A deformation detection system and method for civil air defense doors based on machine vision

Abstract

The present invention discloses a deformation detection system and method for civil air defense doors based on machine vision. The system includes a detection color strip, an intelligent camera, a detection and analysis server, a storage server, and a query terminal. The detection and analysis algorithm of the present invention is an application software installed on the detection and analysis server, and it is a machine vision detection and analysis software. Its core is the detection and analysis algorithm, which mainly consists of several sub-algorithms such as automatic image geometric transformation, automatic image enhancement, edge detection, connected region, smoothing processing, and deformation calculation. The present invention has the advantages of high efficiency, low cost, good timeliness, and high degree of automation in the deformation detection of civil air defense doors, and can fully meet the deformation detection requirements of various types of civil air defense doors. Since this detection and measurement is non-manual operation, the present invention can save a large amount of labor resources and improve the work efficiency of civil air defense project management. Therefore, its application will be more and more extensive.

Description

Technical Field

[0001] The present invention relates to a deformation detection system and method for civil air defense doors based on machine vision, belonging to the technical field of the Internet of Things. Background Art

[0002] Civil air defense projects are a special type of underground construction, and their protective equipment is an important component, playing a key role in the protective function of civil air defense projects. As an important protective equipment for civil air defense projects, whether there is deformation in civil air defense doors is one of the key factors to ensure the protective function. Therefore, how to detect the deformation of civil air defense doors in a timely manner is an important task in the daily management of civil air defense projects.

[0003] Currently, the deformation detection of civil air defense doors can only be carried out by professional personnel through spot checks on a certain range of civil air defense doors at regular intervals, and a completely manual method is used for detection. However, there are deficiencies such as a single means, a large workload, and low efficiency. Such a completely manual detection method can no longer meet the management requirements of civil air defense work and has encountered a major bottleneck in the management of civil air defense projects. Summary of the Invention

[0004] The purpose of the present invention is to solve the above-mentioned deficiencies of the prior art and propose a deformation detection system for civil air defense doors based on machine vision. This system can solve the problems of single means, large workload, high cost, low efficiency, and poor timeliness in manual detection. The deformation detection of civil air defense doors by this system has the advantages of high efficiency, low cost, good timeliness, and high automation, and can fully meet the deformation detection requirements of various types of civil air defense doors. Since this detection and measurement is non-manual operation, the present invention can save a large amount of labor resources and improve the work efficiency of the management of civil air defense projects. Therefore, its application will be more and more extensive. The present invention mainly solves the following problems:

[0005] 1) Manual spot checks, high cost and low efficiency;

[0006] Due to the large number of civil air defense doors, but there are few professional personnel with detection capabilities, at present, only spot checks can be used to carry out regular detection work. At the same time, due to the rapid increase in labor costs and the rapid growth of civil air defense project construction, problems such as shortage of professional personnel and high costs are becoming more and more prominent. At present, this method can no longer meet the daily management requirements of civil air defense projects.

[0007] 2) Single detection means, large workload and long maintenance cycle;

[0008] Manual detection is the method currently used. Due to the single detection means, the detection workload of each civil air defense door is very large on average. Because of the large workload, the maintenance cycle is extended, far exceeding the relevant standards formulated by the state. Therefore, this detection method and technology simply cannot meet the requirements of the management of civil air defense projects.

[0009] 3) Low level of informatization, poor timeliness, and weak qualitative ability;

[0010] The manual inspection method can only collect a small amount of data on civil air defense doors. Moreover, due to poor timeliness and a long collection cycle, problems cannot be discovered and maintained in a timely manner. Therefore, the current informatization level of civil air defense doors is very low and almost non-existent in most areas.

[0011] 4) Poor planning and unable to guarantee the implementation strength;

[0012] Due to the high professionalism and large workload of manual inspection, it is impossible to formulate corresponding work plans according to relevant national standards. Only irregular sampling inspections can be adopted based on the local personnel and the number of civil air defense doors. At the same time, due to the manual operation method, the implementation strength and effectiveness cannot be guaranteed.

[0013] 5) Lack of predictability, poor timeliness and pertinence of maintenance;

[0014] Manual inspection can only conduct sampling inspections with a long interval period. Therefore, the collected data is small and predictive analysis cannot be carried out through such data. At the same time, due to the poor timeliness of inspection, the timeliness and pertinence of maintenance are also poor. The current methods and means adopted cannot discover problems in a timely manner, nor can they predict the occurrence of problems. Therefore, timely maintenance cannot be achieved, which greatly reduces the actual effective service life of civil air defense doors and thus increases the overall cost.

[0015] The technical solution adopted by the present invention to solve its technical problems is: a deformation detection system for civil air defense doors based on machine vision. The system includes a detection color strip, an intelligent camera, a detection analysis server, a storage server, and a query terminal, specifically including the following:

[0016] Detection color strip: During the production process, after completion of production or installation of the civil air defense door, a four-sided detection color strip with a width of about 5 cm needs to be generated at the edge of the civil air defense door using tools. The chromaticity value adopted by the detection color strip needs to form an obvious color difference from the chromaticity value of the civil air defense door body.

[0017] Intelligent Camera: It is used to acquire images of the product under test and upload them to the image analysis server. The intelligent camera is built-in with an SD card and supports local storage. In case of power failure or network disconnection and other emergencies, it can save product pictures and video information. It uses high-performance rate distortion compression technology, with high image quality and low bit rate to save bandwidth. It adopts an ultra-high-resolution progressive scan image sensor to provide high-quality and high-definition pictures for the detection and analysis server, improving the detection success rate. It supports WEB-based browsing and management, facilitating managers to query the real-time status of the product to be tested at any time. It provides an external sensor trigger alarm linkage method, which can integrate various external sensors to provide alarms according to actual on-site requirements. The intelligent camera must have the following functions: (1) Strong light suppression; (2) Automatic white balance; (3) Automatic infrared switching; (4) Automatic gain; (5) Automatic noise reduction; (6) Automatic exposure.

[0018] Detection and Analysis Server: A specific machine vision detection and analysis software runs in the detection and analysis server, and data interfaces between the software and other devices in the system are provided. The detection and analysis server receives the image data collected by the intelligent camera, and through the detection, analysis, and comparison of the software, it independently judges whether the civil air defense door under test is deformed. The detection and analysis software is the core of the detection and analysis server, and its main functions include:

[0019] a. Automatic Image Geometric Transformation: Due to errors in actual operations, it is impossible to ensure that the target object is always in the same position and direction. First, detect the position and direction information of the target object in the image, that is, the pose of the target object, and then perform corresponding geometric transformations. The transformations are divided into four types: translation, scaling, rotation, and skew.

[0020] b. Automatic Image Enhancement: Image enhancement is an essential preprocessing process for images. The purpose is to enhance certain features of the images collected by the intelligent camera to facilitate subsequent image processing. It usually includes: grayscale value transformation, sharpening, and noise reduction.

[0021] c. Edge Detection: The process of performing edge detection on the image to obtain the edges of the detection color bands on the civil air defense door.

[0022] d. Connected Regions: The regions obtained through image edge detection are only subsets of one or more images in the algorithm. In order to obtain the detection marking lines required later, these segmented subsets must be marked as connected regions one by one. By connecting adjacent regions, the edge connection lines of the detection color bands in the entire image can be obtained.

[0023] e. Smoothing Processing: The obtained detection color band connection lines may not be smooth. In order to obtain smooth detection color band edge lines, it is necessary to perform smoothing processing on the above-mentioned obtained detection color band edge connection lines, so as to obtain two continuous edge lines of the detection color band of the entire civil air defense door.

[0024] f. Deformation calculation: Compare the processed target image with the initial template of the personnel protection door. Mainly based on the differences between the two detection color band edges and the angular differences of the four diagonals, and extract the relevant previous data of the personnel protection door from the storage server, so as to calculate values such as the change rate, and synchronously calculate the deformation degree and deformation rate of the relevant parts of the personnel protection door, etc., and then automatically judge whether the currently detected personnel protection door has deformation or the probability of deformation.

[0025] g. Product information transmission: When the judgment of the personnel protection door is completed this time, the software will send the detection information of the personnel protection door to the storage server for storage or relevant applications outside the system through the data interface.

[0026] Storage server: A data server used to store detection results;

[0027] Query terminal: Supports mobile phones, tablets and ordinary PCs to query analysis data and statistical reports at a specific time from the storage server.

[0028] The present invention also provides a method for implementing a deformation detection system for a personnel protection door based on machine vision, including: First, check whether the personnel protection door has a detection color band. If not, add it manually as required; The intelligent camera accurately collects the image of the current personnel protection door at regular intervals according to the detection plan requirements, and preprocesses the image through automatic exposure, automatic white balance, automatic gain, and automatic noise reduction technologies, and synchronously uploads the processed image to the detection analysis server; A specific machine vision application software runs in the detection analysis server. This application software will analyze specific areas on the image, correct the image deformation caused by the shooting angle through automatic image geometric transformation, and then obtain the feature image for subsequent calculation through automatic image enhancement, edge detection, connected region, and smoothing processing; Finally, the deformation data of the personnel protection door is obtained through deformation calculation, so as to detect the quality result of the current personnel protection door; After the calculation is completed, if the detection result reaches the warning condition, relevant warning information will be issued outside the system; If the detection result reaches the alarm condition, relevant alarm information will be issued outside the system; At the same time, the detection analysis server will store the detection result of the personnel protection door in the storage server. The storage server is a database server that can store data content for up to one year or even longer. At the same time, the system is equipped with query software to support different terminals to query the quality results at a specific time and relevant statistical reports.

[0029] The detection analysis algorithm of the present invention is an application software installed on the detection analysis server, which is a machine vision detection analysis software. Its core is the detection analysis algorithm, which mainly includes several sub-algorithms such as automatic image geometric transformation, automatic image enhancement, edge detection, connected region, smoothing processing, and deformation calculation.

[0030] Automatic Image Geometric Transformation: Through the sub-algorithm of the present invention, the image collected and processed by the camera can be transformed into an image with the same angle as the sample image of the civil air defense door, ensuring the consistency of the geometric features of the image; in particular, it is necessary to ensure that the color band area in the image is consistent with the sample image.

[0031] Automatic Image Enhancement: Through the sub-algorithm of the present invention, the color difference between the color band area and the background color area of the civil air defense door in the image is increased, thus highlighting the color band area.

[0032] Edge Detection: Through the sub-algorithm of the present invention, the periphery of the color band area is marked to form two nearly quadrilateral marked points including the inner four sides and the outer four sides.

[0033] Connected Region: Through the sub-algorithm of the present invention, regions with similar features are connected to form two connected nearly quadrilateral connections.

[0034] Smoothing Process: Through the sub-algorithm of the present invention, the uneven connections are formed into two flat inner and outer quadrilaterals.

[0035] Deformation Calculation: Through the sub-algorithm of the present invention, the deformation rate and deformation speed of each side of the inner and outer quadrilaterals can be calculated; based on these 8 values, it can be determined whether there is deformation or the possibility of deformation in the civil air defense door.

[0036] Beneficial Effects:

[0037] 1. The cost of the present invention is a one-time investment, and the annual average cost is low: For a civil air defense project, the number of civil air defense doors is large, but after classification, typical points can be selected to install corresponding intelligent cameras, and this camera is also used for detection such as illegal occupation at the same time. Therefore, the overall cost is low and the social benefit is greater.

[0038] 2. The detection means of the present invention is advanced and fully automated: Since the detection method of this system is realized through software algorithms, and the system has strong self-adaptability and a high level of intelligence, this system has achieved full automation, greatly improving the overall efficiency.

[0039] 3. The present invention has a high level of informatization, strong timeliness and reliability: Complete, timely and reliable detection data can provide basic data support for the intelligent civil air defense system of the civil air defense management department.

[0040] 4. The detection method of the present invention has strong planning and reliable execution results: The detection results can be automatically extracted according to the plan through the automatic inspection function; the detection results can also be automatically extracted outside the plan by setting manual inspection.

[0041] 5. Strong prediction ability and timely maintenance: The present invention has a strong prediction ability, can timely detect potential risks, give warnings and alerts, and provide relevant handling suggestions. Thus, the timeliness and planning of maintenance are improved, the service life of the civil air defense door is greatly extended, and the overall construction cost and maintenance cost are reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is the system architecture diagram of the present invention.

[0043] Figure 2 It is the processing flow chart of the detection and analysis algorithm of the present invention.

[0044] Figure 3 It is the working flow chart of the civil air defense door deformation detection system based on machine vision of the present invention.

[0045] Figure 4 It is the specific implementation schematic diagram of the system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The following will disclose the embodiments of the present invention with diagrams. For the sake of clarity, many practical details will be described together in the following narrative. However, it should be understood that these practical details are not used to limit the present invention. That is to say, in some embodiments of the present invention, these practical details are not necessary.

[0047] As Figure 1 shown, the present invention provides a civil air defense door deformation detection system based on machine vision. The system includes a detection color strip, an intelligent camera, a detection and analysis server, a storage server and a query terminal, and specifically includes the following:

[0048] Detection color strip: During the production process, after production or after installation of the civil air defense door, a four-sided detection color strip with a width of about 5 cm needs to be generated at the edge of the civil air defense door using tools. The chromaticity value adopted by the detection color strip needs to form an obvious color difference with the chromaticity value of the civil air defense door body.

[0049] Intelligent Camera: It is used to acquire images of the product under test and upload them to the image analysis server. The intelligent camera is built-in with an SD card and supports local storage. In case of power failure or network disconnection and other emergencies, it can save product pictures and video information. It uses high-performance rate-distortion compression technology, with high image quality and low bit rate to save bandwidth. It adopts an ultra-high-resolution progressive scan image sensor to provide high-quality and high-definition pictures for the detection and analysis server, improving the detection success rate. It supports WEB-based browsing and management, facilitating managers to query the real-time status of the product to be tested at any time. It provides an external sensor-triggered alarm linkage method, which can integrate various external sensors to provide alarms according to actual on-site requirements. The intelligent camera must have the following functions: (1) strong light suppression; (2) automatic white balance; (3) automatic infrared switching; (4) automatic gain; (5) automatic noise reduction; (6) automatic exposure.

[0050] Detection and Analysis Server: A specific machine vision detection and analysis software runs in the detection and analysis server, and data interfaces between the software and other devices in the system are provided. The detection and analysis server receives the image data collected by the intelligent camera, and through the detection, analysis, and comparison of the software, it independently judges whether the civil air defense door under test is deformed. The detection and analysis software is the core of the detection and analysis server, and its main functions are:

[0051] a. Automatic Image Geometric Transformation: Due to errors in actual operations, it is impossible to ensure that the target object is always in the same position and direction. First, detect the position and direction information of the target object in the image, that is, the pose of the target object, and then perform corresponding geometric transformations. The transformations are divided into four types, namely translation, scaling, rotation, and skew.

[0052] b. Automatic Image Enhancement: Image enhancement is a necessary preprocessing process for images, aiming to enhance certain features of the images collected by the intelligent camera for subsequent image processing. It usually includes: gray value transformation, sharpening, and noise reduction.

[0053] c. Edge Detection: The process of performing edge detection on the image to obtain the edges of the detection color bands on the civil air defense door.

[0054] d. Connected Regions: The regions obtained through image edge detection are only subsets of one or more images in the algorithm. In order to obtain the detection marking lines required later, these segmented subsets must be marked as connected regions one by one. By connecting adjacent regions, the edge connection lines of the detection color bands in the entire image can be obtained.

[0055] e. Smoothing Processing: The obtained detection color band connection lines may not be smooth. In order to obtain smooth detection color band edge lines, the above-obtained detection color band edge connection lines need to be smoothed, so as to obtain two continuous edge lines of the detection color bands of the entire civil air defense door.

[0056] f. Deformation calculation: Compare the processed target image with the initial template of the personnel protection door. Mainly based on the differences in the two detection color band edges and the angular differences of the four diagonals, and extract the relevant previous data of the personnel protection door from the storage server, so as to calculate values such as the change rate, and synchronously calculate the deformation degree and deformation rate of the relevant parts of the personnel protection door, etc., and then automatically judge whether the currently detected personnel protection door has deformation or the probability of deformation.

[0057] g. Product information transmission: After the judgment of the personnel protection door this time is completed, the software will send the detection information of the personnel protection door to the storage server for storage or relevant applications outside the system through the data interface.

[0058] Storage server: A data server used to store detection results;

[0059] Query terminal: Supports mobile phones, tablets and ordinary PCs to query analysis data and statistical reports at specific times from the storage server.

[0060] Such as Figure 2 and Figure 3 As shown, the present invention provides a method for implementing a deformation detection system for a personnel protection door based on machine vision, including: First, check whether the personnel protection door has a detection color band. If not, add it manually as required; The intelligent camera accurately collects the image of the current personnel protection door at regular intervals according to the detection plan requirements, and preprocesses the image through automatic exposure, automatic white balance, automatic gain, and automatic noise reduction technologies, and synchronously uploads the processed image to the detection and analysis server; A specific machine vision application software runs in the detection and analysis server. The application software will analyze specific areas on the image, correct the image deformation caused by the shooting angle through automatic image geometric transformation, and then obtain the feature image for subsequent calculation through automatic image enhancement, edge detection, connected region, and smoothing processing; Finally, the deformation data of the personnel protection door is obtained through deformation calculation, so as to detect the quality result of the current personnel protection door; After the calculation is completed, if the detection result reaches the warning condition, relevant warning information will be released outside the system; If the detection result reaches the alarm condition, relevant alarm information will be released outside the system; At the same time, the detection and analysis server will store the detection result of the personnel protection door in the storage server. The storage server is a database server that can store data content for up to one year or even longer. At the same time, the system is equipped with query software to support different terminals to query the quality results at specific times and relevant statistical reports.

[0061] The detection and analysis algorithm of the present invention is an application software installed on the detection and analysis server, which is a machine vision detection and analysis software. Its core is the detection and analysis algorithm, which mainly includes several sub-algorithms such as automatic image geometric transformation, automatic image enhancement, edge detection, connected region, smoothing processing, and deformation calculation.

[0062] Automatic image geometric transformation: Through the sub-algorithm of the present invention, the image collected and processed by the camera can be transformed into an image with the same angle as the sample image of the air defense door, ensuring the consistency of the geometric features of the image; in particular, it is necessary to ensure that the color band area in the image is consistent with the sample image.

[0063] Automatic image enhancement: Through the sub-algorithm of the present invention, the color difference between the color band area and the background color area of the air defense door in the image is increased, so as to highlight the color band area.

[0064] Edge detection: Through the sub-algorithm of the present invention, the periphery of the color band area is marked to form two nearly quadrilateral marked points including the inner four sides and the outer four sides.

[0065] Connected region: Through the sub-algorithm of the present invention, the regions with similar features are connected to form two connected nearly quadrilateral connections.

[0066] Smoothing processing: Through the sub-algorithm of the present invention, the uneven connections are formed into two flat inner and outer quadrilaterals.

[0067] Deformation calculation: Through the sub-algorithm of the present invention, the deformation rate and deformation speed of each side of the inner and outer two quadrilaterals can be calculated; based on these 8 values, it can be determined whether the air defense door has deformation or the possibility of deformation.

[0068] Figure 4 This is a schematic diagram of the specific implementation of the system of the present invention. The intelligent air defense workstation of the present invention mainly includes three levels, namely: intelligent perception devices, front-end intelligent gateways, and intelligent air defense workstations.

[0069] Intelligent perception devices: By introducing intelligent perception devices related to air defense engineering equipment, whether existing or independently developed, the automatic data collection of air defense engineering equipment information is realized, and the real-time performance, data accuracy, etc. of the system are improved. The intelligent perception devices mainly include: environmental detection devices, wall detection devices, flame monitoring devices, intelligent cameras for deformation detection of air defense doors and detection of illegal occupation of air defense passages / spaces, nuclear, biological and chemical detection devices, etc.

[0070] Front-end intelligent gateway: It realizes the analysis and recognition of the data output by the front-end perception devices and provides the required information data for the back-end system. This front-end intelligent gateway includes all the functions of the detection and analysis server of the present invention, and it is mainly composed of the following parts:

[0071] 1) Based on the exclusive ARM microprocessor array architecture, multi-dimensional and multi-source heterogeneous data load balancing processing is realized.

[0072] 2) The multi-core array realizes the application of special algorithms such as video analysis, image recognition, feature detection, OCR, voiceprint recognition and processing.

[0073] 3) Provide support for edge computing analysis capabilities and business applications, and provide rich physical interfaces and data interfaces to access various types of front-end intelligent perception information data, which is the center of the front-end perception system.

[0074] Intelligent civil air defense workstation: It includes devices and functions such as the storage server and query terminal of the present invention, and mainly realizes the following functions:

[0075] 1) Based on front-end multi-dimensional perception and front-end intelligent gateways, comprehensively and stereoscopically master the basic data and real-time data of civil air defense projects, equipment and facilities, space environment, and illegal behaviors, and give timely early warnings, alarm interventions, and disposals for abnormal situations, forming a complete workflow closed-loop at the workstation level to achieve integrated early warning linkages.

[0076] 2) Through system data tools and services, implement comprehensive inspection plans for civil air defense, security, and fire protection, achieving 80% automatic inspections + 20% manual inspections, greatly reducing the workload of grass-roots personnel and improving work efficiency.

[0077] 3) Through big data aggregation and modeling, construct an intelligent civil air defense management expert human-machine coupling system to automatically generate daily maintenance plans for civil air defense facilities, making the plans more scientific and efficient, reducing maintenance costs, and extending the service life of equipment and facilities.

[0078] 4) Through docking with the management platform of the superior competent department, provide data support, auxiliary decision-making, execution feedback, and information aggregation for the superior platform in aspects such as flood control early warning, emergency evacuation, and wartime management, greatly improving the implementation efficiency of the intelligent civil air defense system in the last mile.

[0079] 5) Use the storage server to store relevant data, such as video data, image data, detection data, calculation result data, basic data, etc., encrypt confidential data according to relevant standards, and also need to ensure the data life cycle according to relevant regulations; at the same time, provide query services for various query terminals.

[0080] The above is only the implementation mode of the present invention and is not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. A deformation detection system for civil air defense doors based on machine vision, characterized in that: the system includes a detection color strip, an intelligent camera, a detection and analysis server, a storage server, and a query terminal; Detection color strip: During the production process of the civil air defense door, after production or installation is completed, a four-sided detection color strip needs to be generated at the edge of the civil air defense door using tools. The chromaticity value of the detection color strip needs to form an obvious color difference from the chromaticity value of the civil air defense door body; Intelligent camera: Used to obtain images of the product to be measured and upload them to the image analysis server. The intelligent camera is built-in with an SD card and supports local storage. In the event of sudden power failure or network disconnection, it saves product pictures and video information; uses high-performance rate-distortion compression technology, with high image quality and low bit rate to save bandwidth; adopts an ultra-high-resolution progressive scan image sensor to provide high-quality and high-definition pictures for the detection and analysis server, improving the detection success rate; supports WEB-based browsing and management, facilitating managers to query the real-time status of the product to be measured at any time; provides an external sensor trigger alarm linkage method, which can integrate various external sensors and provide alarms according to actual on-site requirements; Detection and analysis server: A specific machine vision detection and analysis software runs in the detection and analysis server, and data interfaces between the software and other devices in the system are provided. The detection and analysis server receives the image data collected by the intelligent camera, and through the detection, analysis, and comparison of the software, autonomously judges whether the civil air defense door to be measured is deformed. The detection and analysis software is the core of the detection and analysis server. The functions of the detection and analysis server include: Automatic image geometric transformation: First, detect the position and orientation information of the target object in the image, that is, the pose of the target object, and then perform corresponding geometric transformations. The transformations are divided into four types, namely translation, scaling, rotation, and skew; Automatic image enhancement: Image enhancement is a necessary preprocessing process for images. It enhances certain features of the images collected by the intelligent camera to facilitate subsequent image processing, including: gray value transformation, sharpening, and noise reduction; Edge detection: The process of performing edge detection on the image to obtain the edges of the detection color strip on the civil air defense door; Connected regions: The regions obtained through image edge detection are only subsets of one or more images in the algorithm. In order to obtain the detection marking lines required later, these segmented subsets must be marked as connected regions one by one. By connecting adjacent regions, the edge connection lines of the detection color strip in the entire image can be obtained; Smoothing processing: The obtained connection lines of the detection color strip are not smooth. In order to obtain smooth edges of the detection color strip, the connection lines of the detected color strip edges obtained above need to be smoothed, so as to obtain two continuous edge lines of the detection color strip of the entire civil air defense door; Deformation calculation: Compare the processed target image with the initial template of the civil air defense door itself, based on the gap between the two detection color strip edges and the degree difference of the four diagonals, and extract the relevant previous data of the civil air defense door from the storage server, so as to calculate the value of the change rate, synchronously calculate the deformation degree and deformation rate of the relevant parts of the civil air defense door, and then automatically judge whether the currently detected civil air defense door has deformation or the probability of deformation; Product information transmission: After the judgment of the current civil air defense door is completed, the software will send the detection information of this civil air defense door to the storage server for storage or relevant applications outside the system through the data interface; Storage server: A data server used to store detection results; Query terminal: Supports mobile phones, tablets, and ordinary PCs to query the analysis data and statistical reports at a specific time from the storage server.

2. A deformation detection system for civil air defense doors based on machine vision according to claim 1, characterized in that: The functions of the intelligent camera include strong light suppression, automatic white balance, automatic infrared switching, automatic gain, automatic noise reduction, and automatic exposure.

3. An implementation method of a deformation detection system for civil air defense doors based on machine vision, characterized in that, including: First, check whether the civil air defense door has a detection color strip. If not, add it manually as required; The intelligent camera accurately collects the image of the current civil air defense door at regular intervals according to the detection plan requirements, and preprocesses the image through automatic exposure, automatic white balance, automatic gain, and automatic noise reduction technologies, and synchronously uploads the processed image to the detection analysis server; A specific machine vision application software runs in the detection analysis server. This application software will analyze specific areas on the image, correct the image deformation caused by the shooting angle through automatic image geometric transformation, and then obtain the characteristic image for subsequent calculation through automatic image enhancement, edge detection, connected region, and smoothing processing; finally, the deformation data of the civil air defense door is obtained through deformation calculation, so as to detect the quality result of the current civil air defense door; After the calculation is completed, if the detection result reaches the warning condition, relevant warning information will be issued outside the system; If the detection result reaches the alarm condition, relevant alarm information will be issued outside the system; at the same time, the detection analysis server will store the detection result of the civil air defense door in the storage server; The storage server is a database server that can store data content for up to one year or even longer. At the same time, the system is equipped with query software, which supports different terminals to query the quality results at a specific time and relevant statistical reports; The method includes a detection analysis algorithm; The detection analysis algorithm is an application software installed on the detection analysis server, which is a machine vision detection analysis software. Its core is the detection analysis algorithm, and its sub-algorithms include: automatic image geometric transformation, automatic image enhancement, edge detection, connected region, smoothing processing, and deformation calculation; Automatic image geometric transformation: Through the sub-algorithm, the image collected and processed by the camera is transformed into an image with the same angle as the sample image of this civil air defense door, ensuring the consistency of the geometric features of the image and ensuring that the color strip area in the image is consistent with the sample image; Automatic image enhancement: Through the sub-algorithm, the color difference between the color strip area and the background color area of the civil air defense door in the image is increased, so as to highlight the color strip area; Edge detection: Through the sub-algorithm, the periphery of the color strip area is marked to form two nearly quadrilateral marking points including the inner four sides and the outer four sides; Connected region: Through the sub-algorithm, regions with similar features are connected to form two connected nearly quadrilateral connections; Smoothing process: Through a sub-algorithm, the uneven connection lines are formed into two flat inner and outer quadrilaterals; Deformation calculation: Through a sub-algorithm, calculate the deformation rate and deformation speed of each side of the inner and outer quadrilaterals. Based on these 8 values, determine whether there is deformation or the possibility of deformation of the air defense door.

Images