Visual inspection crack identification method and system for gallery
By building a three-dimensional image model of corridors and training related identification and evaluation models, the problems of poor correlation and single attributes of corridor crack identification in the existing technology are solved, efficient and accurate crack identification and evaluation are achieved, and safe operation level is improved.
Patent Information
- Application Number
- CN202411838116.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-13
- Publication Date
- 2025-03-28
AI Technical Summary
The prior art has poor correlation, single attributes, and fixed evaluation models in the identification of corridor cracks, and cannot provide effective evaluation.
Through the collection of corridor image data, an image database and corridor three-dimensional image model are constructed, the crack identification model and evaluation model are trained, the three-dimensional image model is updated in real time and the crack evaluation report is output.
Visual identification of cracks in corridor inspections has been realized, the needs of summary evaluation and joint evaluation have been met, the timeliness and accuracy of inspections have been improved, and the level of safe operation has been significantly improved.
Smart Images
Figure CN119723348A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method and a system for identifying cracks in corridor visual inspection, and belongs to the technical field of corridor detection. Background Art
[0002] During the construction of hydropower, mining and other industries, corridor construction and inspection operations will be involved. It is necessary to understand the distribution of damage such as cracks in the corridor structure, the location of leakage points and the operating status of key facilities. At the same time, regular structural safety such as whether there is any dislocation in the structural joints and whether the drainage ditches are running smoothly also need to be inspected in time to prevent problems before they occur.
[0003] Due to the limitations of manual inspection and the hidden nature of the development of structural diseases and damage, conventional manual inspections are not easy to timely and accurately discover and identify corridor defects and operating conditions, and it is difficult to provide timely and comprehensive support for structural safety management and control. The use of intelligent inspection methods to complete safety inspections of dam corridors, dam shoulder flat tunnels, and corridors at the bottom of cushion ponds is the basis for building a structural safety assessment method that integrates monitoring and detection. Its results are also an important basis for evaluating the safety status of hydropower stations and strengthening and reinforcing treatment, which is of great significance to improving the safe operation level of hydropower stations and enhancing the operational capacity of hydropower stations.
[0004] At present, visual inspection robots are equipped with image acquisition mechanisms to scan the inner walls of corridors, and perform crack identification and early warning after collecting image data. Crack identification basically adopts image recognition technology, that is, establishing an attribute box. The attribute box generally includes the identified crack number and attribute items for the crack number. The attribute items include crack area, crack contour, crack coordinates, etc. It is difficult to truly feedback crack defects and expected defect assessment. Summary of the invention
[0005] The purpose of the present invention is to solve the deficiencies of the above-mentioned prior art. Aiming at the problems that traditional corridor crack identification has poor correlation, single attributes, and a fixed evaluation model that cannot provide effective evaluation, a corridor visual inspection crack identification method and system are proposed.
[0006] In order to achieve the above object, the technical solution adopted by the present invention is: The corridor visual inspection crack identification method includes the following steps: S1 corridor image data collection, construction of image database and corridor 3D image model; S2 builds a crack recognition model trained on an image database and a crack evaluation model that evaluates real-time data in the three-dimensional image model of the corridor based on the crack recognition model; S3 updates the corridor 3D image model based on real-time corridor image data and outputs a crack assessment report through the crack assessment model.
[0007] Preferably, in step S1, a plurality of circular scanning images with identification codes along the corridor direction are stitched together to form the corridor three-dimensional image model, wherein the circular scanning image is stitched together by a plurality of rectangular images numbered one by one along the circular scanning direction; The rectangular image includes a rectangular image stitching coordinate system for partially overlapping and stitching the rectangular images to form a ring scan image; The ring scan image includes an interval coordinate system for partially overlapping and splicing with the ring scan interval displacement reference to form a corridor three-dimensional image model.
[0008] Preferably, in step S1, any of the rectangular images includes an orthographic image image used for generating a circular scan image and a plurality of auxiliary oblique image images matched with the orthographic image image; In step S2, the crack recognition model includes a crack recognition item for recognizing the positive-view image and a crack recognition correction item for correcting the crack recognition item, wherein the recognition sub-items in the crack recognition item include at least the crack area, the coordinates of the center of gravity of the crack, the length of the crack, the maximum depth of the crack, the average depth of the crack, the maximum width of the crack, and the coordinates of two endpoints; The crack identification correction item at least comprises a maximum depth correction sub-item for correcting the maximum depth of the crack and an average depth correction sub-item for correcting the average depth of the crack.
[0009] Preferably, in step S3, the crack evaluation model includes a summary evaluation item for summarizing crack identification of a plurality of orthographic image pictures, and a joint evaluation item for joint evaluation of crack identification of at least two adjacent rectangular pictures; The summary evaluation items include at least the number of cracks, the total area of cracks, and the total length of cracks; The joint evaluation items include at least the average number of cracks, the average crack area, and the total length of continuous cracks.
[0010] Preferably, in step S3, the crack assessment report includes at least crack area warning, crack maximum depth warning, crack maximum width warning, crack center of gravity warning, crack number warning, crack total area warning, crack total length warning, average crack number warning, average crack area warning, and continuous crack total length warning.
[0011] The present invention also proposes a corridor visual inspection crack identification system, comprising an image data acquisition device for corridor image data acquisition, a database storage module for building an image database, and an image model module for generating a corridor three-dimensional image model; A recognition model training module for building a crack recognition model trained on an image database and a crack evaluation module for building a crack evaluation model for evaluating real-time data in a three-dimensional corridor image model based on a crack recognition model; An updating module for updating the corridor three-dimensional image model according to real-time corridor image data, and an evaluation report generating module for outputting a crack evaluation report through a crack evaluation model.
[0012] Preferably, the image data acquisition device includes a step-by-step circular scan image acquisition mechanism for acquiring a plurality of circular scan images with identification codes along the corridor direction, and a three-dimensional image stitching module for stitching the circular scan images to form a three-dimensional image model of the corridor, wherein the three-micro image stitching module includes a circular scan image stitching module for stitching a plurality of rectangular images numbered one by one along the circular scan direction to form a circular scan image; The rectangular image includes a rectangular image stitching coordinate system used for partially overlapping and stitching rectangular images to form a ring-scan image; The circular scan image includes an interval coordinate system for partially overlapping and splicing with the circular scan interval displacement reference to form a corridor three-dimensional image model.
[0013] Preferably, any of the rectangular images includes an orthographic image image for generating a circular scan image and a plurality of auxiliary oblique image images matched with the orthographic image image; The step-by-step circular scanning image acquisition mechanism includes an image collector for acquiring images of a normal viewing angle and images of an auxiliary oblique viewing angle; The crack recognition model includes a crack recognition item for recognizing a positive-view image and a crack recognition correction item for correcting the crack recognition item, wherein the recognition sub-items in the crack recognition item include at least crack area, crack center of gravity coordinates, crack length, maximum crack depth, average crack depth, maximum crack width, and two end point coordinates; The crack identification correction item at least comprises a maximum depth correction sub-item for correcting the maximum depth of the crack and an average depth correction sub-item for correcting the average depth of the crack.
[0014] Preferably, the crack evaluation model includes a summary evaluation item for summarizing crack identification of a plurality of orthographic image pictures, and a joint evaluation item for joint evaluation of crack identification of at least two adjacent rectangular pictures; The summary evaluation items include at least the number of cracks, the total area of cracks, and the total length of cracks; The joint evaluation items include at least the average number of cracks, the average crack area, and the total length of continuous cracks.
[0015] Preferably, the crack assessment report includes at least crack area warning, maximum crack depth warning, maximum crack width warning, crack center of gravity warning, crack number warning, total crack area warning, total crack length warning, average crack number warning, average crack area warning, and continuous crack total length warning.
[0016] The beneficial effects of the present invention are mainly reflected in: 1. It can realize visual recognition of cracks in corridor inspection and generate three-dimensional models through image data to meet the needs of summary evaluation and joint evaluation.
[0017] 2. The requirements for complete feedback on corridor status are met, timeliness and accuracy are guaranteed, and the level of safe operation is significantly improved.
[0018] 3. The crack identification accuracy is reliable, ensuring the accuracy of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 It is a schematic diagram of the flow of the corridor visual inspection crack identification method of the present invention.
[0020] Figure 2 It is a schematic diagram of the corridor three-dimensional image model in the present invention.
[0021] Figure 3 It is a processing schematic diagram of the crack identification model in the present invention.
[0022] Figure 4 It is a schematic diagram of the summary evaluation in the present invention.
[0023] Figure 5 It is a schematic diagram of the joint evaluation in the invention. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0025] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the inventions. It should also be noted that, for ease of description, only the parts related to the relevant inventions are shown in the accompanying drawings. It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.
[0026] The present invention provides a method for identifying cracks during visual inspection of a corridor, such as Figure 1 As shown, the following steps are included: Corridor image data collection, construction of image database and corridor three-dimensional image model.
[0027] A crack recognition model trained based on the image database and a crack evaluation model that evaluates the real-time data in the corridor three-dimensional image model based on the crack recognition model are constructed.
[0028] The corridor three-dimensional image model is updated according to the real-time corridor image data, and the crack assessment report is output through the crack evaluation model.
[0029] Specific implementation process and principle description: During the corridor inspection process, due to the limitations of manual detection and the hidden nature of the development of structural diseases and damage, conventional manual inspections are not easy to promptly and accurately discover and identify corridor defects and operating status, and it is difficult to provide timely and comprehensive support for structural safety management and control.
[0030] At present, corridor inspection mostly adopts regional image acquisition and recognition, that is, the inspection robot is used to set the trajectory and acquisition method, and the acquired image is used to identify local cracks. The recognition is basically real-time judgment, that is, only the crack analysis is performed on the currently identified window image, and basically only the relationship between the current window image and the position of the inspection robot can be realized, so as to mark the current window information and provide the corresponding numbered window recognition data.
[0031] It is difficult to meet the needs of comprehensive judgment, especially it is easy to have identification perspectives and missed collection, and there are also duplications or partial duplications, which leads to distortion of the overall evaluation and inability to make expected judgments.
[0032] In the present invention, after the corridor image data is collected, a corridor three-dimensional image model is constructed, which meets the corridor modeling requirements based on images and realizes the generation of panoramic three-dimensional images without blind spots.
[0033] In addition, a crack recognition model trained on the image database and a crack evaluation model that evaluates the real-time data in the corridor three-dimensional image model based on the crack recognition model are constructed.
[0034] That is, the crack recognition model is trained through the image database, so that the crack recognition model can accurately identify crack-related attributes, and the crack evaluation model can identify the real-time data in the three-dimensional image model through the crack recognition model, so as to generate corresponding evaluation.
[0035] The training of the crack recognition model is explained. YOLOv8 is used as the basis for crack detection. A new SOTA model is provided, including the object detection network with P5 640 and P6 1280 resolutions and the instance segmentation model based on YOLACT. Like YOLOv5, different size models of N / S / M / L / X scales are also provided based on the scaling factor to meet the needs of different scenarios.
[0036] Backbone: The backbone network and Neck part refer to the design concept of YOLOv7 ELAN. The C3 structure of YOLOv5 is replaced with the C2f structure with richer gradient flow, and different numbers of channels are adjusted for models of different scales.
[0037] Loss: YOLOv8 abandons the previous IOU matching or unilateral ratio allocation method, and instead uses the Task-Aligned Assigner positive and negative sample matching method. And introduces Distribution Focal Loss (DFL).
[0038] Train: The data enhancement part of the training introduces the operation of turning off Mosiac enhancement in the last 10 epochs in YOLOX, which can effectively improve the accuracy.
[0039] Model framework: The YOLOv8 network structure consists of three parts: Backbone network, Neck structure and Head structure. Among them, Backbone is responsible for feature extraction; Neck is responsible for feature fusion; Head contains three detection heads, which are responsible for outputting detection information.
[0040] Crack identification model and training belong to the existing technology and will not be described in detail here.
[0041] Finally, the corridor three-dimensional image model is updated according to the real-time corridor image data, and the crack assessment report is output through the crack evaluation model.
[0042] Among them, the 3D image model update is used to record real-time image data, which can realize the retrieval of the latest updated model. The crack assessment report is used to display the information output of the inspection equipment.
[0043] In a specific embodiment, a plurality of circular scanning images with identification codes along the corridor direction are stitched together to form a corridor three-dimensional image model, wherein the circular scanning image is formed by stitching together a plurality of rectangular images numbered one by one along the circular scanning direction.
[0044] The rectangular image includes a rectangular image stitching coordinate system used for partially overlapping and stitching rectangular images to form a circular scan image.
[0045] The ring scan image includes an interval coordinate system for partially overlapping and splicing with the ring scan interval displacement reference to form a corridor three-dimensional image model.
[0046] Specifically speaking, in general, a corridor is a relatively closed channel structure with a rectangular cross-section, and images of its four walls need to be collected.
[0047] In this case, a circumferential plain scan is performed along the corridor with the corridor direction as the axis to obtain a circular scan image. After the current circular scan image is acquired, the next circular scan image is acquired after stepping one displacement.
[0048] The circular scan image is achieved by the multi-axis circumferential deflection of the photographic robotic arm, so it is composed of several rectangular images stitched together.
[0049] When performing circular scan image stitching, the robotic arm itself has precise private server control, so its current position will be consistent with the data built into the currently generated rectangular image, that is, the rectangular image stitching coordinate system of the current rectangular image. The rectangular image stitching coordinate system has a displacement center point, circumferential runout mark point, rectangular image format data, and framing frame edge data. The circular scan image can be formed by partially overlapping and stitching the rectangular images through the above data and numbers.
[0050] After the circular scan image is generated, it contains displacement step data and frame edge data. The above data are spliced with the identification code to form a three-dimensional image model of the corridor.
[0051] Generally, in construction projects, three-dimensional modeling of corridors will be carried out. This three-dimensional modeling is mostly achieved by surveying and mapping, which inherently has certain deviations. In this case, corridor image data is used to generate and update the corridor three-dimensional image model. The constructed three-dimensional image model is completely based on the corridor image data. Therefore, there is no need for additional surveying and mapping modeling. The corridor three-dimensional image model formed by splicing corridor image data does not need to be too precise and can fully meet the inspection and evaluation needs.
[0052] At the same time, when the three-dimensional image model of the corridor is updated, correction amounts based on the crack data are implanted, that is, reference correction is made through the frame edge data of adjacent rectangular images and the local coordinate data of the cracks, making the three-dimensional image model of the corridor more and more accurate and complete.
[0053] In a specific embodiment, the arbitrary rectangular image includes an orthographic image image used for generating a circular scan image and a plurality of auxiliary oblique image images coordinated with the orthographic image image.
[0054] The crack recognition model includes a crack recognition item for identifying a positive-view image and a crack recognition correction item for correcting the crack recognition item. The recognition sub-items in the crack recognition item include at least the crack area, the coordinates of the center of gravity of the crack, the length of the crack, the maximum depth of the crack, the average depth of the crack, the maximum width of the crack, and the coordinates of two endpoints.
[0055] The crack identification correction item at least comprises a maximum depth correction sub-item for correcting the maximum depth of the cracks and an average depth correction sub-item for correcting the average depth of the cracks.
[0056] Specifically, the crack recognition model mainly relies on the orthographic image pictures, and the identification sub-items in the crack recognition item include at least the crack area, crack center of gravity coordinates, crack length, maximum crack depth, average crack depth, maximum crack width, and two endpoint coordinates. The crack recognition accuracy can be achieved through a large number of map pictures.
[0057] However, the walls of the corridor are quite different from traditional single surfaces such as roads. The dark and humid environment often causes some identification items to fail or be distorted, such as the maximum depth of cracks, the average depth of cracks, etc.
[0058] In this case, a crack recognition correction item was introduced, which assisted the introduction of features of the oblique perspective image to assist in the correction of the recognition of the normal perspective image and provide a deviation reference, thereby making the crack information more accurate and the real-time updated three-dimensional image of the corridor more precise.
[0059] In a specific embodiment, the crack evaluation model includes a summary evaluation item for summarizing crack identification of a plurality of orthographic image pictures, and a joint evaluation item for joint evaluation of crack identification of at least two adjacent rectangular pictures.
[0060] The summary evaluation items shall at least include the number of cracks, the total area of cracks, and the total length of cracks; The joint assessment items include at least the average number of cracks, the average crack area, and the total length of continuous cracks.
[0061] Specifically, traditional corridor inspections only satisfy crack judgments based on a local or single image, which makes it difficult to conduct summary evaluations and also fail to satisfy joint evaluations between multiple adjacent frames.
[0062] In this case, the three-dimensional image model of the corridor can achieve an overall seamless summary evaluation, while the identification coding and numbering can realize the joint evaluation between adjacent rectangular images.
[0063] In a specific embodiment, the crack assessment report includes at least crack area warning, maximum crack depth warning, maximum crack width warning, crack center of gravity warning, crack number warning, total crack area warning, total crack length warning, average crack number warning, average crack area warning, and continuous crack total length warning.
[0064] Specifically, the corridor environment changes greatly, and its crack characteristics are changing almost every moment. For the sake of inspection safety, it is more important to check each crack one by one. The overall evaluation is a feedback on the overall situation of the corridor. Local information is difficult to meet the feedback needs and cannot meet the needs of comprehensive judgment or provide preliminary judgment.
[0065] Therefore, the crack assessment in this case includes crack area warning, maximum crack depth warning, maximum crack width warning, crack center of gravity warning, crack number warning, total crack area warning, total crack length warning, average crack number warning, average crack area warning, and continuous crack total length warning.
[0066] Among them, crack area warning, maximum crack depth warning, maximum crack width warning, crack center of gravity warning, crack number warning, total crack area warning, and total crack length warning are mainly used for overall warning settings.
[0067] When any of the above warnings occurs, it is directly manifested as corridor risk.
[0068] The average crack number warning, average crack area warning, and total continuous crack length warning are used to detect the sharp increase in risk in local related areas, thus achieving local joint evaluation.
[0069] The following is a corridor visual inspection crack identification system for implementing a corridor visual inspection crack identification method.
[0070] It includes an image data acquisition device for collecting corridor image data, a database storage module for building an image database, and an image model module for generating a three-dimensional image model of the corridor.
[0071] A recognition model training module is used to carry a crack recognition model trained on an image database, and a crack evaluation module is used to carry a crack evaluation model that evaluates real-time data in a three-dimensional corridor image model based on a crack recognition model.
[0072] An updating module for updating the corridor three-dimensional image model according to real-time corridor image data, and an evaluation report generating module for outputting a crack evaluation report through a crack evaluation model.
[0073] Specifically speaking, the image data acquisition device is an integrated body of a mobile device equipped with a camera mechanism, and all devices capable of acquiring image data are within the protection scope of this case.
[0074] The database storage module is a storage medium such as a hard disk, the image model module is a storage medium with read and write functions; the update module is software and carriers with read, write and modify functions, the recognition model training module is software and carriers with training and modification functions. The crack evaluation module is software and carriers with evaluation functions.
[0075] The reading, writing, storage and other functions of the above modules are all existing technologies and will not be described in detail here.
[0076] The image data acquisition device includes a step-by-step circular scan image acquisition mechanism for acquiring a number of circular scan images with identification codes along the corridor direction, and a three-dimensional image stitching module for stitching the circular scan images to form a three-dimensional image model of the corridor, wherein the three-micro image stitching module includes a circular scan image stitching module for stitching a number of rectangular images numbered one by one along the circular scan direction to form a circular scan image.
[0077] The step-by-step circular scanning image acquisition mechanism is generally a mobile vehicle guided by a ceiling rail or a ground rail, equipped with an image acquisition robotic arm with multi-axis displacement, while the three-dimensional image stitching module and the circular scanning image stitching module are computing and processing carriers with computing and storage capabilities, which are generally built into the processor.
[0078] From the above description, we can find that visual recognition of cracks in corridor inspection can be realized, and a three-dimensional model can be generated through image data, thereby meeting the needs of summary evaluation and joint evaluation. The needs of complete feedback on corridor status are met, timeliness and accuracy are guaranteed, and the level of safe operation is significantly improved. The crack identification accuracy is reliable, which ensures the accuracy of the evaluation.
[0079] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that includes a list of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article, or apparatus / device.
[0080] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
Claims
1. A method for identifying cracks during visual inspection of a corridor, characterized in that The steps include: S1 corridor image data collection, construction of image database and corridor 3D image model; S2 builds a crack recognition model trained on an image database and a crack evaluation model that evaluates real-time data in the three-dimensional image model of the corridor based on the crack recognition model; S3 updates the corridor 3D image model based on real-time corridor image data and outputs a crack assessment report through the crack assessment model.
2. The corridor visual inspection crack identification method according to claim 1 is characterized by: In the step S1, a plurality of circular scanning images with identification codes along the corridor direction are stitched together to form the corridor three-dimensional image model, wherein the circular scanning image is stitched together by a plurality of rectangular images numbered one by one along the circular scanning direction; The rectangular image includes a rectangular image stitching coordinate system for partially overlapping and stitching the rectangular images to form a ring scan image; The ring scan image includes an interval coordinate system for partially overlapping and splicing with the ring scan interval displacement reference to form a corridor three-dimensional image model.
3. The corridor visual inspection crack identification method according to claim 2 is characterized by: In step S1, any of the rectangular images includes an orthographic image image used for generating a circular scan image and a plurality of auxiliary oblique image images matched with the orthographic image image; In step S2, the crack recognition model includes a crack recognition item for recognizing the positive-view image and a crack recognition correction item for correcting the crack recognition item, wherein the recognition sub-items in the crack recognition item include at least the crack area, the coordinates of the center of gravity of the crack, the length of the crack, the maximum depth of the crack, the average depth of the crack, the maximum width of the crack, and the coordinates of two endpoints; The crack identification correction item at least comprises a maximum depth correction sub-item for correcting the maximum depth of the crack and an average depth correction sub-item for correcting the average depth of the crack.
4. The corridor visual inspection crack identification method according to claim 3 is characterized by: In step S3, the crack evaluation model includes a summary evaluation item for summarizing crack identification of a plurality of orthographic image pictures, and a joint evaluation item for joint evaluation of crack identification of at least two adjacent rectangular pictures; The summary evaluation items include at least the number of cracks, the total area of cracks, and the total length of cracks; The joint evaluation items include at least the average number of cracks, the average crack area, and the total length of continuous cracks.
5. The corridor visual inspection crack identification method according to claim 4 is characterized by: In step S3, the crack assessment report includes at least crack area warning, crack maximum depth warning, crack maximum width warning, crack center of gravity warning, crack number warning, crack total area warning, crack total length warning, average crack number warning, average crack area warning, and continuous crack total length warning.
6. Corridor visual inspection crack identification system, characterized by: It includes an image data acquisition device for collecting corridor image data, a database storage module for building an image database, and an image model module for generating a three-dimensional image model of the corridor; A recognition model training module for building a crack recognition model trained on an image database and a crack evaluation module for building a crack evaluation model for evaluating real-time data in a three-dimensional corridor image model based on a crack recognition model; An updating module for updating the corridor three-dimensional image model according to real-time corridor image data, and an evaluation report generating module for outputting a crack evaluation report through a crack evaluation model.
7. The corridor visual inspection crack identification system according to claim 6 is characterized by: The image data acquisition device includes a step-by-step circular scanning image acquisition mechanism for acquiring a plurality of circular scanning images with identification codes along the corridor direction, and a three-dimensional image stitching module for stitching the circular scanning images to form a three-dimensional image model of the corridor, wherein the three-micro image stitching module includes a circular scanning image stitching module for stitching a plurality of rectangular images numbered one by one along the circular scanning direction to form a circular scanning image; The rectangular image includes a rectangular image stitching coordinate system used for partially overlapping and stitching rectangular images to form a ring-scan image; The circular scan image includes an interval coordinate system for partially overlapping and splicing with the circular scan interval displacement reference to form a corridor three-dimensional image model.
8. The corridor visual inspection crack identification system according to claim 7 is characterized by: Any of the rectangular images includes an orthographic image image used for generating a circular scan image and a plurality of auxiliary oblique image images matched with the orthographic image image; The step-by-step circular scanning image acquisition mechanism includes an image collector for acquiring images of a normal viewing angle and auxiliary oblique viewing angle images; The crack recognition model includes a crack recognition item for recognizing a positive-view image and a crack recognition correction item for correcting the crack recognition item, wherein the recognition sub-items in the crack recognition item include at least crack area, crack center of gravity coordinates, crack length, maximum crack depth, average crack depth, maximum crack width, and two end point coordinates; The crack identification correction item at least comprises a maximum depth correction sub-item for correcting the maximum depth of the crack and an average depth correction sub-item for correcting the average depth of the crack.
9. The corridor visual inspection crack identification system according to claim 8, characterized in that: The crack evaluation model includes a summary evaluation item for summarizing crack identification of a plurality of orthographic image pictures, and a joint evaluation item for joint evaluation of crack identification of at least two adjacent rectangular pictures; The summary evaluation items include at least the number of cracks, the total area of cracks, and the total length of cracks; The joint evaluation items include at least the average number of cracks, the average crack area, and the total length of continuous cracks.
10. The corridor visual inspection crack identification system according to claim 9, characterized in that: The crack assessment report includes at least crack area warning, maximum crack depth warning, maximum crack width warning, crack center of gravity warning, crack number warning, total crack area warning, total crack length warning, average crack number warning, average crack area warning, and continuous crack total length warning.