A method, system, equipment and medium for intelligent detection of construction technology in distribution station area
By acquiring and matching the head-up and upward viewing angle images of the distribution station area, calculating the perspective transformation matrix, and detecting the equipment position and parameters, the problem of relying on experience and incomplete scene coverage of construction technology detection is solved, and efficient construction quality inspection is achieved.
Patent Information
- Application Number
- CN202411144844.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-08-20
AI Technical Summary
The construction process inspection in the distribution station area relies on the experience of the supervisor, resulting in different inspection results. Some technologies cannot cover the panoramic scenes, affecting the construction quality.
By acquiring head-up and upward viewing angle images, extracting feature sets for feature matching, calculating perspective transformation matrix, detecting equipment position and parameters, and realizing standardized detection of equipment position and process parameters.
It improves the quality inspection effect of construction process, covers all equipment of the construction process, reduces the limitations of inspection scenarios, and realizes standardized inspection without relying on supervision experience.
Smart Images

Figure CN118657779B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent detection technology, and in particular to an intelligent detection method, system, equipment and medium for construction process of a distribution station area. Background Art
[0002] During the construction of the distribution station area, in order to ensure that the quality of the project meets the acceptance standards and eliminate potential safety hazards in a timely manner, when the project is partially or fully completed, it is necessary to conduct construction process quality inspection and acceptance of the project in accordance with the acceptance standards.
[0003] At present, the construction process inspection of distribution substations mainly relies on the experience of supervisors. However, the process standards involved in distribution substations are complicated, and some process points need to be inspected by professional equipment. The supervisors have a heavy task of process inspection. At the same time, the professional levels of supervisors vary, making it difficult to strictly implement standardized inspections of construction processes. As a result, the construction process quality inspection results do not meet the expected results, affecting the project construction process quality.
[0004] Some regions have also tried to use intelligent terminals for acceptance, mainly by combining image preprocessing algorithms and deep learning algorithms to identify the number and location of each device in the distribution area, and then compare them with relevant acceptance standards to achieve process acceptance inspection of the distribution area. However, these technologies generally have the problem of incomplete coverage of process points, only identifying the number of components and component defects, and not considering the on-site construction scene, which greatly limits the application scenarios of the technology. Therefore, the use of intelligent terminals for acceptance will also lead to the construction process quality inspection effect not achieving the expected effect, affecting the project construction process quality. Summary of the invention
[0005] The present application aims to propose a method, system, equipment and medium for intelligent detection of construction technology in distribution station areas, which can improve the detection effect of construction technology quality, thereby improving the quality of project construction technology.
[0006] In a first aspect, an embodiment of the present application provides a method for intelligently detecting construction technology of a distribution station area, the method comprising:
[0007] Acquire a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives;
[0008] Extracting a first feature set from the head-up perspective image, and extracting a second feature set from the upward perspective image; and performing feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs;
[0009] Calculating a target perspective transformation matrix according to the plurality of feature matching pairs;
[0010] Detecting the position coordinates corresponding to each device in the upward viewing angle image;
[0011] Based on the target perspective transformation matrix, the position coordinates of each device are mapped to a head-up viewing angle to obtain the transformed position coordinates of each device;
[0012] Calculating a ratio value between an actual size and image pixels corresponding to a first device among the devices according to the transformed position coordinates corresponding to the first device, wherein the first device is a device with a standard size;
[0013] Calculating the process parameters of each device according to the transformed position coordinates of each device and the ratio value;
[0014] According to the process parameters of each device, the construction process of each device in the distribution station area is detected to obtain the construction process detection result of the distribution station area.
[0015] Compared with the prior art, the first aspect of the present application has the following beneficial effects:
[0016] This method calculates the target perspective transformation matrix based on multiple feature matching pairs, detects the position coordinates corresponding to each device in the upward-looking perspective image, maps the position coordinates of each device to the head-up perspective based on the target perspective transformation matrix, and maps the equipment position parameters of the distribution station area in the head-up perspective to the head-up perspective by associating the head-up perspective image and the upward-looking perspective image of the distribution station area, thereby avoiding directly shooting the panoramic head-up perspective image of the distribution station area, so as to reduce the shooting distance of the distribution station area image and solve the problem of limited detection scene. According to the transformed position coordinates corresponding to the first device in each device, the ratio value of the actual size and image pixel corresponding to the first device is calculated, and the process parameters of each device are calculated according to the transformed position coordinates and ratio values of each device. According to the process parameters of each device, the construction process of each device in the distribution station area is detected, which can cover the construction process detection of each device without relying on the experience of the supervisor, realize the standardized detection of the construction process, and improve the effect of the construction process quality detection, thereby improving the construction process quality of the project.
[0017] In some implementations, acquiring a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives includes:
[0018] An initial head-up perspective image and an initial upward perspective image of the distribution station area are collected through a preset semi-transparent mask and a preset angle range;
[0019] The initial head-up perspective image is input into a trained perspective matching model for perspective matching to obtain a head-up perspective image of a distribution station area with a perspective match, and the initial upward perspective image is input into the trained perspective matching model for perspective matching to obtain an upward perspective image of a distribution station area with a perspective match, wherein the trained perspective matching model is obtained by training multiple groups of head-up perspective images and upward perspective images with standard perspectives, and the standard perspective is a perspective that meets the preset angle range.
[0020] In some implementations, the extracting a first feature set from the head-up perspective image and the extracting a second feature set from the upward perspective image; and performing feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs include:
[0021] According to the preset semi-transparent mask, an outline area of the power distribution station area in the head-up perspective image and the upward perspective image is acquired;
[0022] Using a feature extraction method to extract a first feature set of an outline area of a power distribution station area in the head-up perspective image;
[0023] A feature extraction method is used to extract a second feature set of the outer shape area of the power distribution station area in the upward viewing angle image;
[0024] Calculating the Euclidean distance between each first feature point in the first feature set and each second feature point in the second feature set to obtain a Euclidean distance matrix, wherein the Euclidean distance matrix is used to indicate feature matching pairs between each first feature point in the first feature set and each second feature point in the second feature set;
[0025] The feature matching pairs in the Euclidean distance matrix are screened to obtain a plurality of screened feature matching pairs.
[0026] In some implementations, calculating a target perspective transformation matrix according to the plurality of feature matching pairs includes:
[0027] Randomly selecting a plurality of first feature matching pairs from the plurality of feature matching pairs, wherein the number of the first feature matching pairs is less than the number of the feature matching pairs;
[0028] Calculating a perspective transformation matrix according to the plurality of first feature matching pairs;
[0029] Calculating a reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix, wherein the second feature matching pair is used to indicate a feature matching pair in the plurality of feature matching pairs other than the plurality of first feature matching pairs;
[0030] Sorting the reprojection errors corresponding to each second feature matching pair, and selecting a median value among the sorted reprojection errors;
[0031] If the median value is smaller than a preset error threshold, the perspective transformation matrices corresponding to the plurality of first feature matching pairs associated with the median value are used as the target perspective transformation matrix.
[0032] In some implementations, calculating the reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix includes:
[0033] Projecting the coordinates of the second feature points in each second feature matching pair by using the perspective transformation matrix to obtain the projection coordinates of the second feature points in each second feature matching pair;
[0034] Acquire the coordinates of a first feature point that matches the second feature point;
[0035] A reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated according to the projection coordinates of the second feature point and the coordinates of the first feature point matching the second feature point.
[0036] In some implementations, the reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated in the following manner:
[0037]
[0038] in, Indicates The coordinates of the first feature point that matches the second feature point, Indicates The projection coordinates of the second feature point, represents the reprojection error corresponding to each second feature matching pair, Represents the number of second feature matching pairs.
[0039] In some implementations, calculating the ratio between the actual size and the image pixel corresponding to the first device among the devices according to the transformed position coordinates corresponding to the first device includes:
[0040]
[0041]
[0042] in, Indicates the ratio between the actual size of the first device and the image pixels. Indicates the actual size of the first device. Indicates the image pixel length corresponding to the first device, Indicates the transformed position coordinates corresponding to the upper left position of the first device coordinate, Indicates the transformed position coordinates corresponding to the lower right position of the first device coordinate.
[0043] In a second aspect, the embodiment of the present application further provides an intelligent detection system for construction technology of a distribution station area, the system comprising:
[0044] An image acquisition unit, used to acquire a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives;
[0045] A feature extraction unit, configured to extract a first feature set from the head-up perspective image and a second feature set from the upward perspective image; and perform feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs;
[0046] A first calculation unit, configured to calculate a target perspective transformation matrix according to the plurality of feature matching pairs;
[0047] A position detection unit, used to detect the position coordinates corresponding to each device in the upward viewing angle image;
[0048] A coordinate mapping unit, used to map the position coordinates of each device to a head-up viewing angle based on the target perspective transformation matrix to obtain the transformed position coordinates of each device;
[0049] A second calculation unit, configured to calculate a ratio value between an actual size and image pixels corresponding to a first device among the devices according to the transformed position coordinates corresponding to the first device, wherein the first device is a device with a standard size;
[0050] A third calculation unit, used for calculating the process parameters of each device according to the transformed position coordinates of each device and the ratio value;
[0051] The process detection unit is used to detect the construction process of each device in the distribution substation area according to the process parameters of each device, and obtain the construction process detection result of the distribution substation area.
[0052] In the third aspect, an embodiment of the present application also provides an electronic device, comprising at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the intelligent detection method for construction technology of a distribution station area as described above.
[0053] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned intelligent detection method for construction technology of a distribution station area.
[0054] It can be understood that the beneficial effects of the second to fourth aspects compared with the related art are the same as the beneficial effects of the first aspect compared with the related art. Please refer to the relevant description in the first aspect, and no further details will be given here. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0056] Figure 1 It is a flow chart of an embodiment of the intelligent detection method for construction technology of a distribution station area provided by the present application;
[0057] Figure 2 is a schematic diagram of the shooting mask guidance in the embodiment of the intelligent detection method for the construction process of the distribution station area provided by the present application, wherein: Figure 2 Figure a in the middle is a schematic diagram of the shooting mask at a horizontal viewing angle. Figure 2 Middle b is a schematic diagram of mask-guided shooting;
[0058] Figure 3 It is a schematic diagram of defining a spatial attitude angle in an embodiment of the intelligent detection method for construction technology of a distribution station area provided by the present application;
[0059] Figure 4 Schematic diagram of pictures taken from a horizontal perspective and a downward perspective in an embodiment of the intelligent detection method for construction technology of a distribution station area provided by the present application, wherein: Figure 4 Figure a is a schematic diagram of a head-up view image that meets the requirements. Figure 4 b is a schematic diagram of an upward-looking perspective image that meets the requirements;
[0060] Figure 5 is a schematic diagram of the matching area between the head-up perspective image and the upward perspective image in the embodiment of the intelligent detection method for the construction process of the distribution station area provided by the present application, wherein: Figure 5 Figure a is a schematic diagram of the matching area of the head-up view image. Figure 5 Middle b is a schematic diagram of the matching area of the upward viewing angle image;
[0061] Figure 6 is a schematic diagram of matching point pairs between a head-on perspective image and an upward perspective image in an embodiment of the intelligent detection method for construction technology of a distribution station area provided by the present application, wherein: Figure 6Figure a is a schematic diagram of matching points in the head-up view image. Figure 6 b is a schematic diagram of matching points in the upward viewing angle image;
[0062] Figure 7 It is a flow chart of an embodiment of the intelligent detection method for construction technology of a distribution station area provided in the present application. DETAILED DESCRIPTION
[0063] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.
[0064] In the description of this application, if there is a description of first, second, etc., it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.
[0065] It should also be understood that the references to "one embodiment" or "some embodiments" described in the specification of the embodiments of the present application mean that one or more embodiments of the embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Thus, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.
[0066] In the description of the present application, it should be understood that the descriptions involving orientation, such as the orientation or positional relationship indicated as up, down, etc., are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.
[0067] In the description of this application, it should be noted that, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meaning of the above terms in this application based on the specific content of the technical solution.
[0068] First, some nouns involved in this application are analyzed:
[0069] Reprojection error: refers to the difference between the projection of the real three-dimensional space point on the image plane (that is, the pixel point on the image) and the reprojection (actually a virtual pixel point obtained by calculating the value).
[0070] During the construction of the distribution station area, in order to ensure that the quality of the project meets the acceptance standards and eliminate potential safety hazards in a timely manner, when the project is partially or fully completed, it is necessary to conduct construction process quality inspection and acceptance of the project in accordance with the acceptance standards.
[0071] At present, the construction process inspection of the distribution area mainly relies on the experience of the supervisors. However, the process standards involved in the distribution area are complicated, and some process points need to be inspected by professional equipment. The supervisors have a heavy task of process inspection. At the same time, the professional levels of the supervisors are different, and it is difficult to strictly implement the standardized inspection of the construction process, resulting in the construction process quality inspection effect not meeting the expected effect, affecting the quality of the project construction process. Some areas have also tried to use intelligent terminal acceptance, mainly by combining image preprocessing algorithms and deep learning algorithms to realize the identification of the number and location of each device in the distribution area, and then compare with the relevant acceptance standards to realize the process acceptance inspection of the distribution area. However, these technologies generally have the problem of incomplete coverage of process points, only identifying the number of components and component defects, and not considering the on-site construction scene, resulting in the application scenarios of the technology being greatly limited.
[0072] In order to solve the above-mentioned problems of inconsistent manual inspection levels, construction scene limitations, and inspection results that do not meet expected results, the present application proposes an intelligent inspection method, system, equipment, and medium for construction technology in a distribution station area.
[0073] Reference Figure 1 The present application embodiment provides a method for intelligent detection of construction technology of a distribution station area, and the method comprises the following steps:
[0074] Step S100, obtaining a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives;
[0075] Step S200, extracting a first feature set from the head-up perspective image, and extracting a second feature set from the upward perspective image; and performing feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs;
[0076] Step S300, calculating a target perspective transformation matrix according to a plurality of feature matching pairs;
[0077] Step S400, detecting the position coordinates corresponding to each device in the upward viewing angle image;
[0078] Step S500: Mapping the position coordinates of each device to a head-up viewing angle based on a target perspective transformation matrix to obtain the transformed position coordinates of each device;
[0079] Step S600: Calculate the ratio of the actual size and image pixels corresponding to the first device according to the transformed position coordinates corresponding to the first device among the devices, wherein the first device is a device with a standard size;
[0080] Step S700, calculating the process parameters of each device according to the transformed position coordinates and ratio values of each device;
[0081] Step S800: detect the construction process of each device in the distribution substation area according to the process parameters of each device, and obtain the construction process detection result of the distribution substation area.
[0082] In this embodiment, by calculating the target perspective transformation matrix based on multiple feature matching pairs, the position coordinates corresponding to each device in the upward-looking perspective image are detected, and based on the target perspective transformation matrix, the position coordinates of each device are mapped to the head-up perspective. By associating the head-up perspective image and the upward-looking perspective image of the distribution station area, the equipment position parameters of the distribution station area in the upward perspective are mapped to the head-up perspective, which can avoid directly shooting the panoramic head-up perspective image of the distribution station area, so as to reduce the shooting distance of the distribution station area image and solve the problem of limited detection scene. According to the transformed position coordinates corresponding to the first device in each device, the ratio value of the actual size and image pixel corresponding to the first device is calculated, and the process parameters of each device are calculated according to the transformed position coordinates and ratio values of each device. According to the process parameters of each device, the construction process of each device in the distribution station area is detected, which can cover the construction process detection of each device without relying on the experience of the supervisor, realize the standardized detection of the construction process, and improve the effect of the construction process quality detection, thereby improving the project construction process quality.
[0083] The above-mentioned head-up perspective image may be an image of the distribution station area taken at a head-up angle.
[0084] The above-mentioned upward-looking perspective image may be an image of the distribution station area taken from an upward-looking angle.
[0085] The above-mentioned first feature set is extracted from the head-on perspective image, and the second feature set is extracted from the upward perspective image; and feature matching is performed on the first feature set and the second feature set to obtain multiple feature matching pairs. Feature extraction can be performed by using a scale-invariant feature transformation method, or by using methods such as ORB (Oriented FAST and Rotated BRIEF), SURF (Speeded Up Robust Features) or FAST (Features from Accelerated Segment Test). This embodiment does not make any specific restrictions.
[0086] The target perspective transformation matrix is calculated based on multiple feature matching pairs. Several feature matching pairs are randomly selected from the multiple feature matching pairs, and then the target perspective transformation matrix is directly calculated using a perspective transformation method based on the selected feature matching pairs.
[0087] The first device is a device with a standard size, and may be a device in accordance with a unified standard specification, and the standard size of the device is a fixed and known size.
[0088] The above-mentioned method of detecting the construction technology of each equipment in the distribution substation area according to the process parameters of each equipment to obtain the distribution substation area construction technology detection result can be to compare the process parameters of each equipment with the corresponding relevant enterprise standards, and use the obtained qualified status of each equipment and the relevant error deviation range as the distribution substation area construction technology detection result.
[0089] In some implementations, obtaining a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives includes:
[0090] An initial head-up perspective image and an initial upward perspective image of the distribution station area are collected through a preset semi-transparent mask and a preset angle range;
[0091] The initial head-up perspective image is input into the trained perspective matching model for perspective matching to obtain a head-up perspective image of the distribution station area with perspective matching, and the initial upward perspective image is input into the trained perspective matching model for perspective matching to obtain an upward perspective image of the distribution station area with perspective matching, wherein the trained perspective matching model is obtained by training multiple groups of head-up perspective images and upward perspective images with standard perspectives, and the standard perspective is a perspective that meets a preset angle range.
[0092] In this embodiment, the head-up perspective image and the upward perspective image of the distribution station area are collected through a preset semi-transparent mask and a preset angle range, and then the head-up perspective image and the upward perspective image are perspective matched through a trained perspective matching model to determine whether the shooting angles of the head-up perspective image and the upward perspective image meet the shooting requirements (i.e., meet the preset angle range), so that the captured image is more in line with the requirements, the quality of the image captured in the distribution station area is improved, and a good data foundation is laid for subsequent feature matching.
[0093] Specifically, an initial horizontal viewing angle image and an initial upward viewing angle image of the power distribution station area are collected through a preset semi-transparent mask and a preset angle range.
[0094] The trained perspective matching model is obtained by training multiple groups of head-on perspective images and upward perspective images with standard perspectives (i.e., perspectives that meet a preset angle range).
[0095] The initial head-up perspective image is input into the trained perspective matching model for perspective matching. If the trained perspective matching model determines that the initial head-up perspective image meets the perspective matching requirements (i.e., the preset angle range), the trained perspective matching model will output a head-up perspective image of the distribution station area with a perspective match; if the trained perspective matching model determines that the initial head-up perspective image does not meet the perspective matching requirements, the trained perspective matching model will output a result that does not meet the perspective matching requirements. When receiving a result that does not meet the perspective matching requirements, the photographer will re-collect a new head-up perspective image of the distribution station area through a preset semi-transparent mask and a preset angle range, and then re-input the new head-up perspective image into the trained perspective matching model for perspective matching until the trained perspective matching model outputs a head-up perspective image of the distribution station area with a perspective match.
[0096] The initial upward-looking perspective image is input into the trained perspective matching model for perspective matching. If the trained perspective matching model determines that the initial upward-looking perspective image meets the perspective matching requirements (i.e., the preset angle range), the trained perspective matching model will output an upward-looking perspective image of the distribution station area with a perspective match; if the trained perspective matching model determines that the initial upward-looking perspective image does not meet the perspective matching requirements, the trained perspective matching model will output a result that does not meet the perspective matching requirements. Upon receiving a result that does not meet the perspective matching requirements, the photographer will re-collect a new upward-looking perspective image of the distribution station area through a preset semi-transparent mask and a preset angle range, and then re-input the new upward-looking perspective image into the trained perspective matching model for perspective matching, until the trained perspective matching model outputs an upward-looking perspective image of the distribution station area with a perspective match.
[0097] The above-mentioned perspective matching model can be an existing image classification algorithm, or it can be other existing deep learning models capable of classification matching, which is not described in detail in this embodiment.
[0098] The above-mentioned preset semi-transparent mask and preset angle range can be set according to actual needs, and this embodiment does not impose any specific limitations.
[0099] The above-mentioned preset semi-transparent mask can guide the user to adjust the position of the distribution station area in the picture by adjusting the position of the smart mobile terminal relative to the distribution station area, ensuring that the distribution station area is placed within the mask area, thereby obtaining a captured image that meets subsequent calculation and detection requirements.
[0100] In some implementations, extracting a first feature set from a head-up perspective image, and extracting a second feature set from an upward perspective image; and performing feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs, including:
[0101] According to a preset semi-transparent mask, the outer shape area of the power distribution station area in the head-up perspective image and the upward perspective image is obtained;
[0102] A feature extraction method is used to extract a first feature set of the outer shape area of the power distribution station area in the head-up perspective image;
[0103] A second feature set of the outer shape area of the distribution station area in the upward viewing angle image is extracted by using a feature extraction method;
[0104] Calculating the Euclidean distance between each first feature point in the first feature set and each second feature point in the second feature set to obtain a Euclidean distance matrix, wherein the Euclidean distance matrix is used to indicate feature matching pairs between each first feature point in the first feature set and each second feature point in the second feature set;
[0105] The feature matching pairs in the Euclidean distance matrix are screened to obtain a plurality of screened feature matching pairs.
[0106] In this embodiment, the outer shape area of the distribution station area in the head-on view image and the upward view image is obtained according to a preset semi-transparent mask, which is used to limit the area of feature point detection and matching, thereby ensuring that the matched feature points are on the same spatial plane as much as possible, reducing the interference of the background area, and improving the accuracy of the perspective transformation matrix calculation in the later stage. The feature matching pairs in the Euclidean distance matrix are screened to obtain a plurality of screened feature matching pairs, and only those matching items whose best match is significantly better than the second best match are retained, thereby reducing errors caused by mismatching or noise.
[0107] In some implementations, calculating a target perspective transformation matrix based on a plurality of feature matching pairs includes:
[0108] Randomly selecting a plurality of first feature matching pairs from the plurality of feature matching pairs, wherein the number of the first feature matching pairs is less than the number of the feature matching pairs;
[0109] Calculating a perspective transformation matrix based on the plurality of first feature matching pairs;
[0110] Calculate, according to the perspective transformation matrix, a reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs, wherein the second feature matching pair is used to indicate a feature matching pair in the plurality of feature matching pairs excluding the plurality of first feature matching pairs;
[0111] Sort the reprojection errors corresponding to each second feature matching pair, and select the median value of the sorted reprojection errors;
[0112] If the median value is less than a preset error threshold, the perspective transformation matrices corresponding to the plurality of first feature matching pairs associated with the median value are used as the target perspective transformation matrix.
[0113] In this embodiment, the reprojection error corresponding to each second feature matching pair in multiple feature matching pairs is calculated according to the perspective transformation matrix; then the reprojection error corresponding to each second feature matching pair is sorted, and the median of the sorted reprojection error is selected. If the median is less than the preset error threshold, the perspective transformation matrices corresponding to multiple first feature matching pairs related to the median are used as the target perspective transformation matrix. By selecting the perspective transformation matrix with a small error as the target perspective transformation matrix, and then performing subsequent construction process quality inspection, the construction process quality inspection effect can be improved, thereby improving the project construction process quality.
[0114] In some implementations, calculating the reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix includes:
[0115] Projecting the coordinates of the second feature point in each second feature matching pair through the perspective transformation matrix to obtain the projection coordinates of the second feature point in each second feature matching pair;
[0116] Obtaining coordinates of a first feature point that matches a second feature point;
[0117] A reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated according to the projection coordinates of the second feature point and the coordinates of the first feature point matching the second feature point.
[0118] In this embodiment, the coordinates of the second feature point in each second feature matching pair are projected through the perspective transformation matrix, and the median corresponding to the perspective transformation matrix is calculated according to the reprojection error corresponding to each second feature matching pair, and then the median with small error is used to determine whether the perspective transformation matrix is the target perspective transformation matrix, and the perspective transformation matrix with small error can be selected as the target perspective transformation matrix, and then the subsequent construction process quality inspection can be carried out, which can improve the effect of construction process quality inspection, thereby improving the project construction process quality. This embodiment lays a good data foundation for the later construction process quality inspection.
[0119] In some implementations, the reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated as follows:
[0120]
[0121] in, Indicates The coordinates of the first feature point that matches the second feature point, Indicates The projection coordinates of the second feature point, represents the reprojection error corresponding to each second feature matching pair, Represents the number of second feature matching pairs.
[0122] In some implementations, calculating the ratio of the actual size corresponding to the first device and the image pixel value according to the transformed position coordinates corresponding to the first device in each device includes:
[0123]
[0124]
[0125] in, Indicates the ratio between the actual size of the first device and the image pixels. Indicates the actual size of the first device. Indicates the image pixel length corresponding to the first device, Indicates the transformed position coordinates corresponding to the upper left position of the first device coordinate, Indicates the transformed position coordinates corresponding to the lower right position of the first device coordinate.
[0126] In this embodiment, the ratio value of the actual size and image pixels corresponding to the first device is calculated based on the transformed position coordinates corresponding to the first device in each device, so as to use a unified standard for subsequent construction process quality inspection, thereby realizing standardized inspection of the construction process.
[0127] To facilitate understanding by those skilled in the art, a set of best embodiments is provided below:
[0128] In this embodiment, the multi-view association and target detection technology in computer vision is used to realize the intelligent detection of the equipment location parameters of the distribution station area and complete the calculation of the on-site construction process parameters, aiming to solve the technical defects in the construction process detection of the distribution station area. The specific implementation steps of this embodiment are as follows:
[0129] (1) Use the smart mobile terminal to collect the head-up and head-down images of the distribution station area according to the collection instructions set in advance, and ensure that the shooting angle and shooting direction meet the input requirements of the subsequent detection algorithm. The collection instructions include shooting mask instructions, real-time posture control, and perspective matching control, as follows:
[0130] 1) Shooting mask guidance: By integrating a pre-designed semi-transparent mask into the shooting interface of the smart mobile terminal, the user is guided to adjust the position of the distribution station area in the picture by adjusting the position of the smart mobile terminal relative to the distribution station area, ensuring that the distribution station area is placed within the mask area, thereby obtaining a shot image that meets the algorithm requirements.
[0131] The specific design process of the semi-transparent mask is as follows: ① Use a smart mobile terminal without a mask to obtain a standard distribution station area image that meets the requirements of the subsequent detection algorithm; ② Use image processing software to use straight lines to outline the outer shape of the standard distribution station area image; ③ Set all pixels of the standard distribution station area image to background color pixel values, such as RGB (0, 0, 0), and set its opacity, and then set the station area outer shape area to transparent, and you can get a semi-transparent mask image of the station area; ④ Cover the semi-transparent mask image on the layer of the shooting interface and use it when shooting. Figure 2 As shown, Figure 2 Figure a in the middle is a schematic diagram of a shooting mask (i.e., a semi-transparent mask) at a level viewing angle. Figure 2 Figure b in the middle is a schematic diagram of mask-guided shooting.
[0132] 2) Real-time attitude control: The spatial attitude angle at the time of collection is obtained in real time through the orientation sensor on the smart mobile terminal, mainly including the pitch angle , yaw angle ,like Figure 3 As shown, the pitch angle Is around smart mobile terminal X b When the axis rotates, Y b The angle between the axis and the horizontal plane, the yaw angle is the coordinate system Z of the smart mobile terminal b The angle of axis rotation is controlled by the pre-set pitch angle and yaw angle allowable range to control the shooting angle of the smart mobile terminal within the pre-set angle allowable range (i.e. the preset angle range), ensuring that the image imaging angle meets the requirements for the shooting posture angle for looking straight ahead and looking up.
[0133] 3) Through the perspective matching algorithm module (i.e., perspective matching model) integrated in the smart mobile terminal, further verify whether the user shoots according to the shooting mask and perspective. The perspective matching algorithm is an image classification algorithm that matches the head-up perspective image and the upward perspective image of the distribution station area, and is used to identify whether the head-up perspective image and the upward perspective image collected by the smart mobile terminal match the corresponding perspectives of the standard head-up perspective image or the standard upward perspective image. The image classification algorithm is trained using standard head-up perspective images and standard upward perspective images, wherein the standard head-up perspective image may be in accordance with the allowable angle range set in advance in the above step 2), and the standard upward perspective image may be in accordance with the allowable angle range set in advance in the above step 2). The schematic diagram of the head-up perspective image that meets the requirements is shown in the figure below. Figure 4 As shown in a, the schematic diagram of the upward viewing angle image that meets the requirements is as follows Figure 4 As shown in b.
[0134] It should be noted that the viewing angle matching algorithm of this embodiment adopts the existing image classification algorithm, which is not described in detail in this embodiment.
[0135] (2) Feature point detection and matching are performed on the head-on view image and the upward view image, as follows:
[0136] 1) Obtain the matching area. Based on the shooting mask (i.e., semi-transparent mask) set in advance in step (1), obtain the outer shape area of the distribution station area in the head-up view image and the upward view image respectively, which is used to limit the area of feature point detection and matching, thereby ensuring that the matched feature points are on the same spatial plane as much as possible, reducing the interference of the background area, and improving the accuracy of the subsequent perspective transformation matrix calculation. The matching area between the head-up view image and the upward view image is as follows: Figure 5 shown.
[0137] 2) Calculate the scale-invariant feature points (SIFT features) of the outer shape of the distribution station area in the head-up view image and the upward view image, that is, the first feature set , the second feature set ,in and are the number of feature points in the horizontal perspective image and the upward perspective image, respectively. is the dimension of SIFT features;
[0138] 3) Match the feature points by calculating the Euclidean distance between feature point pairs. The specific calculation is as follows:
[0139] ① Calculate the Euclidean distance matrix between each feature point on the upward-looking image (i.e., each first feature point in the first feature set) and each feature point on the horizontal-looking image (i.e., each second feature point in the second feature set) ,
[0140] (1)
[0141] (2)
[0142] in, Represents the Euclidean distance matrix The Euclidean distance of a pair of points in , Indicates the first The feature point dimensions, Indicates the first The feature point dimensions.
[0143] ②Based on the Euclidean distance matrix Remove noisy matching points, reducing errors due to false matches or noise by keeping only those matches where the best match is significantly better than the second best match.
[0144] Specifically, when the ratio of the second closest distance to the closest distance of the matching point is greater than the threshold When , keep this matching point pair, the steps are as follows: a) find The minimum and second minimum values of each row; b) Calculate the ratio of the second minimum value to the minimum value r ; c) If , then keep the matching point pair, otherwise delete it. The matching feature point pairs that are retained (i.e., the multiple feature matching pairs after screening) are , the schematic diagram of matching point pairs is as follows Figure 6 shown.
[0145] It should be noted that the threshold value of this embodiment The method can be modified according to the actual situation, and this embodiment does not impose any specific restrictions. This embodiment can use Euclidean distance, Manhattan distance, Hamming distance, or cosine similarity, and this embodiment does not impose any specific restrictions.
[0146] (3) Based on matching feature point pairs (i.e., multiple feature matching pairs after screening) , calculate the perspective transformation matrix , the specific steps are as follows:
[0147] 1) Randomly select 4 groups of matching feature point pairs (i.e., 4 first feature matching pairs) from all the retained matching feature point pairs (i.e., multiple feature matching pairs after screening) as the minimum sample set;
[0148] 2) Use these 4 sets of feature point pairs to calculate the temporary perspective transformation matrix , the specific formula is as follows:
[0149] (3)
[0150] in,( , )、( , ) are the coordinate points of the same feature point of the object in the horizontal and upward viewing angles, Represents the perspective transformation matrix.
[0151] Expand formula (3) as follows:
[0152] (4)
[0153] We can get:
[0154] (5)
[0155] From formula (5), we can get a set of matching feature point pairs ( , )、( , ) can obtain 2 sets of equations.
[0156] In perspective transformation, and ( ) have the same transformation effect, as follows:
[0157] (6)
[0158] (7)
[0159] Normally, ,but There are only 8 unknowns:
[0160] (8)
[0161] According to formula (5), formula (8) and four randomly selected matching feature point pairs, we can get The only solution of .
[0162] 3) Use formula (4) to project the remaining matching feature point pairs (i.e., the second feature matching pairs) except the four randomly selected group feature matching point pairs among the retained matching feature point pairs, and calculate the reprojection error of each matching feature point pair (i.e., the second feature matching pair): , get the reprojection error of multiple feature point pairs, and calculate the median of the reprojection error of multiple feature point pairs ;
[0163] (9)
[0164] (10)
[0165] in, Represents the number of remaining matching feature point pairs (i.e., second feature matching pairs), represents the coordinates of the matching point (i.e., the first feature point) of the feature point of the upward-looking image in the remaining matching feature point pairs on the horizontal-looking image. Represents the coordinates of the feature point (second feature point) in the upward viewing angle image in the remaining matching point pair after projection (i.e., projection coordinates).
[0166] 4) The median of the reprojection error The error threshold set in advance For comparison, if , then stop the iterative optimization; otherwise, continue to repeat steps 1) to 3);
[0167] 5) Take the iterative process Minimum value (i.e. The perspective transformation matrix corresponding to as the optimal solution (i.e., the target perspective transformation matrix).
[0168] (4) Based on the target detection model framework, a target detection model for distribution area equipment is constructed to detect the types and position coordinates of each device in the upward-looking image. For example, the equipment types are poles, transformers, integrated distribution boxes, and crossarms. Based on the target perspective transformation matrix Mapping the device position coordinates detected by the upward-viewing perspective image to the horizontal-viewing perspective to obtain the position coordinates of each device after the upward-viewing perspective image is transformed (i.e., the transformed position coordinates);
[0169] (11)
[0170] in, , It is the detection rectangle coordinate point of a device in the distribution station area from the upward perspective. , for , Map to the coordinate point corresponding to the head-up viewing angle.
[0171] It should be noted that the target detection model of this embodiment adopts the existing technology, which is not described in detail in this embodiment.
[0172] (5) Calculate the process point parameter detection results based on the position coordinates of all equipment transformed to the head-up perspective, as follows;
[0173] 1) Calculation of actual size and image pixel ratio. The specific calculation steps are as follows: a) First, select a device with standard size as a reference, specifically the transformer double-pole support stand (i.e. the first device). According to the enterprise standard specification, its length b) Obtain the position coordinates of the transformer double-rod support stand after perspective transformation (i.e., the transformed position coordinates corresponding to the first device) through step (4), and calculate the image pixel length of the transformer double-rod support stand ; c) Calculate the actual size and image pixel ratio The calculation formula is as follows:
[0174] (12)
[0175] (13)
[0176] in, Represents the upper left coordinate point of the transformer double-rod support stand detection rectangle after perspective transformation Coordinates (i.e., the transformed position coordinates corresponding to the upper left position of the first device coordinate), Represents the lower right coordinate point of the transformer double-rod support stand detection rectangle after perspective transformation Coordinates (i.e., the transformed position coordinates corresponding to the lower right position of the first device coordinate), Indicates the actual length of the transformer double-pole support stand (i.e. the actual size corresponding to the first device), Indicates the actual size and image pixel ratio (that is, the ratio of the actual size and image pixels corresponding to the first device).
[0177] 2) Calculation of process point parameters. According to step (4), the position coordinate points of each device after perspective transformation and the actual size and image pixel ratio calculated by formula (12) can be used to calculate the parameter values of different process points, such as root-to-ground distance, pole height, cross arm height from the ground, etc. Here, the root-to-ground distance of the distribution station area is taken as an example. Assuming that there are two poles in the distribution station area, that is, there are two detection rectangles, so four horizontal coordinate points can be obtained. The root-to-ground distance of the distribution station area is calculated as follows:
[0178] (14)
[0179] in, Indicates the actual size of the root opening distance. , They represent the upper left and lower right horizontal coordinate points of the left pole after perspective transformation. , They respectively represent the upper left and lower right horizontal coordinate points of the right pole after perspective transformation.
[0180] (6) By establishing a standard process knowledge base, the standard values and allowable error ranges of various process parameters in the distribution substation area are structured, and the process point parameter test results are compared with the standard values and error ranges in the standard process knowledge base, so as to automatically obtain the qualified status and error deviation range of each process point in the distribution substation area.
[0181] It should be noted that the standard process knowledge base of this embodiment is established according to relevant enterprise standards, in which the standard specifications of each process point are described. This is to structure the relevant description, such as the upper and lower limits of the value of a process point. Since the relevant standards of different enterprises are different, this embodiment does not make specific restrictions on the standard process knowledge base.
[0182] In this embodiment, the method of this embodiment has the following beneficial effects:
[0183] 1) Significantly improve image quality: Integrated shooting guidance, including semi-transparent mask, posture control and perspective matching algorithm, ensures clear and accurate images, providing reliable data for subsequent analysis to improve process inspection quality.
[0184] 2) Break through the limitations of detection scenarios: Establish a correlation between the upward-looking and horizontal-looking images of the distribution station area, and map the equipment location parameters of the distribution station area from the upward-looking perspective to the horizontal-looking perspective, in order to avoid directly taking panoramic horizontal-looking pictures of the distribution station area, so as to reduce the shooting distance of the distribution station area image and solve the problem of limited detection scenarios.
[0185] 3) Realize comprehensive automated detection: Based on the target detection model, the category and location of each process equipment in the distribution station area are automatically identified, and the process point parameters associated with each equipment are calculated, aiming to realize automatic detection of the construction process, which can fully cover the acceptance process points and improve work efficiency.
[0186] Reference Figure 7 The embodiment of the present application also provides a distribution station area construction process intelligent detection system, the system includes an image acquisition unit 100, a feature extraction unit 200, a first calculation unit 300, a position detection unit 400, a coordinate mapping unit 500, a second calculation unit 600, a third calculation unit 700 and a process detection unit 800, wherein:
[0187] An image acquisition unit 100 is used to acquire a head-up perspective image and an upward perspective image of a distribution station area with matching perspectives;
[0188] The feature extraction unit 200 is used to extract a first feature set from the head-up perspective image and a second feature set from the upward perspective image; and perform feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs;
[0189] A first calculation unit 300 is used to calculate a target perspective transformation matrix according to a plurality of feature matching pairs;
[0190] The position detection unit 400 is used to detect the position coordinates corresponding to each device in the upward viewing angle image;
[0191] A coordinate mapping unit 500 is used to map the position coordinates of each device to a head-up viewing angle based on a target perspective transformation matrix to obtain the transformed position coordinates of each device;
[0192] A second calculation unit 600 is used to calculate a ratio value between an actual size and image pixels corresponding to the first device according to the transformed position coordinates corresponding to the first device in each device, wherein the first device is a device with a standard size;
[0193] The third calculation unit 700 is used to calculate the process parameters of each device according to the transformed position coordinates and proportion values of each device;
[0194] The process detection unit 800 is used to detect the construction process of each device in the distribution substation area according to the process parameters of each device, and obtain the construction process detection result of the distribution substation area.
[0195] It should be noted that since the intelligent detection system for construction technology of a distribution station area in this embodiment and the above-mentioned intelligent detection method for construction technology of a distribution station area are based on the same inventive concept, the corresponding contents in the method embodiment are also applicable to the system embodiment and will not be described in detail here.
[0196] An embodiment of the present invention further provides an electronic device, including: at least one control processor and a memory for communicating with the at least one control processor.
[0197] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0198] The non-transient software program and instructions required to implement the above-mentioned method for intelligent detection of construction technology in a distribution substation area are stored in the memory. When executed by the processor, the above-mentioned method for intelligent detection of construction technology in a distribution substation area is executed. For example, the above-mentioned method is executed. Figure 1 The method comprises steps S100 to S800.
[0199] The system embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0200] The embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. The computer-executable instructions are executed by one or more control processors, which can enable the one or more control processors to execute a distribution station area construction process intelligent detection method in the above method embodiment, for example, to execute the above described Figure 1 The functions of method steps S100 to S800 in the method.
[0201] It will be appreciated by those skilled in the art that all or some of the steps and systems in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.
[0202] The above is a specific description of the preferred implementation of the embodiments of the present application, but the embodiments of the present application are not limited to the above-mentioned implementation methods. Technical personnel familiar with the field can also make various equivalent modifications or substitutions without violating the spirit of the embodiments of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the embodiments of the present application.
Claims
1. An intelligent detection method for construction technology of a distribution station area, characterized in that: The method comprises: Obtain the head-up view image and the upward view image of the distribution station area with matching view angles; wherein: An initial head-up perspective image and an initial upward perspective image of the distribution station area are collected through a preset semi-transparent mask and a preset angle range; Inputting the initial head-up perspective image into a trained perspective matching model for perspective matching to obtain a head-up perspective image of a distribution station area with a perspective matching, and inputting the initial upward perspective image into the trained perspective matching model for perspective matching to obtain an upward perspective image of a distribution station area with a perspective matching, wherein the trained perspective matching model is obtained by training multiple groups of head-up perspective images and upward perspective images with standard perspectives, and the standard perspective is a perspective that meets the preset angle range; Extracting a first feature set from the head-up perspective image of the distribution station area with the matching perspective, and extracting a second feature set from the upward perspective image of the distribution station area with the matching perspective; and performing feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs; specifically comprising: According to the preset semi-transparent mask, an outline area of the power distribution station area in the head-up perspective image and the upward perspective image is acquired; A feature extraction method is used to extract a first feature set of an outline area of the power distribution station area in the head-up perspective image of the power distribution station area with the perspective matching; A feature extraction method is used to extract a second feature set of the outer shape area of the power distribution station area in the upward viewing angle image of the power distribution station area with the matching viewing angle; Calculating the Euclidean distance between each first feature point in the first feature set and each second feature point in the second feature set to obtain a Euclidean distance matrix, wherein the Euclidean distance matrix is used to indicate feature matching pairs between each first feature point in the first feature set and each second feature point in the second feature set; Screening the feature matching pairs in the Euclidean distance matrix to obtain a plurality of screened feature matching pairs; Calculating a target perspective transformation matrix according to the plurality of feature matching pairs; specifically comprising: Randomly selecting a plurality of first feature matching pairs from the plurality of feature matching pairs, wherein the number of the first feature matching pairs is less than the number of the feature matching pairs; Calculating a perspective transformation matrix according to the plurality of first feature matching pairs; Calculating a reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix, wherein the second feature matching pair is used to indicate a feature matching pair in the plurality of feature matching pairs other than the plurality of first feature matching pairs; Sorting the reprojection errors corresponding to each second feature matching pair, and selecting a median value among the sorted reprojection errors; If the median is less than a preset error threshold, taking the perspective transformation matrices corresponding to the plurality of first feature matching pairs associated with the median as the target perspective transformation matrix; Detecting the position coordinates corresponding to each device in the upward viewing angle image; Based on the target perspective transformation matrix, the position coordinates of each device are mapped to a head-up viewing angle to obtain the transformed position coordinates of each device; Calculating a ratio value between an actual size and image pixels corresponding to a first device among the devices according to the transformed position coordinates corresponding to the first device, wherein the first device is a device with a standard size; Calculating the process parameters of each device according to the transformed position coordinates of each device and the ratio value; According to the process parameters of each device, the construction process of each device in the distribution station area is detected to obtain the construction process detection result of the distribution station area.
2. The intelligent detection method for construction technology of distribution station area according to claim 1 is characterized in that: The step of calculating the reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix includes: Projecting the coordinates of the second feature points in each second feature matching pair using the perspective transformation matrix to obtain the projection coordinates of the second feature points in each second feature matching pair; Acquire the coordinates of a first feature point that matches the second feature point; A reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated according to the projection coordinates of the second feature point and the coordinates of the first feature point matching the second feature point.
3. The intelligent detection method for construction technology of distribution station area according to claim 2 is characterized in that: The reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs is calculated in the following manner: ; in, Indicates The coordinates of the first feature point that matches the second feature point, Indicates The projection coordinates of the second feature point, represents the reprojection error corresponding to each second feature matching pair, Represents the number of second feature matching pairs.
4. The intelligent detection method for construction technology of distribution station area according to claim 1 is characterized in that: The calculating, according to the transformed position coordinates corresponding to the first device among the devices, a ratio value between the actual size corresponding to the first device and the image pixel, comprises: ; ; in, Indicates the ratio between the actual size of the first device and the image pixels. Indicates the actual size of the first device. Indicates the image pixel length corresponding to the first device, Indicates the transformed position coordinates corresponding to the upper left position of the first device coordinate, Indicates the transformed position coordinates corresponding to the lower right position of the first device coordinate.
5. An intelligent detection system for construction technology of distribution station area, characterized in that: The system comprises: An image acquisition unit is used to acquire a head-up view image and an upward view image of a distribution station area with matching view angles; wherein: An initial head-up perspective image and an initial upward perspective image of the distribution station area are collected through a preset semi-transparent mask and a preset angle range; Inputting the initial head-up perspective image into a trained perspective matching model for perspective matching to obtain a head-up perspective image of a distribution station area with a perspective matching, and inputting the initial upward perspective image into the trained perspective matching model for perspective matching to obtain an upward perspective image of a distribution station area with a perspective matching, wherein the trained perspective matching model is obtained by training multiple groups of head-up perspective images and upward perspective images with standard perspectives, and the standard perspective is a perspective that meets the preset angle range; A feature extraction unit is used to extract a first feature set from the head-up perspective image of the distribution station area of the matching perspective, and to extract a second feature set from the upward perspective image of the distribution station area of the matching perspective; and to perform feature matching on the first feature set and the second feature set to obtain a plurality of feature matching pairs; specifically comprising: According to the preset semi-transparent mask, an outline area of the power distribution station area in the head-up perspective image and the upward perspective image is acquired; A feature extraction method is used to extract a first feature set of an outline area of the power distribution station area in the head-up perspective image of the power distribution station area with the perspective matching; A feature extraction method is used to extract a second feature set of the outer shape area of the power distribution station area in the upward viewing angle image of the power distribution station area with the matching viewing angle; Calculating the Euclidean distance between each first feature point in the first feature set and each second feature point in the second feature set to obtain a Euclidean distance matrix, wherein the Euclidean distance matrix is used to indicate feature matching pairs between each first feature point in the first feature set and each second feature point in the second feature set; Screening the feature matching pairs in the Euclidean distance matrix to obtain a plurality of screened feature matching pairs; The first calculation unit is used to calculate the target perspective transformation matrix according to the multiple feature matching pairs; specifically comprising: Randomly selecting a plurality of first feature matching pairs from the plurality of feature matching pairs, wherein the number of the first feature matching pairs is less than the number of the feature matching pairs; Calculating a perspective transformation matrix according to the plurality of first feature matching pairs; Calculating a reprojection error corresponding to each second feature matching pair in the plurality of feature matching pairs according to the perspective transformation matrix, wherein the second feature matching pair is used to indicate a feature matching pair in the plurality of feature matching pairs other than the plurality of first feature matching pairs; Sorting the reprojection errors corresponding to each second feature matching pair, and selecting a median value among the sorted reprojection errors; If the median is less than a preset error threshold, taking the perspective transformation matrices corresponding to the plurality of first feature matching pairs associated with the median as the target perspective transformation matrix; A position detection unit, used to detect the position coordinates corresponding to each device in the upward viewing angle image; A coordinate mapping unit, used to map the position coordinates of each device to a head-up viewing angle based on the target perspective transformation matrix to obtain the transformed position coordinates of each device; A second calculation unit, configured to calculate a ratio value between an actual size and image pixels corresponding to a first device among the devices according to the transformed position coordinates corresponding to the first device, wherein the first device is a device with a standard size; A third calculation unit, used for calculating the process parameters of each device according to the transformed position coordinates of each device and the ratio value; The process detection unit is used to detect the construction process of each device in the distribution substation area according to the process parameters of each device, and obtain the construction process detection result of the distribution substation area.
6. An electronic device, characterized in that: It includes at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the intelligent detection method for construction technology of distribution station area as described in any one of claims 1 to 4.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the intelligent detection method for construction technology of a distribution station area as described in any one of claims 1 to 4.
