Point cloud image fusion-based power transmission line channel hidden danger distance measurement method

Abstract

The invention discloses a power transmission line channel hidden danger distance measurement method based on point cloud image fusion. The method comprises the steps of data joint calibration, target recognition and modeling of a power transmission line channel scene, target segmentation and positioning of the power transmission line channel scene and judgment of the distance between a hidden danger target and a wire. The method has the advantages that the point cloud data and the image data are fused, the space coordinate conversion relation between the point cloud data and the image data is calibrated and solved, the point cloud data are projected to the image view angle to generate a new point cloud projection mapping graph, training recognition is conducted through the image data and the point cloud data, the recognition accuracy of the hidden danger target is improved, the SAM model serves as a basic model, and the recognition efficiency is improved. The point cloud projection mapping graph and the hidden danger target in the image data are used as samples, training is carried out, a hidden danger target segmentation model is generated, accurate contour information of the hidden danger target is extracted, and effective early warning information is provided for the safety of a power transmission line channel.

Description

Technical Field

[0001] The present invention relates to a transmission line channel hidden danger distance measurement method, in particular to a transmission line channel hidden danger distance measurement method based on point cloud image fusion, belonging to the technical field of transmission line hidden danger distance measurement. Background Art

[0002] Generally speaking, traditional manual inspections are carried out manually. Due to the heavy workload and high difficulty, inadequate inspections and substandard inspections are common. In addition, the inspection cycle is long and there is a regulatory gap.

[0003] In the prior art, the commonly used inspection method is to install multiple monitoring devices at fixed positions on the transmission tower for inspection and snapshot, identify potential hazards in the image through deep learning methods, and have the inspection personnel judge the safety level based on the distance between the potential hazards in the image and the transmission wire. This inspection method has two defects: first, it cannot automatically calculate the distance and requires manual intervention, which increases labor costs; second, there is no depth information on the monitoring screen, and the distance estimated by the operation and maintenance personnel is obviously not accurate enough.

[0004] The comprehensive deployment and in-depth application of transmission line visualization devices play an increasingly important role in operation and maintenance control. However, the visualization devices can qualitatively identify external damage hazards such as construction vehicles, but cannot perceive the size, direction, position and other information of external damage hazards, and cannot calculate their clearance distance to the conductors in real time. There are a large number of hidden danger alarms, but the operation and maintenance personnel cannot judge the threat level of the hidden dangers in time, and cannot accurately warn and conduct targeted operation and maintenance control according to the hidden danger level.

[0005] In the prior art, there is an automatic inspection method for power transmission channels based on the fusion of visible light and laser radar point clouds, such as the one disclosed in announcement number CN115240093A. The transmission channel is photographed from the air to collect three-dimensional point clouds and two-dimensional visible light image data. The data sets are labeled and constructed according to the three-dimensional point cloud segmentation task and the two-dimensional tower hidden danger target detection task, and model training is performed. The three-dimensional point cloud segmentation model PCI-Seg and the tower hidden danger target detection model YOLOV5 are generated. The tower position is obtained through the point cloud segmentation result. When encountering the tower, the drone flies around the tower and shoots two-dimensional visible light images and three-dimensional point clouds through cameras and laser scanning radars. The shooting points on the tower are located through three-dimensional point cloud segmentation and two-dimensional images are shot at the shooting points. The image information and shooting point location information are transmitted back to the server. The tower hidden danger target detection model is used to detect whether there are hidden dangers in the tower in the picture. Through the point cloud segmentation results, check whether there are hidden dangers of external damage in the transmission channel. If there are, the distance between the hidden danger and the tower and the main body of the transmission channel is calculated through the position information of the laser radar point cloud to determine whether there is an invasion of the transmission channel. This method directly determines the distance of the hidden danger target through the drone point cloud, takes pictures after locating the tower through point cloud segmentation, and then recognizes the tower body target based on the image, which requires the drone to fly in real time. The drone inspection method is greatly affected by flight environment factors (radio environment, meteorological environment, geographical environment), the amount of data collected at a single time is limited, the collection effectiveness is low, and the operating cost is high.

[0006] A method and terminal for calculating the distance of hidden dangers in transmission lines disclosed in announcement number CN114066985A obtains two-dimensional image data and three-dimensional point cloud data of the transmission line channel, uses a depth estimation network to perform depth estimation on the two-dimensional image, and obtains a corresponding depth map. A target contour extraction network is used to extract contours from the two-dimensional image and the depth map to obtain a two-dimensional hidden danger contour. Based on ground marker points, two-dimensional image data and three-dimensional point cloud data are collected in advance to establish a ground mapping model. Using the ground mapping model, the two-dimensional image contour is converted into the corresponding three-dimensional hidden danger contour. The transmission line location information is determined based on the three-dimensional point cloud data, and the shortest distance from the hidden danger target contour to the transmission line is calculated. This method estimates the depth of the target through a depth estimation grid, extracts the contour and performs conversion, and does not make full use of the point cloud information.

[0007] A method for measuring the hidden danger of mechanical damage in power transmission disclosed in announcement number CN116091494A establishes a monocular depth estimation model, a transmission line segmentation model, and a transmission line mechanical damage hidden danger target detection model; uses the monocular depth estimation model to perform monocular depth estimation of scene scenery on the visible light picture of the transmission channel to obtain a depth map of the visible light picture of the transmission channel; uses the transmission line segmentation model to perform image segmentation on the wires in the visible light picture of the transmission channel to obtain the wire segmentation pixel position; combines the wire segmentation pixel position with the depth map of the visible light picture of the transmission channel to obtain the wire depth information; converts the wire depth information into a wire point cloud through a camera intrinsic parameter matrix; uses a transmission line mechanical damage hidden danger target detection model to obtain a mechanical damage hidden danger frame in the visible light picture of the transmission channel; obtains the plane position information of the representative point set of the mechanical damage hidden danger target; uses the plane position information to obtain the spatial position information of the representative point set of the mechanical damage hidden danger target through the camera imaging principle, and calculates the minimum distance between the mechanical damage hidden danger and the transmission line with the wire point cloud. This method uses a monocular depth estimation model to perform depth estimation and three-dimensional conversion. The monocular cannot directly obtain depth information, and the estimation accuracy is limited.

[0008] Based on this, the present application proposes a method for measuring hidden dangers in transmission line channels based on point cloud image fusion. Summary of the invention

[0009] The purpose of the present invention is to provide a method for measuring hidden dangers in a transmission line channel based on point cloud image fusion in order to solve at least one of the above technical problems.

[0010] The present invention achieves the above-mentioned purpose through the following technical scheme: a transmission line channel hidden danger ranging method based on point cloud image fusion, the transmission line channel hidden danger ranging method comprises the following steps:

[0011] S1. Data joint calibration: obtain image data (2D image) through a monocular camera and point cloud data (3D point cloud) through a laser point cloud. Take advantage of the dense information of the monocular camera and the accuracy of the laser point cloud to fuse the image data with the point cloud data, and calibrate and solve the spatial coordinate transformation relationship between the two.

[0012] S2, target recognition and modeling of transmission line channel scenes, using spatial coordinate transformation relationships to generate point cloud projection mapping from the image perspective;

[0013] S3. Target segmentation and positioning of transmission line channel scenes. The hidden danger target segmentation model is trained based on the SAM model deep learning method. The image data and point cloud data are used for training and recognition to improve the recognition accuracy of hidden danger targets. The hidden danger targets in the point cloud projection map and the original two-dimensional image (image data) are used as samples for fine-tuning training. A hidden danger target segmentation model dedicated to transmission line channels is generated to extract the accurate contour information of hidden danger targets and realize the positioning of hidden danger targets.

[0014] S4. Determine the distance between the hidden danger target and the conductor. By calculating the distance between the hidden danger target and the conductor, the danger level of the hidden danger target is determined to provide early warning information for the safety of the transmission line channel.

[0015] As a further solution of the present invention: data joint calibration specifically includes:

[0016] S11, acquiring point cloud data of a transmission line channel scene, and correspondingly acquiring image data of the transmission line channel scene;

[0017] S12, camera calibration, calibrating the camera by Zhang Zhengyou calibration method to obtain the camera internal parameters;

[0018] S13, reading the point cloud data, adjusting the viewing angle of the point cloud data to be consistent with the viewing angle of the image data, selecting feature points in the transmission line channel scene based on the image data, and finding the corresponding feature point positions in the point cloud data;

[0019] S14, using the coordinates of the point cloud data as world coordinate input, combining the coordinates of the point cloud data with the pixel coordinates of the camera by using a camera pose estimation algorithm, and obtaining external parameters of the point cloud data and the camera rotation and translation;

[0020] S15. Calculate and output the spatial coordinate transformation relationship between the two based on the joint calculation of the camera's internal and external parameters.

[0021] As a further solution of the present invention: the spatial coordinate transformation relationship is described as follows:

[0022] The origin coordinate system of the point cloud data is used as the unified world coordinate system; it is defined as Xw, Yw, Zw, and the unit is the length unit;

[0023] The camera coordinate system uses the optical center as the origin of the camera coordinate system, the x and y directions parallel to the image data as the Xc axis and Yc axis, the Zc axis is parallel to the optical axis, the Xc axis, Yc axis and Zc axis are perpendicular to each other, and the unit is the length unit.

[0024] As a further solution of the present invention: the pixel coordinate system of the image data takes the vertex of the image data as the coordinate origin, the u and v directions are parallel to the x and y directions, and the unit is pixel;

[0025] The final spatial coordinate transformation relationship is:

[0026]

[0027] Where R is the rotation matrix (3 degrees of freedom), t is the translation matrix, and the two form a 3×4 matrix, which is the external parameter matrix of the camera; f x , f y , c x , c y is the internal parameter of the camera, where f x ,f y is the focal length parameter of the camera, c x ,c y is the camera optical center parameter.

[0028] As a further solution of the present invention: target recognition and modeling of the transmission line channel scene specifically includes:

[0029] S21, collecting image data and point cloud data of the transmission line channel scene;

[0030] S22, marking potential danger targets in the image data;

[0031] S23, using the calibration fusion relationship, generating a point cloud projection mapping diagram under the image perspective from the corresponding point cloud data;

[0032] S24, automatically annotating the point cloud projection mapping image using the image annotation information;

[0033] S25, generating fusion sample data;

[0034] S26. Use YOLOV5 to train hidden danger targets and generate a target recognition model for the transmission line channel scene.

[0035] As a further solution of the present invention: target segmentation and positioning of the power transmission line channel scene specifically includes:

[0036] S31. Construct a hidden danger sample set of the transmission line channel scenario, which includes but is not limited to large construction machinery, poles, tower cranes, cranes, trees or building targets;

[0037] S32, using the SAM model to build a target recognition and segmentation network for the transmission line channel scene;

[0038] S33. Train and output a recognition, segmentation and positioning model for hidden danger targets.

[0039] As a further solution of the present invention: the SAM model consists of the following three parts:

[0040] The image encoder, which is responsible for processing the image and creating an embedding that represents the image; this part consists of the VIT transformer and is the largest component of the network;

[0041] The prompt encoder handles additional input to the network;

[0042] The mask decoder receives the output of the image encoder and the hint encoder and generates the final segmentation mask.

[0043] As a further solution of the present invention: the determination of the distance between the hidden danger target and the wire specifically includes:

[0044] S41, hidden danger target detection, obtaining the location information of the hidden danger target in the image data and point cloud data;

[0045] S42, importing the recognition segmentation positioning model according to the recognition position frame to accurately extract the target contour information;

[0046] S43, obtaining the lowest point and the highest point of the hidden danger target, using the calibration fusion relationship, reverse calculation, and obtaining the corresponding three-dimensional position information;

[0047] S44, calculating the clearance distance between the hidden danger target and the conductor;

[0048] S45. Determine the threat level of hidden dangers at the target based on the clearance distance and the voltage level of the transmission line.

[0049] The beneficial effects of the present invention are:

[0050] 1) The present invention comprehensively utilizes image data and point cloud data, projects the point cloud data to a two-dimensional visualization perspective, generates a point cloud projection map of a new image perspective, and utilizes the fusion calibration relationship to project the point cloud projection map to the image data perspective. After preprocessing, a new point cloud projection map is generated, and the image data and the new point cloud projection map are used for target training to improve the recognition effect of hidden danger targets;

[0051] 2) The invention is based on the accurate segmentation and positioning method of hidden danger targets of the point cloud projection map and the original image fine-tuned by the SAM model. The SAM model is used as the basic model, and the hidden danger targets in the point cloud projection map and the original two-dimensional image are used as samples for fine-tuning training to generate a hidden danger target segmentation model for the dedicated transmission line channel scene, and extract accurate contour information;

[0052] 3) The present invention comprehensively utilizes the advantages of dense information of images and precision of laser point cloud, optimizes the accurate identification and positioning method of hidden danger targets, fuses point cloud data with image data, calibrates and solves the spatial coordinate transformation relationship between the two, and adds point cloud projection mapping as a training sample on the basis of existing image data recognition to realize the hidden danger distance measurement function, which can improve the accurate distance measurement of hidden danger targets, thereby further judging the danger level of hidden dangers and providing more effective early warning information for the safety of transmission line channels. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0054] Figure 2 It is a schematic diagram of the data joint calibration process of the present invention;

[0055] Figure 3 A schematic diagram of a monocular image and a point cloud viewing angle of the present invention;

[0056] Figure 4 Schematic diagram of the pixel coordinate system structure of image data of the present invention;

[0057] Figure 5 This is a schematic diagram of precise distance measurement for construction vehicles in a certain line channel in the second embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Embodiment 1, as Figure 1 As shown, this embodiment provides a transmission line channel hidden danger distance measurement method based on point cloud image fusion, and the transmission line channel hidden danger distance measurement method includes the following steps:

[0060] First: joint data calibration. The image data (2D image) obtained by the monocular camera and the point cloud data (3D point cloud) obtained by the laser point cloud are fused by taking advantage of the dense information of the monocular camera and the precision of the laser point cloud, and the spatial coordinate transformation relationship between the two is calibrated and solved.

[0061] Data joint calibration specifically includes:

[0062] 1) Obtaining point cloud data of the transmission line channel scene, and correspondingly obtaining image data of the transmission line channel scene;

[0063] 2) Camera calibration: Use Zhang Zhengyou calibration method to calibrate the camera and obtain the camera internal parameters;

[0064] 3) Read the point cloud data, adjust the point cloud data perspective to be consistent with the image data perspective, and select the feature points in the transmission line channel scene based on the image data. At the same time, find the corresponding feature point positions in the point cloud data and select 10 groups of point pairs. The schematic diagram of the monocular image and point cloud perspective is shown in the figure. Figure 3 As shown;

[0065] 4) Using the coordinates of the point cloud data as the world coordinate input, the coordinates of the point cloud data are combined with the pixel coordinates of the camera by using the camera pose estimation algorithm to obtain the external parameters of the point cloud data and the camera rotation and translation;

[0066] 5) Based on the joint calculation of the camera's internal and external parameters, the 3D-2D spatial coordinate transformation relationship is calculated and output.

[0067] Among them, the 3D-2D spatial coordinate transformation relationship is described as follows:

[0068] 51) The origin coordinate system of the point cloud data is used as the unified world coordinate system, defined as Xw, Yw, Zw, and the unit is the length unit;

[0069] 52) The camera coordinate system takes the optical center as the origin of the camera coordinate system, the x and y directions parallel to the image data as the Xc axis and Yc axis, the Zc axis is parallel to the optical axis, Xc, Yc and Zc are perpendicular to each other, and the unit is the length unit;

[0070] 53) The physical coordinate system of the image data takes the intersection of the principal optical axis and the image plane as the origin, and the x and y directions are as follows: Figure 4 As shown, the units are length units.

[0071] 54) The pixel coordinate system of the image data takes the vertex of the image as the coordinate origin, the u and v directions are parallel to the x and y directions, and the unit is pixel;

[0072] The final 3D-2D spatial coordinate transformation relationship is:

[0073]

[0074] Where: R is the rotation matrix (3 degrees of freedom), t is the translation matrix, and the two form a 3×4 matrix, which is the external parameter matrix of the camera; f x , f y , c x , c y is the internal parameter of the camera, where f x ,f y is the focal length parameter of the camera, c x ,cy is the camera optical center parameter.

[0075] Second: Target recognition and modeling of transmission line channel scenes. The 3D-2D spatial coordinate transformation relationship is used to generate a point cloud projection map from the image perspective, that is, the point cloud data is projected to the image perspective to generate a new point cloud projection map.

[0076] Among them, the target recognition and modeling of the transmission line channel scene specifically includes:

[0077] 1) Collect image data and point cloud data of transmission line channel scenes;

[0078] 2) Label hidden danger targets in image data;

[0079] 3) Using the calibration fusion relationship, the corresponding point cloud data is used to generate a point cloud projection mapping diagram under the image perspective;

[0080] 4) Automatically annotate the point cloud projection map using image annotation information;

[0081] 5) Generate fusion sample data;

[0082] 6) YOLOV5 is used to train hidden danger targets and generate a target recognition model for the transmission line channel scenario.

[0083] Third: Target segmentation and positioning in the transmission line channel scenario. The deep learning method based on the SAM model is used to train the hidden danger target segmentation model, and the image data and point cloud data are used for training and recognition to improve the recognition accuracy of hidden danger targets. The hidden danger targets in the point cloud projection map and the original two-dimensional image are used as samples for fine-tuning training to generate a hidden danger segmentation model dedicated to the transmission line channel, and the accurate contour information of the hidden danger target is extracted to achieve the positioning of the hidden danger target.

[0084] Among them, the scene target segmentation and positioning of the transmission line channel specifically includes:

[0085] 1) Construct a hidden danger sample set of the transmission line channel scenario, which includes but is not limited to large construction machinery, poles, tower cranes, cranes, trees and building targets;

[0086] 2) The SAM model is used to build an object recognition and segmentation network; the SAM model consists of three parts: 21) Image encoder, which is responsible for processing the image and creating an embedding representing the image. This part consists of VIT transformer and is the largest component of the network; 22) Prompt encoder, which processes additional inputs to the network; 23) Mask decoder, which receives the output of the image encoder and prompt encoder and generates the final segmentation mask;

[0087] 3) Train and output the recognition, segmentation and positioning model of hidden danger targets.

[0088] Fourth: Determine the distance between the hidden danger target and the conductor. By calculating the distance between the hidden danger target and the conductor, the danger level of the hidden danger target can be determined, providing more effective early warning information for the safety of the transmission line channel.

[0089] Among them, the determination of the distance between the hidden danger target and the wire specifically includes:

[0090] 1) Hidden danger target detection, obtaining the location information of hidden danger targets in image data and point cloud data;

[0091] 2) According to the identification position frame, the identification segmentation positioning model is imported to accurately extract the contour information of the hidden danger target;

[0092] 3) Obtain the lowest and highest points of the hidden danger target, use the calibration fusion relationship, reversely calculate, and obtain the three-dimensional position information of the hidden danger target;

[0093] 4) Calculate the clearance distance between the hidden danger target and the conductor;

[0094] 5) Determine the threat level of hidden dangers at the target based on the clearance distance and the voltage level of the transmission line.

[0095] Embodiment 2: This embodiment is a method for measuring hidden dangers in a transmission line channel based on point cloud image fusion in Embodiment 1.

[0096] Reasons for implementing this embodiment: With the rapid development of national infrastructure, the number of equipment tripping events caused by external damage such as construction vehicles hitting the line on the transmission line is increasing year by year. The original visual monitoring equipment does not have an accurate distance measurement function, and only recognizes the monitoring drawing, resulting in construction machinery outside the range of 500 meters from the line still reporting alarms, with a very high false alarm rate, accounting for more than 80% of false alarms per day, increasing the workload of monitoring personnel.

[0097] The specific implementation process of this embodiment is as follows: Figure 5 As shown in the figure, by studying the matching technology of monitoring video and laser point cloud data, the monitoring video and laser point cloud data are aligned and integrated to form a basic data base; the deep learning external damage and foreign object recognition technology, the non-contact measurement technology based on ordinary cameras, accurately judges the distance between the risk point and the conductor and tower, thereby reducing the visual false alarm rate and improving the level of line intelligence. The 500 kV XX line channel operated and maintained by the company has a new high-speed railway crossing construction. The original visual monitoring equipment relies on manual investigation and is difficult to achieve high-efficiency, high-precision, and timely risk investigation and control. After applying this technology, it can accurately measure the distance of all objects in the line channel, automatically analyze the source of danger, and achieve accurate control of external damage events, providing digital and intelligent capabilities for power grid inspection.

[0098] Working principle: According to the principle of visualization imaging, image data and point cloud data are integrated to establish the spatial coordinate conversion relationship between the two coordinate systems, forming a unified world coordinate system. The three-dimensional point cloud of the transmission line channel scene is used to reconstruct the monitoring system in three dimensions, and the relationship between image data and high-precision three-dimensional space scenes is established. For various types of construction vehicles such as cranes, excavators, bulldozers, pump trucks and pile drivers, the comprehensive data of point clouds and images are fully utilized to accurately identify and locate hidden danger targets, and to accurately calculate and measure hidden danger targets, thereby realizing graded alarms. In the event that the transmission line may be damaged by external forces, an alarm is issued immediately, and the disposal personnel are notified to respond in time; the accuracy and timeliness of observations on external force damage factors are improved; in this way, a large number of hidden danger alarms are issued, but the operation and maintenance personnel cannot judge the threat level of hidden dangers in time, and the hidden danger distance is used as a criterion to assist in screening alarm information and alleviate the current situation of many false alarms on the visualization platform.

[0099] 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 range of 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.

[0100] 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 method for measuring hidden dangers in a transmission line channel based on point cloud image fusion, characterized in that: The transmission line channel hidden danger distance measurement method comprises the following steps: S1. Data joint calibration: obtain image data through a monocular camera and obtain point cloud data through a laser point cloud, fuse the image data and the point cloud data, and calibrate and solve the spatial coordinate transformation relationship between the two; S2, target recognition and modeling of the transmission line channel scene, using the spatial coordinate transformation relationship to generate a point cloud projection mapping diagram under the image perspective; S3, target segmentation and positioning of the transmission line channel scene, using a deep learning method based on the SAM model to train a hidden danger target segmentation model, using the image data and the point cloud data for training and recognition, using the point cloud projection map and the hidden danger targets in the image data as samples, performing fine-tuning training, generating a hidden danger target segmentation model for the transmission line channel, extracting accurate contour information of the hidden danger target, and realizing the positioning of the hidden danger target; S4. Determine the distance between the hidden danger target and the conductor. By calculating the distance between the hidden danger target and the conductor, the danger level of the hidden danger target is determined to provide early warning information for the safety of the transmission line channel.

2. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 1, characterized in that: In S1, the data joint calibration specifically includes: S11, acquiring point cloud data of a transmission line channel scene, and correspondingly acquiring image data of the transmission line channel scene; S12, camera calibration, calibrating the camera by Zhang Zhengyou calibration method to obtain the camera internal parameters; S13, reading the point cloud data, adjusting the viewing angle of the point cloud data to be consistent with the viewing angle of the image data, and selecting feature points in the power transmission line channel scene based on the image data, and finding the corresponding feature point positions in the point cloud data; S14, using the coordinates of the point cloud data as world coordinate input, combining the coordinates of the point cloud data with the pixel coordinates of the camera by using a camera pose estimation algorithm, and obtaining external parameters of the point cloud data and the camera rotation and translation; S15. Calculate and output the spatial coordinate transformation relationship between the two based on the joint calculation of the camera's internal and external parameters.

3. The method for measuring hidden dangers in a transmission line channel according to claim 2, characterized in that: The spatial coordinate transformation relationship is described as follows: The origin coordinate system of the point cloud data is used as the unified world coordinate system; defined as Xw, Yw, Zw; The camera coordinate system uses the optical center as the origin of the camera coordinate system, the x and y directions parallel to the image data as the Xc axis and Yc axis, the Zc axis is parallel to the optical axis, and the Xc axis, Yc axis and Zc axis are perpendicular to each other.

4. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 3, characterized in that: The pixel coordinate system of the image data takes the vertex of the image data as the coordinate origin, the u and v directions are parallel to the x and y directions, and the unit is pixel; The final spatial coordinate transformation relationship is: Where: R is the rotation matrix with 3 degrees of freedom, t is the translation matrix, and the two form a 3×4 matrix, which is the external parameter matrix of the camera; f x , f y , c x , c y is the internal parameter of the camera, where f x ,f y is the focal length parameter of the camera, c x ,c y is the camera optical center parameter.

5. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 1, characterized in that: In S2, the target recognition and modeling of the transmission line channel scene specifically includes: S21, collecting image data and point cloud data of the transmission line channel scene; S22, marking potential danger targets in the image data; S23, using the calibration fusion relationship, generating a point cloud projection mapping diagram under the image perspective from the corresponding point cloud data; S24, automatically annotating the point cloud projection mapping image using the image annotation information; S25, generating fusion sample data; S26. Use YOLOV5 to train hidden danger targets and generate a target recognition model for the transmission line channel scene.

6. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 1, characterized in that: In S3, the target segmentation and positioning of the transmission line channel scene specifically includes: S31. Construct a hidden danger sample set of a transmission line channel scenario, wherein the hidden danger sample set includes but is not limited to large construction machinery, poles, tower cranes, cranes, trees or building targets; S32, using the SAM model to construct a target recognition and segmentation network for the transmission line channel scene; S33. Train and output a recognition, segmentation and positioning model for hidden danger targets.

7. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 6, characterized in that: The SAM model consists of the following three parts: The image encoder is responsible for processing the image and creating an embedding that represents the image, consisting of the VIT transformer; The prompt encoder handles additional input to the network; The mask decoder receives the output of the image encoder and the hint encoder and generates the final segmentation mask.

8. The method for measuring the distance of hidden dangers in a transmission line channel according to claim 1, characterized in that: In S4, the determination of the distance between the hidden danger target and the wire specifically includes: S41, hidden danger target detection, obtaining the location information of the hidden danger target in the image data and point cloud data; S42, importing the recognition segmentation positioning model according to the recognition position frame, and extracting the contour information of the hidden danger target; S43, obtaining the lowest point and the highest point of the hidden danger target, using the calibration fusion relationship, reverse calculation, and obtaining the corresponding three-dimensional position information; S44, calculating the clearance distance between the hidden danger target and the conductor; S45. Determine the hidden danger threat level of the hidden danger target based on the clearance distance and the voltage level of the transmission line.

Images