A fiber optic cable defect detection device based on artificial intelligence computer vision processing

Abstract

This invention proposes a fiber optic cable defect detection device based on artificial intelligence computer vision processing, comprising: a drone (1): a processor and storage device are fixedly connected inside the drone (1); cameras (4): the cameras (4) are arranged in pairs and fixedly installed at the bottom of the drone (1); a searchlight (5): the searchlight (5) is fixedly installed at the bottom of the drone (1); the cameras are connected to the processor to form a machine vision component for shooting the fiber optic cable from different angles, and the processor analyzes the images captured by the machine vision component through the artificial intelligence computer vision processing algorithm of the drone to detect fiber optic cable defects; this invention can detect defects in fiber optic cables erected at high altitudes through the machine vision algorithm built into the drone, and automatically mark defect points in the fiber optic cable images according to the construction factors and actual aging factors of the fiber optic cable at high altitudes.

Description

Technical Field

[0001] This invention relates to the field of optical cable inspection technology, and in particular to an optical cable defect detection device based on artificial intelligence computer vision processing. Background Technology

[0002] With the rapid growth of the national economy and the continuous improvement of people's living standards, information technology is gradually changing our lives. The development and changes of information-based life are inseparable from optical fiber, which can transmit information through light.

[0003] Currently, commonly used optical cable defect detection technologies can only detect unused optical cables. Unused optical cables generally undergo factory and incoming inspections, and they usually have fewer defects. However, in actual operation, some optical cables installed at high altitudes may develop surface cracks due to wind and sun exposure. But because the optical cables are installed at high altitudes, it is inconvenient for maintenance personnel to perform defect detection on optical cables installed at high altitudes. Summary of the Invention

[0004] This invention proposes a fiber optic cable defect detection device based on artificial intelligence computer vision processing. It can detect defects in fiber optic cables erected at high altitudes using machine vision algorithms built into drones, and automatically mark defect points in fiber optic cable images based on construction factors and actual aging factors of the fiber optic cables at high altitudes.

[0005] The present invention adopts the following technical solution.

[0006] A fiber optic cable defect detection device based on artificial intelligence computer vision processing includes:

[0007] Unmanned Aerial Vehicle (1): The unmanned aerial vehicle (1) has a processor and storage device fixedly connected inside its body;

[0008] Camera (4): The cameras (4) are arranged in pairs and are fixedly installed on the bottom of the drone (1);

[0009] Searchlight (5): The searchlight (5) is fixedly installed on the bottom of the UAV (1);

[0010] The camera is connected to the processor to form a machine vision component for shooting the optical cable from different angles. The processor analyzes the images captured by the machine vision component using the artificial intelligence computer vision processing algorithm of the drone to detect defects in the optical cable.

[0011] The drone includes:

[0012] Cover (2): The cover (2) is snapped onto the top of the UAV (1);

[0013] Support (3): Two supports (3) are provided, and the two supports (3) are fixedly installed on the bottom of the drone (1);

[0014] The bracket (3) is taller than the camera (4) and the searchlight (5).

[0015] There are two cameras (4), and the two cameras (4) are positioned between the bracket (3).

[0016] The artificial intelligence computer vision processing algorithm is built into the drone, and its operation steps include:

[0017] Step S1, Optical cable image acquisition: Control the drone (1) to fly above the optical cable and take pictures of the optical cable from different angles through two cameras (4);

[0018] Step S2, Image Processing: The optical cable data captured by the camera (4) is transmitted to the processor of the drone (1);

[0019] Step S3, Defect Detection: The UAV (1) processor analyzes the photos taken by the camera (4);

[0020] Step S4, Defect Marking: The UAV (1) digitally marks the defect locations on the optical cable;

[0021] Step S5, Defect Report: The UAV (1) processor diagnoses and classifies the marked defects and generates a defect report.

[0022] In step S1, the searchlight is a lamp that can emit light of different colors. When the camera of the machine vision component takes a picture of the optical cable, the searchlight illuminates the optical cable.

[0023] The image processing in step S2 specifically includes the following steps:

[0024] Step S2.1: Perform pixel brightness transformation on the original image captured by the camera (4), compare the image after brightness transformation with the original image, and mark the points with large differences in brightness of the optical cable image;

[0025] Step S2.2: Perform grayscale transformation on the original image captured by the camera (4), and then overlay the grayscale transformed image with the original image to mark the positions of the differences on the optical cable image;

[0026] Step S2.3: Compare the marked points of the brightness transformation image with the marked points of the grayscale transformation image, and mark the duplicate marked points.

[0027] In step S3, visual inspection is performed on the marked duplicate points to determine the type of optical cable defect.

[0028] The method for determining the type of optical cable defect includes:

[0029] Method A1: Assume that when the two ends of an optical cable erected at a high place are pulled, they will straighten and there will be no bends in the optical cable. Moreover, the outer sheath of a normal optical cable is smooth. Therefore, the normal optical cable surface photo taken by the camera should be smooth. The two cameras of the machine vision component take pictures of the optical cable from different angles. When the optical cable surface photos taken by the two cameras are not smooth and the difference is greater than a threshold, it indicates that there is an abnormality on the surface of the optical cable.

[0030] Method A2: To further clarify the imaging of surface defects in optical cables, different colored lights are used to illuminate the optical cables with searchlights. This allows the photos taken after different colored lights illuminate the surface of the optical cables to more clearly show the surface defects, avoiding the misinterpretation of some lighting errors as optical cable abnormalities. When there are protrusions and cracks on the surface of the optical cable, the angle of light reflection on the optical cable is different. The light reflected is stronger on smooth areas of the optical cable surface and dimmer on cracked areas. The light reflected is stronger on protrusions and dimmer on depressions.

[0031] Method A3: The captured fiber optic cable photos are processed by a drone processor to remove the influence of light on the fiber optic cable surface, preserving as much as possible the influence of the fiber optic cable's structure on the light in the photos. Defect analysis is then performed on the fiber optic cable photos. The processor has a built-in database of fiber optic cable surface defects. Different defects on the fiber optic cable surface correspond to different photos in the database. When determining the type of fiber optic cable defect, a comparison threshold X is first set. The captured photo is compared with a stored photo. If the difference between the captured photo and the stored photo is lower than the comparison threshold X, the captured photo is determined to have the defect corresponding to the stored photo. If the difference is higher than the comparison threshold X, a new stored photo is selected for comparison. If all stored photo comparisons do not match, the fiber optic cable photo image is retained for subsequent manual processing. The defects of the captured fiber optic cable are determined by similarity. After identifying the defects, the quantity and type of the fiber optic cable defects are marked. Then, a fiber optic cable defect report is generated based on indicators such as quantity, type, and severity. Simultaneously, manual intervention is used to perform feature recognition on unidentified fiber optic cable defect images. The identified fiber optic cable defects and their corresponding images are imported into the photo database to enrich the database.

[0032] Method A4: The method for visually inspecting the repeated marking points includes grayscale transformation and brightness transformation. Grayscale transformation and brightness transformation are used to eliminate the influence of light on the surface of the optical cable. If repeated marking points still exist in the optical cable surface image after processing, the area where the marking point is located is determined to be a defect area caused by the optical cable structure itself. Defects in the defect area include optical cable sheath dents, sheath bulges, or sheath ruptures. The type of optical cable defect is then determined by the optical cables corresponding to different optical cable sheath structures. The types of optical cable defects include cracked optical cable sheath, abnormal unevenness of the sheath, and bending of the sheath.

[0033] In step S4, after determining the type of optical cable defect and the corresponding optical cable image, a database of optical cable images and optical cable defects is established and uploaded to the storage device. The algorithm learning module of the artificial intelligence computer vision processing algorithm uses the database as a basis to compare the subsequently captured images with the images in the database to help determine the type of optical cable defect.

[0034] In step S4, when designing the algorithm learning module for the artificial intelligence computer vision processing algorithm, a relevant algorithm is first established based on existing optical cable defects and their corresponding images. During algorithm establishment, optical cables with different defect types are manually selected, such as those with convex surfaces, concave surfaces, or cracked surfaces. Images of these optical cables with different defects are then captured, and the structural features of the defective parts in the images are obtained. A connection is established between the structural features of the defective parts and the optical cable defects themselves, thereby creating a database of optical cable images and their corresponding defects. This database is then uploaded to the drone's storage device. The algorithm learning module uses this database as a basis to compare subsequently captured images with images in the database to assist in determining the type of optical cable defect, effectively reducing the time required for optical cable defect assessment and accelerating the generation of defect reports.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] This invention utilizes a drone equipped with a camera and searchlight on its underside. When personnel need to inspect high-altitude fiber optic cables, the drone can be controlled to fly above the cables, and the camera will then photograph them. By comparing the images taken by the two cameras, significant defects in the cables can be effectively identified. When the colored searchlight illuminates the cables, differences in surface images are observed when cracks or protrusions are present. Further brightness and grayscale transformations of the images allow for comparison of the original and processed images, effectively highlighting surface differences. These differences can be marked for subsequent processing to determine if they constitute defects. By leveraging drones and computer vision algorithms, large numbers of cables can be inspected quickly and effectively, requiring only initial manual inspection. As the drone's database expands, automatic defect identification becomes possible, saving time and effort and significantly improving the efficiency of high-altitude fiber optic cable maintenance.

[0037] In this invention, the UAV only needs to make a simple judgment on the optical cable and mark the image. At the same time, the algorithm design fully considers the possible defects of optical cable construction and long-term use. Therefore, the detection target is simple and easy to identify, and the computing performance requirements of the UAV are not high. The detection operation can be completed directly on the UAV without the need for data network link support, which facilitates the implementation and deployment of the solution. Attached Figure Description

[0038] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0039] Appendix Figure 1 This is a schematic diagram of the overall structure of the UAV of the present invention;

[0040] Appendix Figure 2 This is another schematic diagram of the overall structure of the UAV from the perspective of the present invention;

[0041] Appendix Figure 3 This is a schematic diagram of the optical cable defect detection method of the present invention;

[0042] In the picture: 1. Drone; 2. Canopy; 3. Support frame; 4. Camera; 5. Searchlight. Detailed Implementation

[0043] As shown in the figure, a fiber optic cable defect detection device based on artificial intelligence computer vision processing includes:

[0044] Unmanned Aerial Vehicle 1: A processor and storage device are fixedly connected inside the body of the unmanned aerial vehicle 1;

[0045] Camera 4: The cameras 4 are arranged in pairs and are fixedly installed on the bottom of the drone 1;

[0046] Searchlight 5: The searchlight 5 is fixedly installed on the bottom of the UAV 1;

[0047] The camera is connected to the processor to form a machine vision component for shooting the optical cable from different angles. The processor analyzes the images captured by the machine vision component using the artificial intelligence computer vision processing algorithm of the drone to detect defects in the optical cable.

[0048] The drone includes:

[0049] Cover 2: The cover 2 is snapped onto the top of the drone 1;

[0050] Support 3: Two supports 3 are provided, and the two supports 3 are fixedly installed on the bottom of the drone 1;

[0051] The height of the bracket 3 is greater than the height of the camera 4 and the searchlight 5.

[0052] There are two cameras 4, and the two cameras 4 are arranged between the bracket 3.

[0053] The artificial intelligence computer vision processing algorithm is built into the drone, and its operation steps include:

[0054] Step S1, Optical Cable Image Acquisition: Control the drone 1 to fly above the optical cable and take pictures of the optical cable from different angles through the two cameras 4;

[0055] Step S2, Image Processing: The optical cable data captured by camera 4 is transmitted to the processor of drone 1;

[0056] Step S3, Defect Detection: The processor of UAV 1 analyzes the photos taken by camera 4;

[0057] Step S4, Defect Marking: UAV 1 digitally marks the locations of defects on the optical cable;

[0058] Step S5, Defect Report: The UAV 1 processor diagnoses and classifies the marked defects and generates a defect report.

[0059] In step S1, the searchlight is a lamp that can emit light of different colors. When the camera of the machine vision component takes a picture of the optical cable, the searchlight illuminates the optical cable.

[0060] The image processing in step S2 specifically includes the following steps:

[0061] Step S2.1: Perform pixel brightness transformation on the original image captured by camera 4, compare the image after brightness transformation with the original image, and mark the points with large differences in brightness in the optical cable image;

[0062] Step S2.2: Perform grayscale transformation on the original image captured by camera 4, and then overlay the grayscale transformed image with the original image to mark the positions of the differences on the optical cable image;

[0063] Step S2.3: Compare the marked points of the brightness transformation image with the marked points of the grayscale transformation image, and mark the duplicate marked points.

[0064] In step S3, visual inspection is performed on the marked duplicate points to determine the type of optical cable defect.

[0065] The method for determining the type of optical cable defect includes:

[0066] Method A1: Assume that when the two ends of an optical cable erected at a high place are pulled, they will straighten and there will be no bends in the optical cable. Moreover, the outer sheath of a normal optical cable is smooth. Therefore, the normal optical cable surface photo taken by the camera should be smooth. The two cameras of the machine vision component take pictures of the optical cable from different angles. When the optical cable surface photos taken by the two cameras are not smooth and the difference is greater than a threshold, it indicates that there is an abnormality on the surface of the optical cable.

[0067] Method A2: To further clarify the imaging of surface defects in optical cables, different colored lights are used to illuminate the optical cables with searchlights. This allows the photos taken after different colored lights illuminate the surface of the optical cables to more clearly show the surface defects, avoiding the misinterpretation of some lighting errors as optical cable abnormalities. When there are protrusions and cracks on the surface of the optical cable, the angle of light reflection on the optical cable is different. The light reflected is stronger on smooth areas of the optical cable surface and dimmer on cracked areas. The light reflected is stronger on protrusions and dimmer on depressions.

[0068] Method A3: The captured fiber optic cable photos are processed by a drone processor to remove the influence of light on the fiber optic cable surface, preserving as much as possible the influence of the fiber optic cable's structure on the light in the photos. Defect analysis is then performed on the fiber optic cable photos. The processor has a built-in database of fiber optic cable surface defects. Different defects on the fiber optic cable surface correspond to different photos in the database. When determining the type of fiber optic cable defect, a comparison threshold X is first set. The captured photo is compared with a stored photo. If the difference between the captured photo and the stored photo is lower than the comparison threshold X, the captured photo is determined to have the defect corresponding to the stored photo. If the difference is higher than the comparison threshold X, a new stored photo is selected for comparison. If all stored photo comparisons do not match, the fiber optic cable photo image is retained for subsequent manual processing. The defects of the captured fiber optic cable are determined by similarity. After identifying the defects, the quantity and type of the fiber optic cable defects are marked. Then, a fiber optic cable defect report is generated based on indicators such as quantity, type, and severity. Simultaneously, manual intervention is used to perform feature recognition on unidentified fiber optic cable defect images. The identified fiber optic cable defects and their corresponding images are imported into the photo database to enrich the database.

[0069] Method A4: The method for visually inspecting the repeated marking points includes grayscale transformation and brightness transformation. Grayscale transformation and brightness transformation are used to eliminate the influence of light on the surface of the optical cable. If repeated marking points still exist in the optical cable surface image after processing, the area where the marking point is located is determined to be a defect area caused by the optical cable structure itself. Defects in the defect area include optical cable sheath dents, sheath bulges, or sheath ruptures. The type of optical cable defect is then determined by the optical cables corresponding to different optical cable sheath structures. The types of optical cable defects include cracked optical cable sheath, abnormal unevenness of the sheath, and bending of the sheath.

[0070] In step S4, after determining the type of optical cable defect and the corresponding optical cable image, a database of optical cable images and optical cable defects is established and uploaded to the storage device. The algorithm learning module of the artificial intelligence computer vision processing algorithm uses the database as a basis to compare the subsequently captured images with the images in the database to help determine the type of optical cable defect.

[0071] In step S4, when designing the algorithm learning module for the artificial intelligence computer vision processing algorithm, a relevant algorithm is first established based on existing optical cable defects and their corresponding images. During algorithm establishment, optical cables with different defect types are manually selected, such as those with convex surfaces, concave surfaces, or cracked surfaces. Images of these optical cables with different defects are then captured, and the structural features of the defective parts in the images are obtained. A connection is established between the structural features of the defective parts and the optical cable defects themselves, thereby creating a database of optical cable images and their corresponding defects. This database is then uploaded to the drone's storage device. The algorithm learning module uses this database as a basis to compare subsequently captured images with images in the database to assist in determining the type of optical cable defect, effectively reducing the time required for optical cable defect assessment and accelerating the generation of defect reports.

[0072] In this example, the camera has image stabilization.

[0073] In this example, when the camera takes pictures of the optical cable, it takes pictures corresponding to different light colors according to the different colors of the searchlight, so that the defects on the surface of the optical cable can be more easily identified under different light colors.

[0074] In this example, the drone's processor includes a flight control module. The flight control module determines the position and distance of the optical cable based on the images captured by the camera of the machine vision component. In the absence of a remote control signal, it automatically controls the drone to fly along the optical cable erected in the high altitude, and keeps the drone's flight position above the optical cable and at the same distance from the optical cable. In other words, the high-altitude optical cable is used as the flight path marker to ensure that the images captured by the machine vision component are more standardized.

Claims

1. A fiber optic cable defect detection device based on artificial intelligence computer vision processing, characterized in that: include: Unmanned Aerial Vehicle (1): The unmanned aerial vehicle (1) has a processor and storage device fixedly connected inside its body; Camera (4): The cameras (4) are arranged in pairs and are fixedly installed on the bottom of the drone (1); Searchlight (5): The searchlight (5) is fixedly installed on the bottom of the UAV (1); The camera is connected to the processor to form a machine vision component for shooting the optical cable from different angles. The processor analyzes the images captured by the machine vision component through the artificial intelligence computer vision processing algorithm of the drone to detect defects in the optical cable. The artificial intelligence computer vision processing algorithm is built into the drone, and its operation steps include: Step S1, Optical cable image acquisition: Control the drone (1) to fly above the optical cable and take pictures of the optical cable from different angles through two cameras (4); Step S2, Image Processing: The optical cable data captured by the camera (4) is transmitted to the processor of the drone (1); Step S3, Defect Detection: The UAV (1) processor analyzes the photos taken by the camera (4); Step S4, Defect Marking: The UAV (1) digitally marks the location of defects on the optical cable; Step S5, Defect Report: The UAV (1) processor diagnoses and classifies the marked defects and generates a defect report; In step S1, the searchlight is a lamp that can emit light of different colors. When the camera of the machine vision component takes a picture of the optical cable, the searchlight illuminates the optical cable. The image processing in step S2 specifically includes the following steps: Step S2.1: Perform pixel brightness transformation on the original image captured by the camera (4), compare the image after brightness transformation with the original image, and mark the points with large differences in brightness of the optical cable image; Step S2.2: Perform grayscale transformation on the original image captured by the camera (4), and then overlay the grayscale transformed image with the original image to mark the positions of the differences on the optical cable image; Step S2.3: Compare the marked points of the brightness transformation image with the marked points of the grayscale transformation image, and mark the duplicate marked points; In step S3, the marked duplicate points are visually inspected to determine the type of optical cable defect. The method for visually inspecting the repeated markers includes grayscale transformation and brightness transformation. Grayscale transformation and brightness transformation are used to eliminate the influence of light on the surface of the optical cable. If repeated markers still exist on the surface of the optical cable after processing, the area where the marker is located is determined to be a defect area caused by the optical cable structure itself. Defects in the defect area include optical cable sheath dents, sheath bulges, and sheath ruptures. The type of optical cable defect is then determined by the optical cables corresponding to different optical cable sheath structures. The types of optical cable defects include cracked outer sheath, abnormal unevenness of the sheath, and bending of the sheath.

2. The optical cable defect detection device based on artificial intelligence computer vision processing according to claim 1, characterized in that: The drone includes: Cover (2): The cover (2) is snapped onto the top of the UAV (1); Support (3): Two supports (3) are provided, and the two supports (3) are fixedly installed on the bottom of the drone (1); The bracket (3) is taller than the camera (4) and the searchlight (5).

3. The optical cable defect detection device based on artificial intelligence computer vision processing according to claim 2, characterized in that: There are two cameras (4), and the two cameras (4) are positioned between the bracket (3).

Images