Box-type transformer inspection model establishment method

Abstract

The invention provides a box-type transformer inspection model establishment method, and relates to the technical field of data processing, and the method comprises the steps: controlling inspection equipment to carry out the information collection of target equipment, and obtaining the inspection data; performing element identification on the inspection data to obtain environment elements, performing multi-dimensional screening on the environment elements to obtain static reference elements, and generating an environment model based on the static reference elements; obtaining a position relationship between a target element corresponding to the target equipment and the static reference element, and obtaining a target orientation of the target element based on the position relationship and the reference orientation of the static reference element; performing model fusion on the standard box model corresponding to the target element and the environment model based on the target orientation to generate an inspection model; when it is determined that the image comparison result of the standard box-type model and the target element is inconsistent, the inspection model is updated according to the target element, an updated inspection model is obtained, efficient model construction can be conducted according to automatically collected data, and the accuracy of the inspection model is improved.

Description

Technical Field

[0001] This invention relates to data processing technology, and more particularly to a method for establishing an inspection model for a box-type transformer. Background Technology

[0002] As a critical node in modern power distribution networks, the stable operation of prefabricated transformers directly impacts the reliability and security of power supply. With the development of smart grids and unmanned inspection technologies, utilizing drones or inspection robots equipped with high-definition cameras to automatically collect data from prefabricated transformers and construct their 3D digital inspection models has become an important means of modernizing condition assessment, fault early warning, and operation and maintenance management. This digital inspection model can intuitively reproduce the equipment and its surrounding environment, providing remote and visualized decision support for maintenance personnel, greatly improving inspection efficiency and safety.

[0003] However, most existing methods for building inspection models of box-type transformers rely on directly reconstructing 3D from acquired images or point cloud data. This approach is highly susceptible to interference from complex on-site environments. For example, reflections caused by changes in lighting, dirt or occlusion on equipment surfaces can make it difficult for image recognition algorithms to accurately extract the complete outline of the target equipment, leading to model construction failure or a significant decrease in accuracy.

[0004] Therefore, how to efficiently build models based on automatically collected data and improve the accuracy of inspection models has become an urgent problem to be solved. Summary of the Invention

[0005] This invention provides a method for establishing an inspection model for a box-type transformer, which can efficiently construct the model based on automatically collected data and improve the accuracy of the inspection model.

[0006] A first aspect of the present invention provides a method for establishing an inspection model for a box-type transformer, comprising: The system receives inspection coordinates and controls the inspection equipment to collect information from the target equipment, thus obtaining inspection data. The inspection data is subjected to element identification to obtain environmental elements. The environmental elements are then filtered in multiple dimensions to obtain static reference elements. An environmental model is generated based on the static reference elements. Obtain the positional relationship between the target element and the static reference element corresponding to the target device, and obtain the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element; Based on the target orientation, the standard box model corresponding to the target element is fused with the environmental model to generate an inspection model; The standard box model is compared with the target element. If the comparison results are inconsistent, the inspection model is updated according to the target element to obtain the updated inspection model.

[0007] Optionally, in one possible implementation of the first aspect, the receiving inspection coordinate control inspection device collects information from the target device to obtain inspection data, including: Based on the inspection coordinates, the center point of the top surface contour of the target equipment is determined, and the initial top-view acquisition point of the target equipment is obtained according to the center point and the preset acquisition height. The inspection equipment is controlled to move to the initial overhead acquisition point. When it is determined that there is no complete target outline of the target equipment in the overhead data corresponding to the inspection equipment, the height of the inspection equipment is adjusted to obtain the reference acquisition height. Based on the aforementioned reference acquisition height, a top-down view of the target device is captured to obtain a top-down reference video. The side-view acquisition points of the inspection equipment are determined based on the top-view reference video, and the side-view reference video is obtained based on the side-view acquisition points. The inspection data includes the top-view reference video and the side-view reference video.

[0008] Optionally, in one possible implementation of the first aspect, determining the side-view acquisition points of the inspection equipment based on the top-view reference video, and obtaining the side-view reference video based on the side-view acquisition points, includes: Based on any apex angle of the target device's outline in the top-view reference video, control the inspection device to move to the corresponding apex angle of the outline, and determine the height outline line corresponding to the apex angle of the target device's outline. The side-view acquisition point is determined based on the preset side-view distance and the height contour line. When it is determined that there is no complete boundary contour of the target device in the side-view data at the side-view acquisition point, the spacing of the inspection equipment is adjusted to obtain the reference acquisition spacing. The target device is photographed from the side based on the reference acquisition interval to obtain a side-view reference video.

[0009] Optionally, in one possible implementation of the first aspect, the step of identifying elements in the inspection data to obtain environmental elements, and then performing multi-dimensional filtering on the environmental elements to obtain static reference elements, includes: The inspection data is subjected to element identification. The image element corresponding to the target device is taken as the target element, and the remaining image elements in the inspection data other than the target element are taken as environmental elements. The environmental elements are statistically analyzed to obtain an environmental element set. Based on a preset judgment time, the inspection data is statically identified to obtain the first reference element. The physical parameters of the first reference element are obtained from the inspection images in the inspection data. The first reference element is then filtered based on the physical parameters to obtain static reference elements. The physical parameters include physical height, physical width, and physical length.

[0010] Optionally, in one possible implementation of the first aspect, it also includes: When it is determined that there is no static reference element in the environmental data, the reference acquisition height and reference acquisition spacing are increased and adjusted to obtain the current inspection data. Based on the current inspection data, repeat the steps of determining the static reference element until the environmental elements in the current inspection data contain a static reference element.

[0011] Optionally, in one possible implementation of the first aspect, generating the environment model based on the static reference elements includes: A model is constructed based on the environmental elements to obtain an environmental sub-model, and the relative positional relationship between each environmental element is obtained based on the inspection data. Based on the relative positional relationship, the environmental sub-models are located to obtain the environmental model.

[0012] Optionally, in one possible implementation of the first aspect, the step of statically identifying the inspection data based on a preset judgment time to obtain the first reference element includes: Based on a preset judgment duration, video segments are extracted from the inspection data to obtain comparison video segments; Based on the set of environmental elements, the contours of environmental elements in each video frame of the comparison video segment are identified sequentially to obtain the element contours of each environmental element corresponding to each video frame. Obtain the contour coordinates of the pixels corresponding to the element contour to obtain a set of multiple contour coordinates corresponding to each video frame; The contour coordinate sets corresponding to the same environmental element in each video frame are compared to obtain the comparison results. Based on the comparison results, the first reference element is determined.

[0013] Optionally, in one possible implementation of the first aspect, comparing the set of contour coordinates corresponding to the same environmental element in each video frame to obtain a comparison result, and determining the first reference element based on the comparison result, includes: When the contour coordinate sets are determined to be the same, a consistent comparison result is generated; when the contour coordinate sets are determined to be different, an inconsistent comparison result is generated. The comparison result includes consistent comparison results and inconsistent comparison results. Based on the inconsistency results, the corresponding environmental elements are treated as non-reference elements; Once the comparison results are confirmed to be consistent, the corresponding environmental element is used as the first reference element.

[0014] Optionally, in one possible implementation of the first aspect, obtaining the physical parameters of the first reference element based on the inspection image in the inspection data includes: According to the target model of the target device, retrieve the corresponding physical parameters, and based on the top view image of the top view reference video, obtain the pixel length of the target device and the outline length and outline width of the first reference element; Based on the physical length of the target device and the pixel length, the conversion ratio is obtained, and the physical length and physical width corresponding to the first reference element are determined according to the contour length, contour width and conversion ratio. Based on the side view image of the side view reference video, the contour heights corresponding to the target device and the first reference element are obtained respectively; Based on the top view image, the interval distance between each first reference element and the target element is determined, and the first distance between the inspection equipment and the first reference element is determined according to the interval distance and the reference acquisition spacing. The physical height of the first reference element is obtained based on the first distance, the contour height and physical height of the target device, and the contour height of the first reference element.

[0015] Optionally, in one possible implementation of the first aspect, the step of filtering the first reference elements based on the physical parameters to obtain static reference elements includes: The physical length in the physical parameters of the first reference element is compared with the length threshold to obtain the first comparison result; The physical height is compared with the height threshold to obtain the second comparison result, and the physical width is compared with the width threshold to obtain the third comparison result; When any comparison result is determined to be greater than a preset threshold, the corresponding first reference element is used as a static reference element. The preset threshold includes a length threshold, a height threshold, and a width threshold.

[0016] Optionally, in one possible implementation of the first aspect, obtaining the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element includes: The positional relationship between the target element and each static reference element is determined based on the inspection data; Compare the physical length and physical width of the static reference elements in the top view image, and select the orientation of the boundary contour corresponding to the maximum value as the reference orientation of the corresponding static reference element. Based on the reference orientation, the corresponding boundary contour is determined as the reference contour, and the long side contour corresponding to the target element in the top view image is used as the comparison contour. The boundary angle is obtained based on the reference contour and the comparison contour, and the target orientation corresponding to the target element is obtained based on the reference orientation and the boundary angle.

[0017] Optionally, in one possible implementation of the first aspect, the step of fusing the standard box model corresponding to the target element with the environment model based on the target orientation to generate an inspection model includes: Based on the target model, a model is selected from the preset model library to obtain a standard box model. The standard box model is then positioned in the environmental model to obtain the initial inspection model. The standard box model in the initial inspection model is adjusted according to the target orientation to obtain the inspection model.

[0018] Optionally, in one possible implementation of the first aspect, the step of comparing the standard box model with the target element and determining that the comparison results are inconsistent, and then updating the inspection model according to the target element to obtain an updated inspection model, includes: A standard box-shaped image is obtained based on a standard box-shaped model. Image recognition is performed on the target element, and the image elements in the target element that are connected to the standard box-shaped image are taken as connecting elements. A connection model is generated based on the connection elements, and the standard box model in the inspection model is updated based on the connection model to obtain the updated inspection model.

[0019] A second aspect of the present invention provides an electronic device comprising: a memory, a processor, and a computer program, the computer program being stored in the memory, and the processor executing the computer program to perform the methods described in the first aspect of the present invention and various possible methods related to the first aspect.

[0020] A third aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, is used to implement the first aspect of the present invention and various methods possibly involved in the first aspect.

[0021] The beneficial effects of this invention are as follows: 1. This invention enables efficient model building based on automatically collected data, improving the accuracy of inspection models. Firstly, this invention controls inspection equipment to collect data from multiple perspectives. Specifically, it determines the center point of the transformer's upper contour based on inspection coordinates, locks the initial top-view point using a preset acquisition height, and then dynamically adjusts the height to a reference acquisition height based on contour integrity. This ensures the top-view reference video includes the complete top contour, effectively overcoming the problem of inaccurate initial parameters caused by ground undulations or equipment model differences, ensuring the integrity and clarity of the top-view data. For side-view acquisition, the top angle of the top-view contour is used as a reference, and the initial point is located using a preset side-view distance. The spacing is adjusted to a reference acquisition spacing based on boundary contour integrity, avoiding missing side structure images and reducing the risk of data acquisition failure due to excessively close or distant distances, thereby improving the accuracy of subsequent model building.

[0022] 2. This invention can screen out stable and reliable static reference elements to improve the accuracy of target device positioning and orientation in three-dimensional space. Specifically, this invention can first separate the target element from the background environmental elements through element recognition, and then perform static recognition using the time dimension of the video. By comparing the contour coordinate sets of the same element in different video frames of the extracted comparison video segments, this invention can systematically filter out dynamic interference such as pedestrians and vehicles to screen out fixed and distinctive first reference elements. Secondly, this invention can further screen the first reference elements by calculating the corresponding physical parameters to obtain static reference elements. Finally, using geometric calculation methods, the target orientation of the target element is determined based on the boundary angle between the reference orientation of the static reference element and the comparison contour of the target element, thereby improving the accuracy of the target device's positioning and orientation in space and facilitating the improvement of the accuracy of subsequent inspection models.

[0023] 3. This invention can retrieve standard box-type models from a preset model library based on the identified target type, ensuring the standardization and accuracy of the main structure of the model and avoiding the deformation that may be caused by direct image reconstruction. Secondly, the standard model is initially positioned in the pre-constructed environmental model to form an initial inspection model, and the orientation of the standard box-type model is adjusted using the determined target orientation, so that the placement of the model in the 3D twin environment is consistent with the actual placement. Furthermore, this invention can also compare the rendered image of the standard model with the real-world image. When the comparison results are found to be inconsistent, the invention automatically identifies the connecting elements that exist in the real image but not in the standard model, generates connecting models for these connecting elements, and integrates them into the inspection model to obtain an updated inspection model that can truly reflect the current actual state of the equipment, thereby improving the realism and accuracy of the inspection model. Attached Figure Description

[0024] Figure 1A flowchart of a method for establishing an inspection model for a box-type transformer provided by the present invention; Figure 2 This is a schematic diagram of the target outline in a top view image provided by the present invention; Figure 3 A schematic diagram of a height profile provided by the present invention; Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided by the present invention. Detailed Implementation

[0025] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0026] like Figure 1 As shown, this invention provides a flowchart of a method for establishing an inspection model for a box-type transformer. This method includes: S1 receives the inspection coordinates and controls the inspection equipment to collect information from the target equipment and obtain inspection data.

[0027] It should be noted that the construction of the model is inseparable from the original data. Therefore, in order to quickly and accurately construct the inspection model corresponding to the box-type transformer, the inspection equipment can be automatically moved to the location of the corresponding box-type transformer according to the inspection coordinates to collect the original data, so that the inspection model can be automatically generated based on the collected inspection data.

[0028] Among them, the inspection coordinates are the coordinates of the center point of the target device in the geographic information, the inspection device is the device that collects data from the target device, such as a drone, the target device is the device for model building, such as the box transformer corresponding to the inspection coordinates, and the inspection data is the raw information collected by the inspection device, such as video data of the target device and its surrounding environment.

[0029] Through the above implementation methods, the present invention can automatically collect data based on the received inspection coordinates to obtain raw data from multiple perspectives, thereby improving data collection efficiency for subsequent model building.

[0030] In some embodiments, the specific implementation steps of step S1 (where the receiving inspection coordinates control the inspection device to collect information from the target device and obtain inspection data) include: S11, determine the center point of the top surface contour of the target device based on the inspection coordinates, and obtain the initial top view acquisition point of the target device according to the center point and the preset acquisition height.

[0031] Understandably, shooting from directly above the target device can reduce the degree of positional shift of objects caused by the angle. Therefore, directly collecting data from above the target device can capture a complete and detailed layout of the top of the device, which can be used to plan the location of side-view shooting points later.

[0032] Among them, the top surface contour is the top surface area contour corresponding to the target device, the initial top-view acquisition point is the theoretical position of the inspection device when performing the first top-view shooting in three-dimensional space, which is determined by the horizontal coordinates (center point) and the vertical coordinates (preset height), and the preset acquisition height is the pre-set height distance from the target device.

[0033] S12, control the inspection equipment to move to the initial overhead acquisition point, and when it is determined that there is no complete target outline of the target equipment in the overhead data corresponding to the inspection equipment, adjust the height of the inspection equipment to obtain the reference acquisition height.

[0034] It is understandable that the preset acquisition height may be incompatible due to ground undulations or differences in equipment models, which may lead to problems such as not being able to capture the complete equipment or the equipment being too small, resulting in unclear details in the captured images. Therefore, the flight altitude can be automatically adjusted based on the images captured in real time by the inspection equipment. That is, the flight altitude can be automatically adjusted based on whether the image contains the complete outline of the target until a reference acquisition height that can clearly and completely capture the target equipment is found.

[0035] For example, such as Figure 2 The image shown is a schematic diagram of a frame of a top-view reference video provided by the present invention. It includes a target element, environment element 1, environment element 2 and environment element 3. The target outline is the closed outline formed by the edge of the target device in the captured image. The reference acquisition height is the final top-view shooting height that can capture a complete and clear outline of the target device after automatic adjustment.

[0036] Through the above-described embodiments, the present invention can adaptively adjust and automatically overcome problems caused by inaccurate initial parameters or changes in the field environment, thereby improving the accuracy of the collected overhead data.

[0037] S13, Based on the reference acquisition height, take a top-down view of the target device to obtain a top-down reference video.

[0038] Understandably, video contains more information than a single photograph, which facilitates subsequent identification of static elements and exclusion of dynamic objects.

[0039] Among them, the top-down reference video is a continuous video clip containing the target device, taken from directly above at the reference acquisition height.

[0040] S14, determine the side view acquisition points of the inspection equipment based on the top view reference video, and obtain the side view reference video based on the side view acquisition points. The inspection data includes the top view reference video and the side view reference video.

[0041] Understandably, top-view data alone cannot obtain the device's height information and side details, nor can it accurately calculate the height of objects in the environment. In order to build a 3D model, side-view data needs to be introduced. This step uses the clear device outline (especially the top corner) in the top view to automatically plan the position for side-view shooting, ensuring that the full side view of the device can be captured.

[0042] Among them, the side view acquisition point is the spatial location of the inspection equipment when it takes side view shots. It is usually located on the side of the target equipment and at a certain distance from the equipment. The side view reference video is a video clip taken from the side view acquisition point that includes the side of the target equipment. The inspection data is the final raw data set, which is composed of the top view reference video and the side view reference video, forming a multi-view data.

[0043] In some embodiments, the specific implementation steps of step S14 (determining the side-view acquisition points of the inspection equipment based on the top-view reference video, and obtaining the side-view reference video based on the side-view acquisition points) include: S141, based on any contour apex angle of the target device in the top-view reference video, control the inspection device to move to the corresponding contour apex angle, and determine the height contour line corresponding to the corresponding contour apex angle of the target device.

[0044] Understandably, in order to determine as much of the location and orientation of the box-type transformer as possible, it is advisable to collect data at the side corner of the target equipment so as to capture as much data as possible about the different sides and side environment of the transformer.

[0045] Here, the apex angle of the contour is the corner point of the top contour of the target contour in the top-view reference video, for example, as shown in the image. Figure 3 The image shown is a side view image corresponding to a side view reference video provided by the present invention, which is a top view. Figure 2 The diagram shows the corresponding side view of the element. The height contour line is the height boundary contour of the target device that connects to the top corner of the contour.

[0046] It is easy to understand that the apex is the most distinct and easily identifiable geometric feature point on the device's outline, which facilitates the system's positioning and navigation. By moving to the position corresponding to the apex, it is easy to determine the horizontal orientation for side-view shooting and prepare to capture the height outline line under that orientation, that is, the vertical outline line of the device's side.

[0047] S142, determine the side-view acquisition point based on the preset side-view distance and the height contour line. When it is determined that there is no complete boundary contour of the target device in the side-view data at the side-view acquisition point, adjust the spacing of the inspection equipment to obtain the reference acquisition spacing.

[0048] Understandably, if the distance is too close, it may be impossible to capture the complete side of the device; if it is too far, the device will be too small in the image and the details will be unclear. Therefore, you can determine whether to move closer or further away by judging whether the image contains a complete boundary outline, until you find a reference acquisition distance that can clearly and completely capture the side of the device.

[0049] Among them, the preset side view distance is the distance value of the target device that is pre-set for side view acquisition, the side view acquisition point is the point used for side view shooting, that is, the horizontal distance of the point from the height contour line reference acquisition distance, the complete boundary contour is the closed contour formed by the other two height contour lines adjacent to the selected height contour line and the corresponding boundary contours above and below the target device in the side view image, and the reference acquisition distance is the final side view shooting distance that can capture a complete and clear side contour of the target device after automatic adjustment.

[0050] It is easy to understand that the present invention can automatically adjust to the optimal shooting distance according to the actual size of the target device, and through closed-loop control, ensure that the final side-view reference video is valid and usable, avoiding data acquisition failure due to improper distance.

[0051] S143, Based on the reference acquisition spacing, a side-view image of the target device is captured to obtain a side-view reference video.

[0052] Understandably, the side-view reference video is video data that contains complete side-view information of the target device, ultimately captured at optimized side-view acquisition points.

[0053] S2, perform element identification on the inspection data to obtain environmental elements, perform multi-dimensional filtering on the environmental elements to obtain static reference elements, and generate an environmental model based on the static reference elements.

[0054] Understandably, in existing technologies, when constructing inspection models for box-type transformers, the model is usually generated directly from the collected images. However, if the collected images are damaged due to factors such as light reflection or surface damage to the transformer, the corresponding transformer image cannot be accurately identified, affecting the generation of the transformer model. Therefore, this invention can determine the corresponding box-type transformer model based on the collected images and then retrieve it from the model database, improving the accuracy and speed of model construction. Thus, when the standard box-type model is retrieved and placed within a model constructed from surrounding environmental elements, the orientation of the standard box-type model needs to be determined. Furthermore, fixed and relatively large elements can be selected from the collected inspection data as reference elements to determine the orientation of the standard box-type model, thereby improving the accuracy of the constructed inspection model.

[0055] Among them, environmental elements are all the independent image elements identified in the inspection data other than the target equipment, such as building walls, trees, utility poles, grass, road shows, etc. Static reference elements are objects selected from environmental elements after multi-dimensional screening. These objects have fixed positions in space, are relatively large in size, and can serve as stable spatial references, such as building walls and cement bases. The environmental model is a three-dimensional digital scene model constructed based on the position, size, and interrelationship of static reference elements. It represents the stable environment in which the target equipment is located.

[0056] It is easy to understand that this invention can automatically filter out dynamic interference and small objects in the environment, which facilitates the improvement of the accuracy of placing the standard box model in the environmental model. At the same time, the generated environmental model provides a reliable coordinate system and reference for determining the position and orientation of the target equipment, which facilitates the improvement of the accuracy of the inspection model.

[0057] In some embodiments, the specific implementation steps of step S2 (identifying elements in the inspection data to obtain environmental elements, and performing multi-dimensional filtering on the environmental elements to obtain static reference elements) include: S21, perform element identification on the inspection data, take the image element corresponding to the target device as the target element, and take the remaining image elements in the inspection data other than the target element as environmental elements.

[0058] It is understandable that the target element is the image element in the image that corresponds to the target device.

[0059] It is easy to understand that by collecting video image frames, the corresponding elements in each video frame are distinguished, which avoids the target elements being incorrectly counted when the subsequent statistical environment is filtered, thereby improving the accuracy of the model.

[0060] S22, statistically analyze the environmental elements to obtain an environmental element set, and perform static identification on the inspection data based on a preset judgment time to obtain a first reference element.

[0061] It is understandable that dynamic environmental elements are not easy to determine the orientation of the target device in the future. Therefore, when filtering environmental elements, dynamic environmental elements can be excluded in order to save the amount of data processing required for subsequent filtering of reference elements and improve the speed of model building.

[0062] The environmental element set is the set of all identified environmental elements. The preset judgment time is a pre-set comparison time, such as 15 seconds. The first reference element is an environmental element that has been identified as having no change in position and shape within the preset judgment time after static recognition, such as the wall of a building.

[0063] Through the above implementation methods, the present invention distinguishes between static and dynamic objects using an objective and automated method, effectively eliminating interference items such as pedestrians, vehicles, and tree branches swaying in the wind, ensuring that the selected reference elements are fixed objects in the real world, thereby guaranteeing the long-term stability of the environmental model built based on them.

[0064] In some embodiments, the specific implementation steps of step S22 (the static identification of inspection data based on a preset judgment time to obtain the first reference element) include: S221, based on a preset judgment duration, video segments are extracted from the inspection data to obtain comparison video segments.

[0065] Understandably, the video segments being compared are obtained by extracting a pre-defined judgment duration from the original inspection data.

[0066] It's easy to understand that when cropping video, both top-down and side-view reference videos are cropped to facilitate subsequent element identification.

[0067] S222, based on the set of environmental elements, the contours of environmental elements in each video frame of the comparison video segment are identified sequentially to obtain the element contours of each environmental element corresponding to each video frame.

[0068] Understandably, in order to identify static elements from the set of environmental elements, the comparison video segment can be split into multiple video frames, and the outlines of elements in the set of environmental elements appearing in the video frames can be obtained. This allows for subsequent positional comparison of the outlines of the same environmental element in multiple video frames to determine the first reference element.

[0069] Among them, the element outline is the edge outline of the environment element in the video frame image.

[0070] S223, obtain the contour coordinates of the pixel points corresponding to the element contour, and obtain a set of multiple contour coordinates corresponding to each video frame.

[0071] It is understandable that contour coordinates are the coordinates of the contour pixels corresponding to the element contour, and the contour coordinate set is the set of contour coordinates of a single element contour in a video frame, that is, the set used to describe the coordinates of the pixels around the element contour.

[0072] S224, compare the set of contour coordinates corresponding to the same environmental element in each video frame to obtain the comparison result, and determine the first reference element based on the comparison result.

[0073] It is understandable that by comparing the set of contour coordinates corresponding to the same environmental element in different video frames, the dynamic and static state of the corresponding environmental element can be determined. When the set of contour coordinates corresponding to the environmental element is the same in the video frames within the preset judgment time, it can be said that the current element is a static element. Conversely, it can be said that the corresponding environmental element will change position. Therefore, the first reference element can be determined based on the comparison results.

[0074] The comparison result is obtained by comparing the contour coordinates of the same element in different video frames.

[0075] In some embodiments, the specific implementation steps of step S224 (comparing the contour coordinate sets corresponding to the same environmental element in each video frame to obtain a comparison result, and based on the comparison result) include: S2241, when the contour coordinate sets are determined to be the same, a consistent comparison result is generated; when the contour coordinate sets are determined to be different, an inconsistent comparison result is generated. The comparison result includes consistent comparison results and inconsistent comparison results.

[0076] It is understandable that if the set of contour coordinates corresponding to each video frame in the comparison video segment is the same, it can be said that the environment element is static, and therefore the comparison result is consistent. Conversely, if the set of contour coordinates corresponding to the same environment element is different in different video frames, it means that the corresponding environment element has shifted its position, and the comparison result is inconsistent.

[0077] S2242, based on the inconsistency results, the corresponding environmental element is treated as a non-reference element.

[0078] It is understandable that non-reference elements are the environmental elements corresponding to inconsistent alignment results.

[0079] S2243, determine the matching result and use the corresponding environmental element as the first reference element.

[0080] Understandably, the first reference element is the environment element corresponding to the matching result.

[0081] S23, obtain the physical parameters of the first reference element based on the inspection image in the inspection data, and filter the first reference element based on the physical parameters to obtain a static reference element. The physical parameters include physical height, physical width and physical length.

[0082] Understandably, static is only one dimension of the selection criteria for determining reference elements. Although a static blade of grass or a stone has not changed in position, it is not very meaningful as a spatial reference. A stable object with a sufficiently large size and obvious features is needed. Therefore, the actual physical size of the first reference element can be obtained for secondary selection to obtain a static reference element that is actually referable, thereby improving the accuracy of determining the orientation of the target device.

[0083] Among them, the physical parameters are the three-dimensional dimensions of the object in the actual environment, including physical length, physical width and physical height. The static reference elements are the environmental reference objects that are finally determined after two rounds of screening, namely static identification and size screening. They are both static and have a sufficiently large size.

[0084] In some embodiments, the specific implementation steps of step S23 (obtaining the physical parameters of the first reference element based on the inspection image in the inspection data) include: S231, retrieve the corresponding physical parameters according to the target model of the target device, and obtain the pixel length of the target device and the outline length and outline width of the first reference element based on the top view image of the top view reference video.

[0085] It is understandable that the inspection data can determine the model of the target device at the corresponding inspection coordinates, and then determine the corresponding physical parameters. Contour recognition can be performed on the top view image to determine the length of the target device in the image and the contour length of the first reference element in the image. This is then used to obtain the conversion ratio based on the ratio of the physical length and pixel length of the target device. Finally, the actual physical size of the first reference element can be obtained by combining the size of the first reference element in the image with the conversion ratio.

[0086] Wherein, the target model is the device model corresponding to the target device, the top view image is the video frame image in the video reference video, the pixel length is the length of the long side of the target device in the top view image, the outline length is the length of the long side of the first reference element in the top view image, and the outline width is the length of the wide side of the first reference element in the top view image.

[0087] S232, based on the physical length of the target device and the pixel length, a conversion ratio is obtained, and the physical length and physical width corresponding to the first reference element are determined according to the contour length, contour width and conversion ratio.

[0088] It is understandable that the conversion ratio is the ratio of the actual length of the object to the corresponding pixel length in the image, that is, the ratio of the physical length to the pixel length.

[0089] Specifically, the physical length of the first reference element is obtained by multiplying the conversion ratio and the contour length, and the physical width of the first reference element is obtained by multiplying the conversion ratio and the contour width.

[0090] S233, based on the side view image of the side view reference video, obtain the contour heights corresponding to the target device and the first reference element respectively.

[0091] It is understandable that the contour height is the element height corresponding to the target device and the first reference element in the side view image, that is, the difference between the minimum and maximum values ​​of the corresponding vertical coordinates of the corresponding elements is identified to obtain the contour height.

[0092] S234, Based on the top view image, determine the interval distance between each first reference element and the target element, and determine the first distance between the inspection equipment and the first reference element according to the interval distance and the reference acquisition spacing.

[0093] Understandably, the distance between the center point of each first reference element and the center point of the target element can be obtained from the top-view image. This distance is then multiplied by the conversion ratio to calculate the corresponding time interval distance. Next, the first distance between the inspection equipment and the first reference element is determined based on the interval distance and the reference acquisition distance between the inspection equipment and the target equipment. This allows the actual physical height of the first reference element to be determined based on the first distance, reducing image height errors caused by side-view shooting.

[0094] Wherein, the interval distance is the interval length between the first reference element and the target element, and the first distance is the time interval distance between the inspection equipment and the first reference element.

[0095] S235, based on the first distance, the contour height and physical height corresponding to the target device, and the contour height corresponding to the first reference element, the physical height corresponding to the first reference element is obtained.

[0096] Understandably, based on mathematical calculations, the ratio of physical height to contour height can be calculated to obtain the transformation ratio corresponding to the side view image. In turn, the contour height corresponding to the first reference element can be obtained, and the physical height of the first reference element can be calculated based on the focal length parameters captured by the inspection equipment. This reduces the error caused by the acquisition of images with larger objects at near distances and smaller objects at far distances, and improves the accuracy of calculating the physical parameters corresponding to the first reference element.

[0097] In some embodiments, the specific implementation steps of step S23 (the step of filtering the first reference element based on the physical parameters to obtain the static reference element) include: S236, compare the physical length in the physical parameters of the first reference element with the length threshold to obtain the first comparison result.

[0098] Understandably, the first comparison result is the result of comparing the length of the first reference element.

[0099] S237, compare the physical height with the height threshold to obtain the second comparison result, and compare the physical width with the width threshold to obtain the third comparison result.

[0100] Understandably, the second comparison result is the result of comparing the height parameter corresponding to the first reference element, and the third comparison result is the result of comparing the width of the first reference element.

[0101] S238, when any comparison result is determined to be greater than a preset threshold, the corresponding first reference element is used as a static reference element. The preset threshold includes a length threshold, a height threshold, and a width threshold.

[0102] Understandably, by comparing the calculated physical parameters (length, width, height) with preset thresholds, a decision is made to retain or discard the object. If any dimension exceeds the threshold, the object is considered large enough to be used as a static reference element.

[0103] In some embodiments, it also includes: A1, when it is determined that there is no static reference element in the environmental data, the reference acquisition height and reference acquisition spacing are increased and adjusted to obtain the current inspection data.

[0104] Understandably, when data is captured based on the initially adjusted baseline acquisition height and spacing, if no suitable static reference elements are found after element filtering of the captured data, it indicates that there are few elements within the current acquisition environment, which will affect the accuracy of determining the orientation of the target device. Therefore, it is necessary to expand and adjust the baseline acquisition height and spacing to obtain new acquisition points for data acquisition and to obtain the current inspection data. This facilitates subsequent element filtering of the current inspection data to obtain static reference elements, ensuring the accuracy of determining the orientation of the model corresponding to the target element and improving the accuracy of the inspection model construction.

[0105] A2, based on the current inspection data, repeat the above steps of determining the static reference element until the environmental elements in the current inspection data contain a static reference element.

[0106] Understandably, once the spacing has been adjusted and the current inspection data is obtained, the above judgment steps can be repeated until static reference elements are selected, so as to determine the target orientation corresponding to the target element and facilitate the construction of the inspection model.

[0107] In some embodiments, the specific implementation steps of step S2 (generating the environment model based on the static reference elements) include: S24, construct a model based on the environmental elements to obtain an environmental sub-model, and obtain the relative positional relationship between each environmental element based on the inspection data.

[0108] Understandably, when a static reference element with a definite target orientation is identified in the collected inspection data, it indicates that the current inspection data is qualified and valid. This allows for the construction of a model for the environmental elements. Specifically, an independent environmental sub-model is first established for each element. Then, based on their relative positions in the image, these sub-models are positioned and stitched together in a unified coordinate system to form a complete scene model. This allows the standard model corresponding to the target device to be updated into the environmental model, generating an inspection model and improving the efficiency and accuracy of model construction.

[0109] Among them, the environment sub-model is the three-dimensional model corresponding to a single environment element, and the relative positional relationship is the orientation and distance relationship between each environment element in the real three-dimensional space.

[0110] S25, based on the relative positional relationship, perform model positioning on each of the environmental sub-models to obtain the environmental model.

[0111] Understandably, based on the determined relative positional relationships, the various environmental sub-models are positioned and spliced ​​together to obtain the environmental model. This ensures that the generated environmental model is accurate in spatial layout, so as to accurately locate the progress of the standard model corresponding to the target equipment and obtain the inspection model.

[0112] S3, obtain the positional relationship between the target element and the static reference element corresponding to the target device, and obtain the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element.

[0113] Understandably, once the environment model is constructed and the target elements are identified, it is necessary to determine the placement orientation of the model corresponding to the target elements within the environment model. For example, which side of the transformer is the front and the placement angle between it and other elements, etc., in order to perform model fusion and improve the accuracy of the inspection model.

[0114] Among them, the positional relationship refers to the relative orientation and distance between the target element and each static reference element, the reference orientation refers to the orientation of the static reference element itself in space, and the target orientation refers to the orientation of the target device in space.

[0115] Through the above implementation methods, the present invention can determine the spatial orientation of the target device and ensure that the orientation of the standard model finally integrated into the environment is consistent with that of the real world, so as to improve the accuracy of the inspection model.

[0116] In some embodiments, the specific implementation steps of step S3 (obtaining the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element) include: S31, determine the positional relationship between the target element and each static reference element based on the inspection data.

[0117] Understandably, the collected inspection data, namely top-down and side-view reference videos, can identify the relative positional relationship between each static reference element and the target element, so as to determine the location and orientation of the target equipment in the environmental model.

[0118] S32, compare the physical length and physical width corresponding to the static reference elements in the top view image, and select the orientation of the boundary contour corresponding to the maximum value as the reference orientation of the corresponding static reference element.

[0119] It is understandable that, in order to facilitate the determination of the target orientation of the target element based on the reference orientation of the static reference element, the extended orientation of the longer side of the element can be used as the reference orientation of the corresponding static reference element based on the top view image.

[0120] S33, based on the reference orientation, determine the corresponding boundary contour as the reference contour, and use the long side contour corresponding to the target element in the top view image as the comparison contour.

[0121] It is understandable that the reference contour is the boundary contour of the long side corresponding to the reference orientation in the static reference element, and the comparison contour is the long side contour corresponding to the target element.

[0122] Through the above implementation methods, the present invention can obtain the reference contour corresponding to the static reference element and the comparison contour corresponding to the target element, so that the corresponding boundary contour can be extended to determine the corresponding angular relationship, thereby facilitating the determination of the target orientation of the target element.

[0123] S34, obtain the boundary angle based on the reference contour and the comparison contour, and obtain the target orientation corresponding to the target element based on the reference orientation and the boundary angle.

[0124] It is understandable that the boundary angle is the angle between the boundary line of the reference profile and the boundary line of the comparison profile.

[0125] It is easy to understand that by using the boundary angles between multiple static reference elements and the target element, the positioning accuracy of the model corresponding to the target element in the environmental model can be improved. Based on the target orientation, the tilt angle between the model corresponding to the target device and the boundary of each environmental sub-model can be determined, thereby improving the accuracy of the final generated inspection model.

[0126] S4. Based on the target orientation, the standard box model corresponding to the target element is fused with the environmental model to generate an inspection model.

[0127] Understandably, once the orientation of the target equipment is determined, the standard box model corresponding to the target element can be integrated into the environmental model to obtain a complete digital model that can be used for inspection.

[0128] In some embodiments, the specific implementation steps of step S4 (the model fusion of the standard box model corresponding to the target element and the environment model based on the target orientation to generate the inspection model) include: S41, Select a model from the preset model library based on the target model to obtain a standard box model, and position the standard box model in the environmental model to obtain the initial inspection model.

[0129] Understandably, the preset model library is a pre-set model of a large number of box-type transformer models, the standard box-type model is a pre-set standard model corresponding to the target model, and the initial inspection model is a 3D model that positions the standard box-type model at the missing position in the environmental model.

[0130] It is easy to understand that moving the standard box model to a specified position on the environmental model results in an initial inspection model with inaccurate orientation.

[0131] S42, adjust the orientation of the standard box model in the initial inspection model according to the target orientation to obtain the inspection model.

[0132] Understandably, based on the determined target orientation, the standard box model in the initial inspection model is rotated in orientation. This means positioning the boundary line of the model outline with the boundaries of the other environmental sub-models using the calculated orientation angle. For example, the standard box model is rotated so that its front faces the building wall. The rotation stops when the angle between the long side outline of the standard box model and the boundary line of the environmental sub-model in the initial inspection model is equal to the boundary angle collected in the inspection data. This yields the final inspection model.

[0133] S5, compare the standard box model with the target element in an image. If the comparison results are inconsistent, update the inspection model according to the target element to obtain an updated inspection model.

[0134] Understandably, when the box-type transformer at the actual inspection coordinates is modified, for example, by adding a canopy to the box-type transformer, the standard box-type model in the inspection model can be updated in order to make the inspection model more consistent with the actual situation.

[0135] It is easy to understand that when the rendered image of the standard model is superimposed and compared with the real-shot image, if the real image contains parts that are not in the standard model, such as the canopy, the comparison results will be inconsistent. This allows the inspection model to be updated. The updated inspection model is the 3D inspection model after the inspection model has been updated.

[0136] In some embodiments, the specific implementation steps of step S5 (comparing the standard box model with the target element in an image, and updating the inspection model according to the target element when the comparison result is inconsistent, to obtain an updated inspection model) include: S51, a standard box-shaped image is obtained based on the standard box-shaped model, and image recognition is performed on the target element. The image elements in the target element that are connected to the standard box-shaped image are taken as connecting elements.

[0137] It is understandable that the standard box image is the model image corresponding to the standard box model, and the connecting elements are those parts or accessories that are attached to the target device in the real image but are not included in the standard box model.

[0138] S52, Generate a connection model based on the connection elements, and update the standard box model in the inspection model based on the connection model to obtain an updated inspection model.

[0139] It is understandable that the link model is the 3D model corresponding to the link element. The generated link model is used to update the inspection model to obtain the updated inspection model, so that the constructed updated inspection model is more accurate and the integrity of the modeling is improved.

[0140] See Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. The electronic device 40 includes: a processor 41, a memory 42, and a computer program; wherein... The memory 42 is used to store the computer program, and the memory may also be flash memory. The computer program is, for example, an application program or functional module that implements the above method.

[0141] The processor 41 is configured to execute the computer program stored in the memory to implement the various steps performed by the device in the above method. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0142] Alternatively, the memory 42 can be either standalone or integrated with the processor 41.

[0143] When the memory 42 is a device independent of the processor 41, the device may further include: Bus 43 is used to connect the memory 42 and the processor 41.

[0144] The present invention also provides a readable storage medium storing a computer program, which, when executed by a processor, is used to implement the methods provided in the various embodiments described above.

[0145] The readable storage medium can be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of computer programs from one location to another. A computer storage medium can be any available medium accessible to a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application-Specific Integrated Circuit (ASIC). Alternatively, the ASIC can be located in a user equipment. Of course, the processor and the readable storage medium can also exist as discrete components in a communication device. The readable storage medium can be a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0146] The present invention also provides a program product including executable instructions stored in a readable storage medium. At least one processor of the device can read the executable instructions from the readable storage medium, and the at least one processor executes the executable instructions to cause the device to implement the methods provided in the various embodiments described above.

[0147] In the embodiments of the above-described device, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for establishing an inspection model for a box-type transformer, characterized in that, include: The system receives inspection coordinates and controls the inspection equipment to collect information from the target equipment, thus obtaining inspection data. The inspection data is subjected to element identification to obtain environmental elements. The environmental elements are then filtered in multiple dimensions to obtain static reference elements. An environmental model is generated based on the static reference elements. Obtain the positional relationship between the target element and the static reference element corresponding to the target device, and obtain the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element; Based on the target orientation, the standard box model corresponding to the target element is fused with the environmental model to generate an inspection model; The standard box model is compared with the target element. If the comparison results are inconsistent, the inspection model is updated according to the target element to obtain the updated inspection model.

2. The method according to claim 1, characterized in that, The receiving inspection coordinate control inspection device collects information from the target device to obtain inspection data, including: Based on the inspection coordinates, the center point of the top surface contour of the target equipment is determined, and the initial top-view acquisition point of the target equipment is obtained according to the center point and the preset acquisition height. The inspection equipment is controlled to move to the initial overhead acquisition point. When it is determined that there is no complete target outline of the target equipment in the overhead data corresponding to the inspection equipment, the height of the inspection equipment is adjusted to obtain the reference acquisition height. Based on the aforementioned reference acquisition height, a top-down view of the target device is captured to obtain a top-down reference video. The side-view acquisition points of the inspection equipment are determined based on the top-view reference video, and the side-view reference video is obtained based on the side-view acquisition points. The inspection data includes the top-view reference video and the side-view reference video.

3. The method according to claim 2, characterized in that, The step of determining the side-view acquisition points of the inspection equipment based on the top-view reference video, and obtaining the side-view reference video based on the side-view acquisition points, includes: Based on any apex angle of the target device's outline in the top-view reference video, control the inspection device to move to the corresponding apex angle of the outline, and determine the height outline line corresponding to the apex angle of the target device's outline. The side-view acquisition point is determined based on the preset side-view distance and the height contour line. When it is determined that there is no complete boundary contour of the target device in the side-view data at the side-view acquisition point, the spacing of the inspection equipment is adjusted to obtain the reference acquisition spacing. The target device is photographed from the side based on the reference acquisition interval to obtain a side-view reference video.

4. The method according to claim 3, characterized in that, The process involves identifying elements from the inspection data to obtain environmental elements, and then performing multi-dimensional filtering on these environmental elements to obtain static reference elements, including: The inspection data is subjected to element identification. The image element corresponding to the target device is taken as the target element, and the remaining image elements in the inspection data other than the target element are taken as environmental elements. The environmental elements are statistically analyzed to obtain an environmental element set. Based on a preset judgment time, the inspection data is statically identified to obtain the first reference element. The physical parameters of the first reference element are obtained from the inspection images in the inspection data. The first reference element is then filtered based on the physical parameters to obtain static reference elements. The physical parameters include physical height, physical width, and physical length.

5. The method according to claim 4, characterized in that, Also includes: When it is determined that there is no static reference element in the environmental data, the reference acquisition height and reference acquisition spacing are increased and adjusted to obtain the current inspection data. Based on the current inspection data, repeat the steps of determining the static reference element until the environmental elements in the current inspection data contain a static reference element.

6. The method according to claim 4, characterized in that, The generation of the environment model based on the static reference elements includes: A model is constructed based on the environmental elements to obtain an environmental sub-model, and the relative positional relationship between each environmental element is obtained based on the inspection data. Based on the relative positional relationship, the environmental sub-models are located to obtain the environmental model.

7. The method according to claim 4, characterized in that, The first reference element is obtained by statically identifying the inspection data based on a preset judgment time, including: Based on a preset judgment duration, video segments are extracted from the inspection data to obtain comparison video segments; Based on the set of environmental elements, the contours of environmental elements in each video frame of the comparison video segment are identified sequentially to obtain the element contours of each environmental element corresponding to each video frame. Obtain the contour coordinates of the pixels corresponding to the element contour to obtain a set of multiple contour coordinates corresponding to each video frame; The contour coordinate sets corresponding to the same environmental element in each video frame are compared to obtain the comparison results. Based on the comparison results, the first reference element is determined.

8. The method according to claim 7, characterized in that, The step of comparing the set of contour coordinates corresponding to the same environmental element in each video frame to obtain a comparison result, and determining the first reference element based on the comparison result, includes: When the contour coordinate sets are determined to be the same, a consistent comparison result is generated; when the contour coordinate sets are determined to be different, an inconsistent comparison result is generated. The comparison result includes consistent comparison results and inconsistent comparison results. Based on the inconsistency results, the corresponding environmental elements are treated as non-reference elements; Once the comparison results are confirmed to be consistent, the corresponding environmental element is used as the first reference element.

9. The method according to claim 8, characterized in that, The step of obtaining the physical parameters of the first reference element based on the inspection image in the inspection data includes: According to the target model of the target device, retrieve the corresponding physical parameters, and based on the top view image of the top view reference video, obtain the pixel length of the target device and the outline length and outline width of the first reference element; Based on the physical length of the target device and the pixel length, the conversion ratio is obtained, and the physical length and physical width corresponding to the first reference element are determined according to the contour length, contour width and conversion ratio. Based on the side view image of the side view reference video, the contour heights corresponding to the target device and the first reference element are obtained respectively; Based on the top view image, the interval distance between each first reference element and the target element is determined, and the first distance between the inspection equipment and the first reference element is determined according to the interval distance and the reference acquisition spacing. The physical height of the first reference element is obtained based on the first distance, the contour height and physical height of the target device, and the contour height of the first reference element.

10. The method according to claim 9, characterized in that, The process of filtering the first reference elements based on the physical parameters to obtain static reference elements includes: The physical length in the physical parameters of the first reference element is compared with the length threshold to obtain the first comparison result; The physical height is compared with the height threshold to obtain the second comparison result, and the physical width is compared with the width threshold to obtain the third comparison result; When any comparison result is determined to be greater than a preset threshold, the corresponding first reference element is used as a static reference element. The preset threshold includes a length threshold, a height threshold, and a width threshold.

11. The method according to claim 10, characterized in that, The process of obtaining the target orientation of the target element based on the positional relationship and the reference orientation of the static reference element includes: The positional relationship between the target element and each static reference element is determined based on the inspection data; Compare the physical length and physical width of the static reference elements in the top view image, and select the orientation of the boundary contour corresponding to the maximum value as the reference orientation of the corresponding static reference element. Based on the reference orientation, the corresponding boundary contour is determined as the reference contour, and the long side contour corresponding to the target element in the top view image is used as the comparison contour. The boundary angle is obtained based on the reference contour and the comparison contour, and the target orientation corresponding to the target element is obtained based on the reference orientation and the boundary angle.

12. The method according to claim 11, characterized in that, The step of fusing the standard box model corresponding to the target element with the environment model based on the target orientation to generate an inspection model includes: Based on the target model, a model is selected from the preset model library to obtain a standard box model. The standard box model is then positioned in the environmental model to obtain the initial inspection model. The standard box model in the initial inspection model is adjusted according to the target orientation to obtain the inspection model.

13. The method according to claim 12, characterized in that, The step of comparing the standard box model with the target element and determining that the comparison results are inconsistent, and then updating the inspection model based on the target element to obtain an updated inspection model, includes: A standard box-shaped image is obtained based on a standard box-shaped model. Image recognition is performed on the target element, and the image elements in the target element that are connected to the standard box-shaped image are taken as connecting elements. A connection model is generated based on the connection elements, and the standard box model in the inspection model is updated based on the connection model to obtain the updated inspection model.

Images