram of a three-dimensional imaging reconstruction method based on X-ray non-destructive testing. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0049] Traditional power equipment inspection is mainly based on constructing a three-dimensional model by taking two-dimensional photos around the target area. When applied in scenarios where the power equipment installation location is limited and the space is narrow, it is impossible to perform complete sampling and shooting, resulting in large errors in non-destructive testing. In view of this, the present invention provides the following embodiments to solve the above technical problems.

[0050] In the example:

[0051] like Figure 1 As shown, this embodiment proposes a three-dimensional imaging reconstruction system based on X-ray non-destructive testing, which is composed of a historical database unit, an X-ray imaging unit, an image positioning unit, a three-dimensional reconstruction unit and a defect recognition unit.

[0052] like Figure 2 As shown, this embodiment also proposes a three-dimensional imaging reconstruction method based on X-ray non-destructive testing, and the specific steps include:

[0053] Step S1, establish a historical database to facilitate storage and call of the historical database; the historical database storage includes a basic information database of power equipment, a standard defect type database, a historical detection database and a defect database; the basic information database of power equipment stores basic information of power equipment including a three-dimensional model, model, installation location, service life and historical fault records of power equipment; the standard defect type database stores standardized three-dimensional defect type models including cracks, relaxation and corrosion;

[0054] Step S2, using an X-ray imaging device to perform X-ray imaging, image acquisition and image preprocessing; image acquisition performs image acquisition on the detected power equipment, and the image acquisition is shot according to the set N angles, and a two-dimensional slice image is output, wherein N is less than or equal to 10; and 3 angles are perpendicular to each other; image preprocessing performs image denoising and image enhancement on the image acquisition image in sequence, and outputs a two-dimensional X-ray image and transmits it to the image positioning unit; the image preprocessing performs image denoising on the two-dimensional slice image, uses a median filter to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise, improves the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image, enhances the contrast, and outputs a two-dimensional X-ray image;

[0055] Step S3, positioning the X-ray image on a specific device. The image positioning includes an X-ray image feature extraction module extracting geometric features and texture features from the X-ray image to obtain shape information of the device components of the X-ray image, and then extracting feature points from the X-ray image, and using the feature points as the basis for image positioning;

[0056] Image positioning also includes a historical data feature extraction module that slices the three-dimensional model of the power equipment at equal distances along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment;

[0057] Finally, feature point matching is performed, and the similarity value between the X-ray image and the two-dimensional image of the power equipment is calculated, and a threshold is set. The points that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment. The image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data;

[0058] Step S4, performing three-dimensional reconstruction; using an image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model based on the X-ray image collected from the X-ray imaging and the three-dimensional model of the power equipment obtained from the historical database, and using the three-dimensional structure provided by the three-dimensional model of the power equipment as supplementary information;

[0059] Step S5, performing defect identification; according to the features extracted from the 3D reconstructed model and the standardized 3D defect type model in the historical database, an image matching algorithm is used to match the defect features of the 3D reconstructed model with the features of the standardized 3D defect type model to determine the defect type.

[0060] The principle of this embodiment is as follows: an X-ray imaging unit is set up, and several two-dimensional X-ray digital imaging technologies of the target device are collected by the X-ray imaging device. The number of these images is small, usually 4-8, and the shooting angle selection fully considers the space limitation of the target device, and gives priority to the angle that can obtain key information to reduce the requirements for the environment and space; the historical database unit can call the three-dimensional model data and defect information in the historical database according to the type of power equipment, historical detection records and previous defect distribution; the image positioning unit determines the area of ​​the equipment to be inspected by analyzing the historical data, which is convenient for using the existing historical data; the three-dimensional reconstruction module uses the collected two-dimensional X-ray digital imaging technology and the relevant three-dimensional data in the historical database, and adopts the image reconstruction algorithm to perform three-dimensional reconstruction. 3D reconstruction: This module can quickly generate an accurate 3D model based on the existing 2D images. At the same time, it can supplement the missing 3D structure data based on the support of historical data and this information to improve the accuracy and efficiency of reconstruction. The defect recognition unit performs defect detection based on the generated 3D model and defect library. By comparing the defects with the historical data, it analyzes the common defects such as cracks, looseness, wear and tear that may exist on the detection equipment, and can provide accurate detection results in real time. The user interaction and result display module provides a friendly user interface. Users can view the 3D model and defect detection results of the equipment in the system for further analysis and decision-making. Users can choose to view detailed information of a specific location through the interactive interface, or perform subsequent equipment maintenance and repair according to the system's suggestions.

[0061] The historical database unit includes a historical database storage module and a data calling module.

[0062] The historical database storage module adopts a distributed storage architecture to store the basic information library of power equipment, the standard defect type library, the historical detection database and the defect database; the distributed storage architecture is adopted, and it has a massive storage capacity, and can safely and efficiently store various types of X-ray detection data of power equipment accumulated over the years. The distributed storage architecture consists of three parts: a client, a metadata server and a data server. The client is responsible for sending read and write requests, caching file metadata and file data. In the present invention, this system acts as a client to read and write historical database units on the distributed storage architecture. The metadata server is responsible for managing metadata and processing client requests, and is the core component of the entire system. The data server is responsible for storing file data to ensure data availability and integrity. The advantage of this architecture is that performance and capacity can be expanded at the same time, and the system scale is highly scalable. At the same time, relevant data can be accurately retrieved according to demand in a short time.

[0063] The basic information library of power equipment stores basic information of power equipment including three-dimensional model, model, installation location, service life and historical fault records of power equipment; the standard defect type library stores standardized three-dimensional defect type models of cracks, relaxation and corrosion. The standardized three-dimensional defect type model is constructed by high-precision scanning and data collection of a large number of power equipment samples containing typical defects of cracks, relaxation and corrosion, and combining geometric features and physical properties analysis to construct a three-dimensional model.

[0064] The historical detection database stores two-dimensional images of the detected area collected by the X-ray imaging unit and basic information of the detected power equipment; the defect database stores three-dimensional defect images of the detected area identified as having defects by the defect recognition unit.

[0065] The data calling module quickly reads the contents written into the distributed storage architecture by the historical database storage module; when the X-ray imaging unit collects images, the basic information of the inspected power equipment is called according to the inspected area; when the defect recognition unit recognizes that the inspected area has defects, the data calling module retrieves the standardized three-dimensional defect type model.

[0066] The X-ray imaging unit includes an image acquisition module and an image preprocessing module; the image acquisition module acquires images of the detected power equipment, sets N angles for shooting according to the equipment position and space restrictions and geometric symmetry characteristics, and outputs a two-dimensional slice image, where N is less than or equal to 10; three of the angles are perpendicular to each other, and usually, the acquired images are taken from different angles such as the front, side, and top of the equipment; the image preprocessing module performs image denoising and image enhancement on the images acquired by the image acquisition module in turn, and outputs a two-dimensional X-ray image to be transmitted to the image positioning unit.

[0067] The image acquisition module includes an X-ray generator, a detector and a mechanical motion device.

[0068] An X-ray transmitter and a receiver are configured to take X-ray images of the inspected area of ​​the inspected power equipment; in this embodiment, 4-8 two-dimensional X-ray images at different angles are collected to ensure that key geometric information can be captured.

[0069] The X-ray generator and detector are installed on a mechanical motion device. The mechanical motion device adopts a high-precision mechanical motion device. It drives the X-ray generator and the detector to move according to the position of the detected area. When the X-ray generator and the detector reach the position of the detected area, the mechanical motion device drives the X-ray generator and the detector to move to the set angle to start image acquisition. The X-ray generator emits high-energy electromagnetic wave X-rays. The X-rays pass through the detected area. Because different tissues and materials absorb X-rays to different degrees, different degrees of attenuation are produced. These attenuation information will be recorded by the detector. The X-ray detector receives the X-ray penetration signal, records the X-ray intensity passing through the object at each angle and converts it into a digital image signal, and generates a two-dimensional slice image according to the X-ray attenuation coefficient of each point inside the object.

[0070] Equipped with an X-ray generator that can flexibly adjust angles and parameters, the energy can be accurately adjusted according to the material and thickness of the power equipment. In this embodiment, because X-ray imaging is performed on metal power transmission equipment, the X-ray generator is set to 150-300kV, and the exposure time is set to between 5 seconds and 15 seconds, and adjusted according to the ambient light and detector sensitivity.

[0071] The X-ray generator is matched with a corresponding high-resolution, low-noise detector. The X-ray detector receives the X-ray penetration signal, records the X-ray intensity passing through the object at each angle, and converts it into a digital image signal. According to the X-ray attenuation coefficient of each point inside the object, the reconstruction algorithm of the X-ray generator and detector is used to form a series of two-dimensional slice images. These slice images can be further combined into three-dimensional images to show the internal details of the scanned object in the form of grayscale images. Different tissues and materials show different grayscale values. The metal part will appear brighter white, while pores and cracks will appear darker gray. It can detect the internal quality of parts and components, discover hidden defects including weld cracks and pores, and improve product reliability and safety.

[0072] The image preprocessing module performs image denoising on the two-dimensional slice image by using median filtering to remove salt and pepper noise, then uses wavelet denoising to reduce Gaussian noise and improve the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image and output a two-dimensional X-ray image by enhancing the contrast.

[0073] Due to the complex internal structure of the equipment, X-ray images may contain a lot of noise. Through preprocessing of denoising and then enhancement, the image quality can be improved and clearer input can be provided for subsequent three-dimensional imaging.

[0074] First, use median filtering to remove salt and pepper noise. The operating principle of median filtering is to replace the grayscale value of each pixel in the image with the median of the grayscale values ​​of its neighboring pixels. For salt and pepper noise, which is a noise point that obviously deviates from the grayscale value of the surrounding pixels, median filtering can accurately identify and correct it, making its grayscale value more coordinated with the surrounding pixels, thereby effectively removing these noise points and making the image initially smooth.

[0075] Next, wavelet denoising is used to reduce Gaussian noise. Wavelet denoising is based on wavelet transform theory. It first performs multi-scale wavelet decomposition on the image, breaking the image down according to different frequency components, so that the high-frequency wavelet coefficients corresponding to the Gaussian noise can be separated from the low-frequency useful information of the image itself. Subsequently, only the low-frequency part reflecting the image details and main structure is retained, and finally wavelet reconstruction is performed to restore the processed wavelet coefficients to the denoised image, greatly reducing Gaussian noise interference and significantly improving the signal-to-noise ratio of the image.

[0076] Finally, the image enhancement is performed using the contrast-limited adaptive histogram equalization method, which can enhance the contrast of the image while suppressing noise and preventing excessive enhancement from causing detail loss. It is suitable for devices with complex structures.

[0077] The image positioning unit includes: an X-ray image feature extraction module, a historical data feature extraction module and a feature point matching module; the X-ray image feature extraction module extracts geometric features and texture features from the X-ray image to obtain shape information of the equipment components of the X-ray image, and then extracts feature points of the X-ray image, and uses the feature points as the basis for image positioning; the historical data feature extraction module slices the three-dimensional model of the power equipment at equal intervals along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment; the feature point matching module calculates the similarity value between the X-ray image and the two-dimensional image of the power equipment, sets a threshold, and those that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment, and the image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data.

[0078] The X-ray image feature extraction module includes a geometric feature extraction submodule, a texture feature extraction submodule and a feature point extraction submodule.

[0079] The geometric feature extraction submodule uses edge detection algorithms to process the collected X-ray images, finds the maximum value points of the intensity gradient in the X-ray images, and connects the maximum value points into edge lines, so as to accurately extract the internal contours of the detected equipment, such as the outer boundary of the tension clamp, the edges of the internal key structures, and the boundaries of the connection parts. These clear geometric features outline the main shape information in the image, providing a basic framework for subsequent matching with historical data.

[0080] The texture feature extraction submodule uses a gray level co-occurrence matrix to analyze image texture and calculates the joint distribution of pixel gray levels at different directions and distances in the X-ray image to supplement the geometric features of the X-ray image.

[0081] For X-ray images of power equipment, the texture trend presented by the material distribution inside the equipment can be obtained, including the texture uniformity of metal parts and the density changes. These texture features are used to supplement geometric features, enhance image recognition, and help distinguish equipment parts of different materials and different states.

[0082] The feature point extraction submodule uses a scale-invariant feature transformation algorithm to extract feature points with scale and rotation invariance from the X-ray image, calculates extreme points in different scale spaces, and assigns directional information to them, so that the extracted feature points are rich in information about the local area of ​​the image, the features of tiny components on the surface of the equipment, and the local structural changes. They can characterize image details at different scales and angles, and serve as the basis for image positioning. Even if the image has certain rotation and scaling changes, it can still be reliably used for subsequent matching.

[0083] The historical data feature extraction module includes two-dimensional cross-sectional image generation and geometric feature extraction.

[0084] The two-dimensional cross-sectional image generation and geometric feature extraction slice the three-dimensional model of the power equipment in the historical data in a set direction. In this embodiment, the three-dimensional model is converted into a series of two-dimensional cross-sectional images so that its data form can be compared with the X-ray image in the same dimension.

[0085] The same edge detection algorithm is used for these two-dimensional cross-sectional images to extract the geometric features corresponding to the X-ray images, restore the shape contours of the three-dimensional model in different cross-sections, ensure the geometric consistency of key structures, and match the device contours and key component boundaries in the X-ray images.

[0086] The feature point matching module includes a feature point-based initial matching submodule, an error matching elimination submodule, and a space transformation relationship calculation submodule.

[0087] The initial matching based on feature points adopts the nearest neighbor matching algorithm to match the feature points extracted from the X-ray image with the two-dimensional cross-sectional view of the historical data or the voxelized feature points; the Euclidean distance of the nearest neighbor matching algorithm is used for measurement, and the Euclidean distance between the feature points is used as the similarity measurement to quickly search the historical data for the point closest to the feature point of the X-ray image as the matching point, so as to preliminarily establish the connection between the two.

[0088] The error matching elimination adopts a random sampling consistency algorithm to screen the initial matching results. Through multiple random samplings, a geometric transformation model is assumed. This embodiment adopts perspective transformation and uses the number of internal points to evaluate the rationality of the model. The internal points are points that meet the assumed geometric transformation model. After multiple rounds of iterations, feature points that do not meet the assumed model are removed, and the error matching points are removed. The internal points that meet the geometric transformation model are retained to ensure the accuracy and reliability of the matching and avoid interference of the error matching points on the subsequent positioning.

[0089] The spatial transformation relationship calculation uses the least squares method to solve the linear equations based on the screened feature points to calculate an accurate transformation matrix, which is used as the basis for the spatial transformation between the X-ray image and the three-dimensional model of historical data. Translation, rotation and scaling transformations are performed. These parameters are the key information for achieving image positioning. The X-ray image is aligned with the historical model in three-dimensional space.

[0090] The image positioning unit implements corresponding geometric transformations on the collected X-ray images based on these parameters. First, the X-ray image is moved to the corresponding starting position in the three-dimensional coordinate system according to the translation parameters; then, the image direction is adjusted around a specific coordinate axis according to the rotation angle parameters to make it consistent with the orientation of the historical model; finally, the image size is adapted according to the scaling factor to align it with the three-dimensional model of the power equipment in the historical data as the basis for subsequent three-dimensional reconstruction.

[0091] The three-dimensional reconstruction unit uses an image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model based on the X-ray image collected from the X-ray imaging unit and the three-dimensional model of the power equipment obtained from the historical database unit. The three-dimensional model of the power equipment provides supplementary information of the three-dimensional structure, helps to fill in possible missing details in the reconstruction process, and makes up for angles or details that cannot be covered during the image acquisition process, thereby improving the accuracy and completeness of the reconstruction.

[0092] The three-dimensional reconstruction unit image reconstruction algorithm adopts an iterative reconstruction algorithm to obtain a three-dimensional structure of a two-dimensional X-ray image through multiple iterations in the case of incomplete projection data. The iterative reconstruction algorithm is trained using a three-dimensional model of the power equipment to correct errors in the iterative reconstruction process, thereby improving the accuracy and resolution of the three-dimensional reconstruction.

[0093] The defect recognition unit uses the features extracted by the X-ray image feature extraction module and the historical data feature extraction module according to the three-dimensional reconstruction model and the standardized three-dimensional defect type model in the historical database; and uses an image matching algorithm to match the defect features of the three-dimensional reconstruction model with the features of the standardized three-dimensional defect type model.

[0094] The shape feature uses an edge detection algorithm to obtain the defect boundary; the spatial position adopts the position of the defect in the three-dimensional coordinate system.

[0095] The difference between the shape feature and the spatial position is calculated using the Euclidean distance, and the similarity of the obtained Euclidean distance is set at a threshold as the same defect type.

[0096] Based on the geometric parameters, location and direction of the defects, combined with the operating parameters of the power equipment including load, pressure and material performance parameters including strength and toughness, the severity assessment is carried out using a physical model.

[0097] For crack defects, the impact of cracks on equipment strength is evaluated by the stress intensity factor calculation formula based on the crack length, depth and fracture toughness of the equipment material, and a quantitative severity level is given.

[0098] All modules transfer data through message queues to ensure the modularity and scalability of the system. Data storage uses a distributed database system to store a large amount of historical data and 3D reconstruction results in an efficient file storage system.

[0099] In this embodiment, the three-dimensional imaging reconstruction system based on X-ray non-destructive testing also includes a user interaction and result display unit to facilitate human-computer interaction in the system.

[0100] The user interaction and result display unit combines the defect three-dimensional model with the defect marking result based on the detection results and defect three-dimensional model from the defect identification unit to present an intuitive three-dimensional view. The user interaction and result display unit is electrically connected to the historical database unit, and can consult the data of the basic information base of the power equipment and the historical detection database of the detected equipment. The user can view the three-dimensional model and defect detection results of the equipment in the system, rotate and enlarge the defect three-dimensional model, and can deeply view the detailed information of a specific location, such as the size of the crack and the loose location, for reference by maintenance and inspection personnel, which helps to improve the accuracy of fault detection and support subsequent equipment maintenance and decision-making.

[0101] The technology of combining volume rendering and surface rendering is adopted. Volume rendering can show the overall structure and density information inside the power equipment, so that the operator can intuitively understand the internal condition of the equipment; surface rendering focuses on clearly showing the surface contour and defect details of the equipment, which is convenient for accurately locating the specific fault part. For example, for the three-dimensional imaging of the internal winding of the transformer, the overall structure and insulation condition of the winding can be displayed by volume rendering first, and then the specific defect parts between turns that may exist can be highlighted by surface rendering, so that the operator can conduct a more comprehensive analysis.

[0102] Operators can use devices such as mouse and keyboard to rotate, scale, and slice the 3D model and observe the equipment from different angles to fully analyze the health status of the equipment.

[0103] In addition, a detailed inspection report is generated based on the defect identification results of the defect identification unit. The inspection report includes basic information of the inspected equipment, screenshots of key views of the 3D model, defects found, and corresponding treatment suggestions, etc., providing a strong basis for the maintenance and overhaul of power equipment.

[0104] In this embodiment, the three-dimensional imaging reconstruction method based on X-ray non-destructive testing also includes step S6, realizing human-computer interaction, realizing system-user interaction and result display.

[0105] Taking the inspection of a certain type of GIS equipment as an example, first, the operator enters the equipment inspection interface through the user interaction and result display unit, and enters the model, installation location and other information of the GIS equipment. The system automatically determines four key shooting angles based on this information and pre-stored inspection experience of similar equipment. Three of the directions are the front angle perpendicular to the busbar connection, the side angle showing the outline of the tank, and the angle looking down from the top.

[0106] Next, the operator starts the X-ray imaging unit and takes X-ray photos in sequence according to the preset angles, energy exposure time and distance. During the shooting process, the operator records the parameters such as the energy setting of the X-ray source as 200 kV, the exposure time as 5 seconds, the distance between the detector and the equipment as 50 cm, and various shooting angles. The three collected two-dimensional X-ray images are transmitted to the image positioning unit in real time and stored in the historical database unit.

[0107] At the same time, the image positioning unit sends a search request to the historical database unit according to the model and other information of the GIS device. The historical database unit quickly retrieves the past detection data of the same model of GIS equipment and transmits these data back to the image positioning unit.

[0108] Then, the image positioning unit uses the feature extraction method to match the four currently collected images with the relevant images in the historical data, and then constructs a preliminary three-dimensional model, which is continuously refined through iterations to finally construct a three-dimensional model of the GIS equipment.

[0109] Finally, the user interaction and result display unit visualizes the constructed 3D model. By combining volume rendering with surface rendering, operators can clearly see the busbar connection inside the GIS equipment, the internal structure of the tank, and whether there are defects such as cracks and loose connectors.

[0110] Combined with the defect recognition unit, the system uses image matching algorithm to obtain the defect category and grade it. The user interaction and result display unit automatically generates a detailed inspection report based on the inspection results combined with historical data, providing an accurate basis for subsequent maintenance and repair work.

[0111] Through the coordinated work of the above units, this embodiment can effectively solve the problems of traditional detection methods in terms of limited space and heavy computational burden. It not only improves the detection efficiency and effectively achieves the goals of reducing the amount of shooting, lowering system requirements and accelerating reconstruction efficiency, but also improves the accuracy and reliability of defect detection. It is suitable for non-destructive testing of various power equipment and provides more efficient and innovative technical support for the operation and maintenance of power equipment.

[0112] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A three-dimensional imaging reconstruction system based on X-ray nondestructive testing, characterized in that: include: A historical database unit, an X-ray imaging unit, an image positioning unit, a three-dimensional reconstruction unit, and a defect recognition unit; wherein: The historical database unit includes a historical database storage module and a data call module; the historical database storage module stores a basic information library of electric power equipment, a standard defect type library, a historical detection database and a defect database; the basic information library of electric power equipment stores basic information of electric power equipment, and the basic information includes a three-dimensional model, model, installation location, service life and historical fault records of electric power equipment; the standard defect type library stores standardized three-dimensional defect type models including cracks, relaxation and corrosion; The historical detection database stores the two-dimensional image of the detected area collected by the X-ray imaging unit and the basic information of the detected power equipment; the defect database stores the three-dimensional defect image of the detected area identified as defective by the defect recognition unit; The data calling module quickly reads the contents written into the distributed storage architecture by the historical database storage module; when the X-ray imaging unit collects images, the basic information of the detected power equipment is called according to the detected area; when the defect recognition unit recognizes that the detected area has defects, the data calling module retrieves the standardized three-dimensional defect type model; The X-ray imaging unit includes an image acquisition module and an image preprocessing module; the image acquisition module performs image acquisition on the detected power equipment, and the image acquisition is shot according to the set N angles, and a two-dimensional slice image is output, wherein N is less than or equal to 10; and 3 angles are perpendicular to each other; the image preprocessing module performs image denoising and image enhancement on the image acquired by the image acquisition module in sequence, and outputs a two-dimensional X-ray image to be transmitted to the image positioning unit; the image preprocessing module performs image denoising on the two-dimensional slice image by using a median filter to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise, improves the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image, enhances the contrast, and outputs a two-dimensional X-ray image; The image positioning unit includes: an X-ray image feature extraction module, a historical data feature extraction module and a feature point matching module; the X-ray image feature extraction module extracts geometric features and texture features from the X-ray image to obtain shape information of the equipment components of the X-ray image, and then extracts feature points from the X-ray image, and uses the feature points as the basis for image positioning; the historical data feature extraction module slices the three-dimensional model of the power equipment at equal intervals along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment; the feature point matching module calculates the similarity value between the X-ray image and the two-dimensional image of the power equipment, sets a threshold, and those that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment, and the image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data; The three-dimensional reconstruction unit uses an image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model based on the X-ray image collected from the X-ray imaging unit and the three-dimensional model of the power equipment obtained from the historical database unit, and uses the three-dimensional structure provided by the three-dimensional model of the power equipment as supplementary information; The defect recognition unit determines the defect type by matching the defect features of the 3D reconstructed model with the features of the standardized 3D defect type model based on the 3D reconstructed model and the standardized 3D defect type model in the historical database, using the features extracted by the X-ray image feature extraction module and the historical data feature extraction module, and adopting an image matching algorithm.

2. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The image acquisition module includes an X-ray generator and a detector and a mechanical motion device; The image acquisition module is configured with an X-ray emitter and a detector to take X-ray images of the inspected area of ​​the inspected power equipment; The X-ray generator and detector are installed on a mechanical motion device. The mechanical motion device uses a mechanical motion device to drive the X-ray generator and detector to move according to the position of the detected area. When the X-ray generator and detector reach the position of the detected area, the mechanical motion device drives the X-ray generator and detector to move to the set angle to start image acquisition; The X-ray generator emits X-rays, the X-ray detector receives X-ray penetration signals, records the intensity of X-rays passing through the object at each angle and converts it into digital image signals, and generates a two-dimensional slice image based on the attenuation coefficient of the X-rays passing through each point inside the object.

3. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The image preprocessing module performs image denoising on the two-dimensional slice image by using median filtering to remove salt and pepper noise, then uses wavelet denoising to reduce Gaussian noise and improve the signal-to-noise ratio, and then uses an adaptive histogram equalization algorithm to enhance the denoised image and output a two-dimensional X-ray image by enhancing the contrast.

4. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The X-ray image feature extraction module includes a geometric feature extraction submodule, a texture feature extraction submodule and a feature point extraction submodule; The geometric feature extraction submodule processes the collected X-ray image using an edge detection algorithm, finds the maximum value points of the intensity gradient in the X-ray image, connects the maximum value points into edge lines, and thus extracts the internal contour of the detected device; The texture feature extraction submodule uses a gray level co-occurrence matrix to analyze the image texture and calculate the joint distribution of pixel gray levels at different directions and distances in the X-ray image to supplement the geometric features of the X-ray image; The feature point extraction submodule uses a scale-invariant feature transformation algorithm to extract feature points with scale and rotation invariance from the X-ray image, and calculates extreme points in different scale spaces to represent image details as the basis for image positioning.

5. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The historical data feature extraction module includes two-dimensional cross-sectional image generation and geometric feature extraction; The two-dimensional cross-sectional image generation and geometric feature extraction slice the three-dimensional model of the power equipment in the historical data according to the set direction and equal spacing, and convert the three-dimensional model into a series of two-dimensional cross-sectional images; The two-dimensional cross-sectional image uses an edge detection algorithm to extract geometric features corresponding to the X-ray image and restore the shape contours of the three-dimensional model in different cross-sections.

6. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The feature point matching module includes a primary matching submodule based on feature points, an error matching elimination submodule and a space transformation relationship calculation submodule; The initial matching based on feature points adopts the nearest neighbor matching algorithm to match the feature points extracted from the X-ray image with the feature points of the two-dimensional cross-sectional image of the historical data; the Euclidean distance of the nearest neighbor matching algorithm is used for measurement, and the Euclidean distance between the feature points is used as the similarity measurement to quickly search the historical data for the point closest to the feature point of the X-ray image as the matching point; The error matching elimination adopts a random sampling consistency algorithm to screen the initial matching results, and removes the error matching points through multiple random samplings; The spatial transformation relationship calculation uses the least squares method to solve the linear equations based on the selected feature points to calculate the transformation matrix, which is used as the basis for the spatial transformation between the X-ray image and the three-dimensional model of historical data to perform translation, rotation and scaling transformations.

7. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The three-dimensional reconstruction unit image reconstruction algorithm uses an iterative reconstruction algorithm to obtain a three-dimensional structure of a two-dimensional X-ray image through multiple iterations.

8. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to claim 1, characterized in that: The defect recognition unit extracts shape features and spatial positions from the three-dimensional reconstruction model and the standardized three-dimensional defect type model in the historical database; An image matching algorithm is used to match the defect features of the 3D reconstructed model with the features of the standardized 3D defect type model; The shape feature uses an edge detection algorithm to obtain the defect boundary; The spatial position adopts the position of the defect in a three-dimensional coordinate system; The difference between the shape feature and the spatial position is calculated using the Euclidean distance, and the similarity of the obtained Euclidean distance is regarded as the same defect type within a set threshold; The shape characteristics and spatial position of the three-dimensional reconstructed model are used as defect parameters, which include geometric parameters, position and direction. The defect parameters are combined with the operating parameters and material performance parameters of the power equipment. The operating parameters include load and pressure, and the material performance parameters include strength and toughness. The severity is assessed using a physical model.

9. The three-dimensional imaging reconstruction system based on X-ray nondestructive testing according to any one of claims 1 to 8, characterized in that: Data is transferred between all modules through message queues.

10. A three-dimensional imaging reconstruction method based on X-ray nondestructive testing, characterized in that: The method is implemented based on the three-dimensional imaging reconstruction system based on X-ray non-destructive testing according to any one of claims 1 to 9, and comprises the following steps: Step S1, establish a historical database to facilitate storage and call of the historical database; the historical database storage includes a basic information database of power equipment, a standard defect type database, a historical detection database and a defect database; the basic information database of power equipment stores basic information of power equipment including a three-dimensional model, model, installation location, service life and historical fault records of power equipment; the standard defect type database stores standardized three-dimensional defect type models including cracks, relaxation and corrosion; Step S2, using an X-ray imaging device to perform X-ray imaging, image acquisition and image preprocessing; image acquisition performs image acquisition on the detected power equipment, and the image acquisition is shot according to the set N angles, and a two-dimensional slice image is output, wherein N is less than or equal to 10; wherein 3 angles are perpendicular to each other; image preprocessing performs image denoising and image enhancement on the image acquisition image in sequence, and outputs a two-dimensional X-ray image and transmits it to the image positioning unit; the image preprocessing performs image denoising on the two-dimensional slice image, adopts median filtering to remove salt and pepper noise, and then uses wavelet denoising to reduce Gaussian noise, improves the signal-to-noise ratio, and then adopts an adaptive histogram equalization algorithm to enhance the denoised image, enhances the contrast, and outputs a two-dimensional X-ray image; Step S3, positioning the X-ray image on a specific device. The image positioning includes an X-ray image feature extraction module extracting geometric features and texture features from the X-ray image to obtain shape information of the device components of the X-ray image, and then extracting feature points from the X-ray image, and using the feature points as the basis for image positioning; Image positioning also includes a historical data feature extraction module that slices the three-dimensional model of the power equipment at equal distances along the shooting angle of the X-ray imaging unit to obtain a two-dimensional image of the power equipment; Finally, feature point matching is performed, and the similarity value between the X-ray image and the two-dimensional image of the power equipment is calculated, and a threshold is set. The points that reach the threshold are at the same position; the feature points of the X-ray image are spatially transformed and mapped to the corresponding two-dimensional image of the power equipment. The image positioning unit aligns and positions the two-dimensional X-ray image collected and output by the X-ray imaging unit to the spatial position of the three-dimensional model of the power equipment in the historical data; Step S4, performing three-dimensional reconstruction; using an image reconstruction algorithm to convert the two-dimensional X-ray image into a three-dimensional model as a three-dimensional reconstruction model based on the X-ray image collected from the X-ray imaging and the three-dimensional model of the power equipment obtained from the historical database, and using the three-dimensional structure provided by the three-dimensional model of the power equipment as supplementary information; Step S5, performing defect identification; according to the features extracted from the 3D reconstructed model and the standardized 3D defect type model in the historical database, an image matching algorithm is used to match the defect features of the 3D reconstructed model with the features of the standardized 3D defect type model to determine the defect type.

Images