Shaft outer surface disease detection device and detection method based on image recognition

Abstract

The invention discloses a vertical shaft outer surface disease detection device and detection method based on image recognition. Data exchange is carried out between a well end detection device and a mobile terminal through a wireless communication network; a fixing clamping plate is arranged on the first side of a shell of the well end detection device, and a camera is arranged on the second side, opposite to the first side, of the shell. A power supply, a data processing module and a wireless communication module are arranged in the shell; the shell is provided with a charging interface and a data IO interface. And the data IO interface is connected with external equipment through a data line and exchanges data with the external equipment. The well end detection device shoots a plurality of pictures and carries out splicing processing to form a disease distribution cloud picture of one side of the vertical shaft, and then the disease distribution cloud pictures of all sides of the whole vertical shaft are completed. The vertical shaft outer surface disease detection device and method have the advantages that time and labor are saved, dependence on human experience is avoided, operation is convenient, and the disease distribution condition and severity of vertical shaft surface concrete are visually displayed.

Description

Technical Field

[0001] The invention relates to a shaft monitoring technology, in particular to a shaft outer surface disease detection device and a shaft outer surface disease detection method based on image recognition. Background Art

[0002] A vertical shaft is an elevator shaft made of concrete and used to lift materials. It is tall and usually has a regular rectangular cross-section. With the increase of service life, and affected by many factors such as initial defects in construction quality, vertical loads, weathering, shrinkage and creep of concrete, the vertical shaft will naturally age and suffer structural damage. Vertical shafts are prone to concrete cracking, spalling, exposed steel bars, rust and other defects, which will affect the working performance of the shaft and even cause accidents. Therefore, it is necessary to conduct regular inspections and prevent and control vertical shafts in advance.

[0003] In the prior art, the traditional method of shaft inspection is: maintenance personnel take an engineering lift to conduct visual inspections of concrete surface defects. After discovering the defect, they use a marker to mark the defect surface, take photos and make corresponding records, and then organize the materials and write inspection reports later. The shaft is usually tens of meters to hundreds of meters high, or even as high as 500 meters or more. For higher shafts, ordinary engineering lifts usually find it difficult to reach the middle and upper parts of the shaft, and more detailed and accurate inspections cannot be obtained here. In addition, the shaft is composed of multiple shaft walls. During the inspection, the lift needs to be frequently controlled to move back and forth and repeatedly lifted and lowered, and the inspection efficiency is low. Moreover, the inspectors need to work continuously for a long time in the narrow lift platform, and the working comfort and experience are extremely poor.

[0004] With the development of drone technology, a shaft inspection method using drones has emerged. Shaft inspection and maintenance personnel remotely control drones and use drone cameras to collect images of the shaft. The collected images are then summarized and organized for manual inspection. Although this solves the problem of inspectors needing to work in the shaft for a long time, the collected images still require manual screening and inspection. Both of these manual inspection methods have a low degree of intelligence, and the accuracy of the inspection is highly dependent on the personal experience of the inspectors. The inspection results of different people are highly discrete, and the referenceability of the inspection results is poor. Summary of the invention

[0005] The present invention is to avoid the deficiencies existing in the above-mentioned prior art and provide a shaft outer surface defect detection device and detection method based on image recognition to improve the intelligence level of shaft detection, detection accuracy and reliability.

[0006] The present invention adopts the following technical solutions to solve the technical problems.

[0007] The present invention discloses a device for detecting defects on the outer surface of a shaft based on image recognition, comprising a shaft end detection device and a mobile terminal; the shaft end detection device and the mobile terminal exchange data via a wireless communication network;

[0008] The wellhead detection device comprises a housing 1; a fixing card plate 2 is arranged on a first side of the housing 1, and a camera 3 is arranged on a second side of the housing 1 opposite to the first side; a power supply 4, a data processing module 5 and a wireless communication module 6 are arranged inside the housing;

[0009] The housing 1 is provided with a charging interface 7 and a data IO interface 8; the charging interface is connected to the power supply, and the charging interface is connected to an external mains interface via a charging cable 11 to charge the power supply; the data IO interface is connected to an external device via a data cable and exchanges data with the external device.

[0010] The structural features of the shaft outer surface defect detection device based on image recognition of the present invention are also as follows:

[0011] Furthermore, the mobile terminal is a mobile phone or a tablet computer.

[0012] Furthermore, on the second side of the housing, a ring-shaped flash light 9 is provided around the camera 3 .

[0013] Furthermore, a communication antenna 10 is connected to the wireless communication module, and the communication antenna 10 extends out from the top of the housing 1 .

[0014] The present invention also discloses a detection method of a shaft outer surface defect detection device based on image recognition, comprising the following steps:

[0015] Step 1: Install a wellhead detection device on the drone 12; install detection software on the mobile terminal;

[0016] Step 2: The drone moves to the first shooting position, starts the well end detection device, and tests whether the shooting effect and signal transmission are good through the images taken by the mobile phone;

[0017] Step 3: Control the drone to shoot the shaft according to the predetermined shooting route;

[0018] Step 4: The drone moves to the next shooting position and continues to shoot the shaft according to the predetermined shooting route;

[0019] Step 5: Repeat step 4 until one side of the shaft is photographed, and obtain n pictures of one side of the shaft;

[0020] Step 6: Use the "stitching" option of the detection software to pre-process the n images obtained in step 5;

[0021] Step 7: stitch the n pre-processed images to obtain a complete image of one side of the shaft, and send the complete image to the mobile terminal;

[0022] Step 8: The staff receives the complete image on the mobile terminal, confirms that the image quality is good, and then clicks the "Disease Detection" option in the detection software to control the wellhead detection device to perform different types of disease detection. The wellhead detection device processes multiple detection results to obtain image mask data and sends it back to the mobile terminal;

[0023] Step 9: The staff receives the image mask data on the mobile terminal and forms a disease distribution cloud map;

[0024] Step 10: After completing the concrete surface defect detection on one side of the shaft, repeat steps 2 to 9 to detect the next side of the lower shaft until all sides of the shaft are detected.

[0025] Furthermore, in step 3, the predetermined shooting route is: shooting from low to high in a serpentine direction from one side of the shaft in a horizontal direction first and then in a vertical direction.

[0026] Furthermore, in step 3, the overlap rate Or of the two pictures taken continuously by the camera is higher than 30%.

[0027] Furthermore, in step 8, different types of disease detection processes include concrete crack detection, reinforcement leakage detection, and concrete spalling and honeycomb surface detection.

[0028] Furthermore, in step 9, different types of disease detection results are first classified, and then the disease detection results are merged according to the disease severity coefficients Wj of the different types of disease detection results.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] The present invention discloses a device and method for detecting defects on the outer surface of a shaft based on image recognition. The well-end detection device and a mobile terminal exchange data via a wireless communication network. A fixed card is provided on the first side of the housing of the well-end detection device, and a camera is provided on the second side of the housing opposite to the first side. A power supply, a data processing module and a wireless communication module are provided inside the housing. A charging interface and a data IO interface are provided on the housing. The data IO interface is connected to an external device via a data line and exchanges data with the external device. The well-end detection device takes multiple pictures and performs splicing processing to form a defect distribution cloud map of one side of the shaft, thereby completing the defect distribution cloud map of all sides of the entire shaft.

[0031] The shaft outer surface defect detection device and method of the present invention have the advantages of saving time and labor, not relying on human experience, being easy to operate, and intuitively displaying the defect distribution and severity of the shaft surface concrete. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The present invention is a schematic diagram of a well end detection device and a mobile terminal of a vertical shaft outer surface disease detection device based on image recognition.

[0033] Figure 2 The present invention is a schematic structural diagram of a shaft end detection device of a shaft outer surface defect detection device based on image recognition.

[0034] Figure 3 This is a schematic diagram of a shaft outer surface defect detection device based on image recognition of the present invention when the shaft end detection device is mounted on a drone to detect the shaft.

[0035] Figure 4 The invention discloses a control interface of detection software of a shaft outer surface defect detection device based on image recognition.

[0036] Figure 5 The present invention is a schematic diagram of the framework of a shaft end detection device of a shaft outer surface defect detection device based on image recognition.

[0037] Figure 6 This is a defect distribution cloud map of a shaft outer surface defect detection device based on image recognition according to the present invention.

[0038] The present invention will be further described below through specific implementation modes in conjunction with the accompanying drawings. DETAILED DESCRIPTION

[0039] Several technical terms involved in the present invention are as follows.

[0040] Shaft: A concrete shaft used to raise materials.

[0041] Image recognition: technology that uses computers to process, analyze and understand images to identify targets and objects in various patterns;

[0042] Health monitoring: strategies and processes for damage identification and characterization of engineering structures;

[0043] Concrete diseases: concrete cracking, spalling, exposed steel bars, rust and other problems caused by external environment, load and concrete characteristics;

[0044] Wireless communication: A method of transmitting data over long distances between multiple nodes without the use of conductors or cables.

[0045] See also Figure 1 to Figure 6 The present invention discloses a device for detecting defects on the outer surface of a shaft based on image recognition, comprising a shaft end detection device and a mobile terminal; the shaft end detection device and the mobile terminal exchange data via a wireless communication network;

[0046] The wellhead detection device comprises a housing 1; a fixing card plate 2 is arranged on a first side of the housing 1, and a camera 3 is arranged on a second side of the housing 1 opposite to the first side; a power supply 4, a data processing module 5 and a wireless communication module 6 are arranged inside the housing;

[0047] The housing 1 is provided with a charging interface 7 and a data IO interface 8; the charging interface is connected to the power supply, and the charging interface is connected to an external mains interface via a charging cable 11 to charge the power supply; the data IO interface is connected to an external device via a data cable and exchanges data with the external device.

[0048] like Figure 1 is a system block diagram of a shaft outer surface defect detection device based on image recognition according to the present invention. Figure 2 It is a schematic diagram of the well-end detection device of the present invention. An L-shaped fixing card is provided on the first side of the shell of the well-end detection device, and the well-end detection device can be mounted on the bottom of the drone through the fixing card. The camera is placed on the second side of the well-end detection device to facilitate the shooting of the shaft surface. A power supply is provided inside the well-end detection device, and a charging interface is connected. The power supply is charged through the charging interface to facilitate offline work. On the one hand, the data processing module is used for image preprocessing and splicing, and on the other hand, a data IO interface is provided, which can be connected to a computer. Various disease detection algorithms are written into the data processing module of the well-end detection device through the computer, so that specific diseases can be detected in the image. The generated detection result mask data is transmitted to the maintenance personnel's mobile phone by the wireless communication module and the communication antenna for visualization processing. It has the advantages of drone-mobile phone dual terminal, simultaneous detection of multiple disease characteristics, and separate visualization of results and data processing.

[0049] During specific implementation, the mobile terminal is a mobile phone or a tablet computer.

[0050] In a specific implementation, on the second side of the housing, a ring-shaped flash light 9 is arranged around the camera 3 .

[0051] like Figure 2 A circular light strip is set around the camera, which is concentric with the center of the camera and serves as a flash to fill in the camera light. A circle of ring-shaped flash can be used to fill in light in dark shooting environments.

[0052] like Figure 3This is a schematic diagram of mounting the well-end detection device on a drone to inspect the shaft. The well-end detection device is connected to the bracket at the bottom of the drone through a fixed card. After the camera is turned on, the maintenance personnel operate the drone to move the shaft from bottom to top, driving the camera on the side of the well-end detection device to obtain the image of the shaft, and transmit the image to the maintenance personnel's mobile phone in real time through the wireless communication module.

[0053] During specific implementation, a communication antenna 10 is connected to the wireless communication module, and the communication antenna 10 extends out from the top of the housing 1 .

[0054] The present invention also discloses a detection method of a shaft outer surface defect detection device based on image recognition, comprising the following steps:

[0055] Step 1: Install a wellhead detection device on the drone 12; install detection software on the mobile terminal;

[0056] The well end detection device is fixed to the body of the drone 12 by fixing the card plate 2. The mobile terminal is a portable intelligent mobile terminal such as a mobile phone or a tablet computer. The matching detection software is installed on the mobile phone. The drone 12 carries the well end detection device, and the camera 3 starts to take pictures and detect from a certain side of the shaft 13, as shown in the attached figure. Figure 3 shown.

[0057] Step 2: The drone moves to the first shooting position, starts the well end detection device, and tests whether the shooting effect and signal transmission are good through the images taken by the mobile phone;

[0058] Start the well detection device through the detection software to start capturing images. Test the communication signal of the wireless communication network through the image transmission on the mobile phone to see if it is good, whether the well detection device is properly installed, and whether the shooting angle is horizontal. If the image tilt is large, it is necessary to manually readjust the installation position of the fixed card plate 2 and adjust the distance from the camera lens to the shaft 13. If the environment is dark and the image quality is poor, control the flash switch through the detection software on the mobile phone to turn on the flash. Figure 4 It is the control and detection interface of the detection software.

[0059] Step 3: Control the drone to shoot the shaft according to the predetermined shooting route;

[0060] Step 4: The drone moves to the next shooting position and continues to shoot the shaft according to the predetermined shooting route;

[0061] After taking photos of the current location, the operator controls the drone equipped with the detection device to go to the next location for shooting.

[0062] Step 5: Repeat step 4 until one side of the shaft is photographed, and obtain n pictures of one side of the shaft;

[0063] Step 6: Use the "stitching" option of the detection software to pre-process the n images obtained in step 5;

[0064] like Figure 4 , click the "splicing" button on the detection software, and the splicing control signal is sent to the wellhead detection device. The data processing module 5 inside the wellhead detection device first preprocesses all n pictures. The preprocessing process includes but is not limited to Gaussian smoothing noise reduction, highlight removal, shadow removal and other processing methods.

[0065] Step 7: stitch the n pre-processed images to obtain a complete image of one side of the shaft, and send the complete image to the mobile terminal;

[0066] like Figure 5 , set the corner point of the first picture numbered 1 as the coordinate origin, and extract the feature descriptor of the picture numbered 1. Then, extract the feature descriptors of the ii=2~nth pictures, match them with the i-1th picture, generate the transformation matrix Rt of the i-th picture, use the transformation matrix Rt to perform perspective correction on the i-th picture, and then splice it with the i-1th picture. Repeat the above splicing process until the splicing of n pictures is completed, and obtain a complete picture of one side of the shaft, and transmit the complete picture to the mobile phone through the wireless communication module of the wellhead detection device for the staff to check the splicing effect.

[0067] Step 8: The staff receives the complete image on the mobile terminal, confirms that the image quality is good, and then clicks the "Disease Detection" option in the detection software to control the wellhead detection device to perform different types of disease detection. The wellhead detection device processes multiple detection results to obtain image mask data and sends it back to the mobile terminal;

[0068] Step 9: The staff receives the image mask data on the mobile terminal and forms a disease distribution cloud map;

[0069] like Figure 4 As shown, the staff can click the disease display switch button on the mobile phone to switch the shaft surface distribution cloud map of different diseases, or select the comprehensive test results to display the superposition of all disease results. The darker the color of the disease distribution cloud map, the more serious the disease. Clicking the corresponding location of the disease on the touch screen can return the GPS coordinates of the disease point for further inspection and repair by the maintenance personnel.

[0070] Step 10: After completing the concrete surface defect detection on one side of the shaft, repeat steps 2 to 9 to detect the next side of the lower shaft until all sides of the shaft are detected.

[0071] After completing the concrete surface defect detection on one side of the shaft, the detection work of the next side can be carried out. After the detection of all sides of the shaft is completed, there is no need to repeatedly disassemble the well end detection device during the detection process. The defect detection work of one or more shafts can be completed at one time, which improves the efficiency of shaft defect detection.

[0072] like Figure 6 The detection software system of the wellhead detection device includes a shooting module, an image preprocessing module, a disease recognition module, and a disease result visualization and positioning module. The shooting module is used to complete steps 2 to 5, the image preprocessing module is used to complete step 7, the disease recognition module is used to complete step 8, and the disease result visualization and positioning module is used to complete step 9, which is convenient for the staff to view the disease distribution cloud map.

[0073] In the specific implementation, in step 3, the predetermined shooting route is: shooting from low to high from one side of the shaft in a serpentine direction, first horizontally and then vertically.

[0074] The shooting route of the well end detection device is recommended to follow the movement trajectory of the camera in a serpentine direction, first horizontally and then vertically for the shaft, from low to high, and the drone drives the camera to slowly rise for shooting. When shooting, the camera moves horizontally from bottom to top, first horizontally to shoot, and then vertically moves up to a new position after reaching the edge of the shaft, and continues to move horizontally to shoot, and then moves vertically up again after reaching the edge of the other side of the shaft, and continues to move horizontally, winding forward in a serpentine shape.

[0075] During specific implementation, in step 3, the overlap rate Or of the two pictures taken continuously by the camera is higher than 30%.

[0076] When shooting, by controlling the speed and direction of the drone and the shooting frequency of the camera, the overlap rate Or of two pictures with adjacent numbers is above 30%, that is, Or ≥ 30%, which is convenient for later picture calibration and stitching; that is, the overlap rate between the previous picture taken by the camera and the next picture is higher than 30%. When the well-end detection device is brought to a new position by the drone, wait until the lens is stable, and press the shooting button on the mobile phone to start shooting. If you are not satisfied with the shooting quality, you can click the album button to enter the album, delete the existing photos, and shoot again, as shown in the attached figure. Figure 4 shown.

[0077] In specific implementation, in step 8, different types of disease detection processes include concrete crack detection, reinforcement leakage detection, and concrete spalling and honeycomb surface detection.

[0078] After the staff determines that the stitching quality of the complete picture of one side of the shaft is good, click the "Disease Detection" button on the mobile phone, and the mobile phone will transmit the control command to the well-end detection device. The data processing module 5 of the well-end detection device uses different types of disease detection algorithms to detect the complete stitching pictures. For example, the threshold segmentation method is used to detect concrete cracks, the Deeplab v3 language segmentation network is used to detect leaking reinforcement, and the YOLO V5 network is used to detect concrete spalling and honeycomb surface. After the detection results of various types of disease detection algorithms are processed, an image mask is generated and sent wirelessly to the mobile phone for visualization processing, such as Figure 4 shown.

[0079] The disease detection algorithm can be imported into the wellhead detection device through the data IO interface 8 of the data processing module 5, and can be called at any time when performing detection. Different disease detection algorithms can be added, deleted, and replaced at any time for different types of diseases that may exist in different shafts, which can expand the application scenarios of the present invention. For disease detection algorithms involving the application of deep learning, the neural network model should be trained in advance and imported into the wellhead detection device, as shown in the attached figure. Figure 2 shown.

[0080] During specific implementation, in step 9, different types of disease detection results are first classified, and then the disease detection results are merged according to the disease severity coefficients Wj of the different types of disease detection results.

[0081] After receiving the image mask data of the disease detection results, the mobile phone classifies the different types of disease detection results and then multiplies them by the severity coefficients W of various diseases. j , where j = 1~k, k is the total number of disease types detected. j represents the disease severity coefficient of the disease detection result of the jth disease detection algorithm, all W j The sum of ΣW j = 1. Disease severity coefficient W j You can set it in the mobile phone software settings, the default is W j =1 / k. Finally, the results of the image mask data are combined to form a disease distribution cloud map, which is displayed on the mobile phone, as shown in the attached figure. Figure 6 shown.

[0082] The present invention discloses a device for detecting defects on the outer surface of a shaft based on image recognition, including a well-end detection device and a mobile terminal, which transmit data via a wireless communication network. The well-end detection device is mounted on a drone to complete tasks that require high computing power and high power consumption, such as taking pictures, preprocessing and stitching pictures, and detecting defects. The mobile terminal is a device such as a mobile phone or a tablet computer, which is equipped with special application software, issues control instructions to the well-end detection device, receives defect detection results, and visualizes the results on the mobile phone screen for staff to view. The mobile phone has good portability and good screen display effect, and is suitable for processing low computing power tasks.

[0083] Figure 4 It is a mobile phone software interface. When the staff sees that the camera imaging picture is ideal, they click the shooting button to transmit the signal to the well-end detection device to obtain the picture of the location. At the same time, the data processing module in the well-end detection device will write the number and location information into the picture file. When the picture of one side of the shaft is collected completely, click the "stitching" button of the mobile phone software to send the picture stitching signal. The data processing module of the well-end detection device will pre-process the scattered pictures and stitch them according to the shooting number. After obtaining the complete stitching picture of one side of the shaft, input the disease detection algorithm, the data processing module will calculate it, and output the mask of the disease detection result to the mobile phone. The mobile phone uses the mask data to visualize the disease on the shaft picture. The staff can observe the disease cloud map to have an intuitive judgment on the concrete disease condition on the shaft surface. At the same time, after clicking on any position in the picture, the photo GPS data will be used for interpolation calculation, and the location coordinates of the clicked position will be returned, so that the occurrence location of various types of diseases can be clearly identified.

[0084] The device for detecting defects on the outer surface of a shaft based on image recognition of the present invention has the following characteristics.

[0085] 1. The present invention adopts different disease detection algorithms to detect different similar diseases, and can also adjust the disease detection algorithm according to different conditions of the shaft. Compared with the traditional manual detection method, it saves time and labor and does not rely on the personal experience of the maintenance personnel.

[0086] 2. The present invention solves the contradiction between equipment detection performance and equipment size. By wirelessly linking the well-end detection device and the mobile phone, the advantages of the well-end detection device, such as large size, strong computing performance and good stability, and the advantages of the mobile phone, such as portability, good screen display effect and easy operation, are fully utilized.

[0087] 3. The disease cloud map display mode of the present invention uses color depth to mark the more serious the disease is, so that the staff can intuitively judge the distribution and severity of the disease on the surface concrete of the shaft.

[0088] The device for detecting defects on the outer surface of a shaft based on image recognition has the advantages of saving time and effort, being independent of human experience, being easy to operate, and being able to intuitively display the distribution and severity of defects on the surface concrete of the shaft.

[0089] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0090] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A device for detecting defects on the outer surface of a shaft based on image recognition, characterized in that: It includes a well end detection device and a mobile terminal; the well end detection device and the mobile terminal exchange data via a wireless communication network; The wellhead detection device comprises a housing (1); a fixing card (2) is arranged on a first side of the housing (1); a camera (3) is arranged on a second side of the housing (1) opposite to the first side; a power supply (4), a data processing module (5) and a wireless communication module (6) are arranged inside the housing; The housing (1) is provided with a charging interface (7) and a data IO interface (8); the charging interface is connected to the power supply, and is connected to an external mains interface via a charging line (11) to charge the power supply; the data IO interface is connected to an external device via a data line and exchanges data with the external device.

2. According to the image recognition-based vertical shaft outer surface defect detection device of claim 1, it is characterized in that: The mobile terminal is a mobile phone or a tablet computer.

3. The device for detecting defects on the outer surface of a shaft based on image recognition according to claim 1 is characterized in that: On the second side of the housing, a ring-shaped flash light (9) is arranged around the camera (3).

4. The device for detecting defects on the outer surface of a shaft based on image recognition according to claim 1 is characterized in that: The wireless communication module is connected to a communication antenna (10), and the communication antenna (10) extends out of the top of the housing (1).

5. A detection method for the shaft outer surface defect detection device based on image recognition according to claim 1, characterized in that: The steps include: Step 1: Install a wellhead detection device on the drone (12); install detection software on the mobile terminal; Step 2: The drone moves to the first shooting position, starts the well end detection device, and tests whether the shooting effect and signal transmission are good through the images taken by the mobile phone; Step 3: Control the drone to shoot the shaft according to the predetermined shooting route; Step 4: The drone moves to the next shooting position and continues to shoot the shaft according to the predetermined shooting route; Step 5: Repeat step 4 until one side of the shaft is photographed, and obtain n pictures of one side of the shaft; Step 6: Use the "stitching" option of the detection software to pre-process the n images obtained in step 5; Step 7: stitch the n pre-processed images to obtain a complete image of one side of the shaft, and send the complete image to the mobile terminal; Step 8: The staff receives the complete image on the mobile terminal. After confirming that the image quality is good, they click the "Disease Detection" option in the detection software to control the wellhead detection device to perform different types of disease detection. The wellhead detection device processes multiple detection results to obtain image mask data and sends it back to the mobile terminal; Step 9: The staff receives the image mask data on the mobile terminal and forms a disease distribution cloud map; Step 10: After completing the concrete surface defect detection on one side of the shaft, repeat steps 2 to 9 to detect the next side of the lower shaft until all sides of the shaft are detected.

6. The method for detecting defects on the outer surface of a shaft based on image recognition according to claim 5 is characterized in that: In step 3, the predetermined shooting route is: shooting from low to high from one side of the shaft in a serpentine direction, first horizontally and then vertically.

7. The method for detecting defects on the outer surface of a shaft based on image recognition according to claim 5 is characterized in that: In step 3, the overlap rate Or of the two pictures taken continuously by the camera is higher than 30%.

8. The method for detecting defects on the outer surface of a shaft based on image recognition according to claim 5 is characterized in that: In step 8, different types of disease detection processes include concrete crack detection, reinforcement leakage detection, and concrete spalling and honeycomb surface detection.

9. The method for detecting defects on the outer surface of a shaft based on image recognition according to claim 5 is characterized in that: In step 9, different types of disease detection results are first classified, and then the disease detection results are merged according to the disease severity coefficients Wj of the different types of disease detection results.

Images