A tower orientation method, device, apparatus and storage medium

Abstract

The application discloses a tower orientation method, device, equipment and storage medium, and is applied to the technical field of tower orientation, and comprises the following steps: acquiring a tower head image of a target tower; inputting the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results; performing cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability; obtaining a straight line tower angle based on straight line tower data and the cross arm fitting confidence; inputting corner tower data into a turning angle relationship expression to obtain a corner tower angle; performing end key judgment processing on terminal tower data to obtain a terminal tower angle; performing weighted processing on the classification type probability, the straight line tower angle, the corner tower angle and the terminal tower angle to obtain an orientation absolute angle; and determining a next tower direction based on the orientation absolute angle. The method provided in the application can determine a next base tower direction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tower orientation, in particular to a tower orientation method, device, equipment and storage medium. BACKGROUND

[0002] With the advancement of smart grid construction, unmanned aerial vehicle (UAV) inspection has become an important means of distribution network line detection due to its efficiency and flexibility. In the conventional straight section tower inspection scenario, existing algorithms can achieve high-precision target detection and tracking, providing reliable support for power system operation and maintenance. However, when the UAV inspection scenario extends to complex areas containing line corners or branches, the limitations of traditional inspection techniques gradually become apparent.

[0003] Due to the constraints of the camera angle of the UAV gimbal, when the UAV approaches special structures such as corner towers and branch towers, the target tower is easily and quickly out of the detection area due to insufficient turning angle, resulting in target loss and forcing the system to interrupt the inspection process, ultimately leading to insufficient completeness of full autonomous inspection in complex line scenarios, greatly limiting the intelligent and large-scale application process of distribution network UAV inspection. SUMMARY

[0004] The present application provides a tower orientation method, device, equipment and storage medium to solve the technical problem of easy loss of the next base tower target in the current inspection process, to determine the direction of the next base tower.

[0005] To solve the above technical problems, the present application provides a tower orientation method, the method comprising:

[0006] obtaining a tower head image of a target tower;

[0007] inputting the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight tower data, corner tower data and terminal tower data;

[0008] performing cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability;

[0009] based on the straight tower data and the cross arm fitting confidence, obtaining a straight tower angle; inputting the corner tower data into a turning angle relationship to obtain a corner tower angle; performing end key judgment processing on the terminal tower data to obtain a terminal tower angle;

[0010] performing weighted processing on the classification type probability, the straight tower angle, the corner tower angle and the terminal tower angle to obtain an orientation absolute angle;

[0011] Determine a next tower direction of the target tower based on the directional absolute angle.

[0012] As one of the preferred solutions, the tower head image is input into the constructed target detection model to obtain insulator key point detection results and tower type classification results, which include:

[0013] The tower head image is preprocessed to obtain a processed tower head image, wherein the preprocessing process at least includes Gaussian denoising and perspective correction.

[0014] The processed tower head image is input into the target detection model constructed by the improved YOLOv8 algorithm for processing to obtain the insulator key point detection results and the tower type classification results.

[0015] As one of the preferred solutions, the insulator key point detection results are subjected to cross arm division processing to obtain cross arm fitting confidence and classification type probability, which include:

[0016] The insulator key point detection results are subjected to clustering processing to obtain a plurality of cross arm groups.

[0017] The plurality of cross arm groups are subjected to linear fitting processing using the least squares method to obtain cross arm group standard deviation.

[0018] Based on the cross arm group standard deviation, the plurality of cross arm groups are judged to obtain effective cross arm groups.

[0019] The effective cross arm groups are subjected to cross arm division processing to obtain cross arm fitting confidence and classification type probability.

[0020] As one of the preferred solutions, based on the straight line tower data and the cross arm fitting confidence, a straight line tower angle is obtained, which includes:

[0021] The straight line tower data and the cross arm fitting confidence are subjected to cross arm direction fitting and confidence weighting processing to obtain a weighted direction vector.

[0022] The weighted direction vector is processed using an angle conversion model to obtain the straight line tower angle.

[0023] As one of the preferred solutions, the terminal tower data is subjected to end key judgment processing to obtain a terminal tower angle, which includes:

[0024] The terminal tower data is subjected to image recognition judgment processing to obtain outgoing line end data and cross arm midpoint data.

[0025] The outgoing line end data and the cross arm midpoint data are input into a terminal tower angle relationship for processing to obtain the terminal tower angle.

[0026] As one of the preferred schemes, the classification type probability, the straight line tower angle, the corner tower angle and the terminal tower angle are weighted to obtain a directional absolute angle, which comprises:

[0027] The classification type probability and the cross arm fitting confidence are normalized to obtain the weight of the straight line tower angle, the weight of the corner tower angle and the weight of the terminal tower angle, respectively;

[0028] The straight line tower angle, the corner tower angle, the terminal tower angle and the corresponding weight are weighted to obtain the directional absolute angle.

[0029] As one of the preferred schemes, after obtaining the straight line tower angle, the tower orientation method further comprises:

[0030] Obtain historical straight line tower data;

[0031] The historical straight line tower data and the cross arm group standard deviation are input into a horizontal detection model constructed by a neural network algorithm to obtain a confidence influence factor and a mean offset;

[0032] The confidence influence factor, the mean offset and the cross arm fitting confidence are input into a correction coefficient relationship to obtain a correction coefficient;

[0033] Based on the correction coefficient, the straight line tower angle is corrected to obtain a corrected straight line tower angle.

[0034] Another embodiment of the application provides a tower orientation device, comprising:

[0035] An acquisition module is configured to acquire a tower head image of a target tower;

[0036] A detection module is configured to input the tower head image into a target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight line tower data, corner tower data and terminal tower data;

[0037] A division module is configured to divide the insulator key point detection results by cross arm to obtain a cross arm fitting confidence and a classification type probability;

[0038] A processing module is configured to obtain a straight line tower angle based on the straight line tower data and the cross arm fitting confidence, input the corner tower data into a turning angle relationship to obtain a corner tower angle, and perform end key judgment processing on the terminal tower data to obtain a terminal tower angle;

[0039] a weighting module configured to weight the classification type probability, the straight-line tower angle, the corner tower angle, and the terminal tower angle to obtain a directional absolute angle;

[0040] a determination module configured to determine a next tower direction of the target tower based on the directional absolute angle.

[0041] Another embodiment of the present application provides a tower orientation device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the tower orientation method as described above when executing the computer program.

[0042] Still another embodiment of the present application provides a computer readable storage medium storing a computer program, and the device where the computer readable storage medium is located implements the tower orientation method as described above when executing the computer program.

[0043] Compared with the prior art, the beneficial effects of the embodiments of the present application are at least one of the following:

[0044] The present application obtains the tower head image of the target tower; inputs the tower head image into the constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight-line tower data, corner tower data, and terminal tower data; performs cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability; obtains a straight-line tower angle based on the straight-line tower data and the cross arm fitting confidence; inputs the corner tower data into a turning angle relationship to obtain a corner tower angle; performs end key judgment processing on the terminal tower data to obtain a terminal tower angle; weights the classification type probability, the straight-line tower angle, the corner tower angle, and the terminal tower angle to obtain a directional absolute angle; and determines a next tower direction of the target tower based on the directional absolute angle.

[0045] Compared with the prior art, the present application constructs a special processing method for different types of towers, fully considers the structural differences of straight-line towers, corner towers, and terminal towers, makes the directional angle calculation more targeted, effectively improves the directional accuracy under different tower types, combines the tower type prediction probability and the cross arm fitting confidence through a multi-type weighting fusion strategy, enhances the adaptability in complex scenes, avoids the single model misjudgment problem, improves the stability of the unmanned aerial vehicle in the face of different types of towers, and realizes the unmanned aerial vehicle full autonomous adaptive distribution network tower inspection task by combining the target detection model and the multi-type tower weighting processing to obtain the next base tower direction in the unmanned aerial vehicle adaptive inspection. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 is a flowchart of a tower orientation method in one of the embodiments of the present application;

[0047] Figure 2 is a structural schematic diagram of a tower orientation device in one of the embodiments of the present application;

[0048] Figure 3 is a structural schematic diagram of a tower orientation device in one of the embodiments of the present application;

[0049] Reference Signs:

[0050] Wherein, 11, an acquisition module; 12, a detection module; 13, a division module; 14, a processing module; 15, a weighting module; 16, a determination module; 21, a processor; 22, a memory. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0052] In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0053] The terms "vertical", "horizontal", "left", "right", "up", "down", and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. The term "and / or" used herein includes any and all combinations of one or more related listed items. The specific meanings of the above terms in the present application can be understood by those of ordinary skill in the art on a case-by-case basis.

[0054] In the description of the present application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present application have the same meaning as understood by those skilled in the art. The terms used in the specification of the present application are only for the purpose of describing the specific embodiments, and are not intended to limit the present application. The specific meanings of the above terms in the present application can be understood by those of ordinary skill in the art on a case-by-case basis.

[0055] An embodiment of the present application provides a tower orientation method, and specifically, refer to Figure 1 , Figure 1 The tower orientation method in an embodiment of the present application is shown as a flowchart, and the method comprises the following steps:

[0056] S1: obtaining a tower head image of a target tower;

[0057] S2: inputting the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least comprise straight tower data, corner tower data and terminal tower data;

[0058] S3: performing cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability;

[0059] S4: obtaining a straight tower angle based on the straight tower data and the cross arm fitting confidence; inputting the corner tower data into a turning angle relationship to obtain a corner tower angle; performing end key judgment processing on the terminal tower data to obtain a terminal tower angle;

[0060] S5: performing weighted processing on the classification type probability, the straight tower angle, the corner tower angle and the terminal tower angle to obtain an orientation absolute angle;

[0061] S6: determining a next tower direction of the target tower based on the orientation absolute angle.

[0062] Specifically, a tower head image of a target tower is obtained, and the tower head (tower top) image usually contains key components such as insulator strings, cross arms (horizontal support structures) and the like, which are core visual features for judging tower type and direction.

[0063] The target tower top image is obtained by a UAV, and specifically, before the UAV starts to inspect a current tower, the initial absolute angle of flying into the tower is recorded ; after the UAV ends the current tower inspection, the UAV hovers at a position 2m-5m above the tower, the gimbal angle is adjusted to -90°, and the tower head image, i.e. the target tower top image, is obtained by taking a picture from below.

[0064] In step S2, the tower head image is input into the built target detection model to obtain insulator key point detection results and tower type classification results. Specifically, the tower head image is input into the built target detection model to obtain insulator key point detection results and classification type probability, including: performing image preprocessing on the tower head image to obtain a processed tower head image, wherein the preprocessing process at least includes Gaussian denoising and perspective correction; inputting the processed tower head image into the target detection model constructed by the improved YOLOv8 algorithm for processing to obtain the insulator key point detection results and the tower type classification results.

[0065] Preferably, the insulator key point detection results include the center point coordinates of the target tower, and the tower type classification results at least include straight line tower data, corner tower data and terminal tower data.

[0066] Specifically, YOLOv8 is a single-stage target detection algorithm, and its original structure includes feature extraction, multi-scale feature fusion and target prediction. For the insulator detection requirement, the improved model mainly optimizes the following directions:

[0067] On the basis of the original target detection (locating the insulator region), a key point detection branch (predicting the key coordinates of the insulator) and a classification branch (predicting the insulator type probability) are added; the anchor box size is adjusted (matching the size distribution of the insulator), the loss function is optimized (fusing the classification loss, the key point positioning loss and the cross arm fitting loss), and the detection capability in the complex background is enhanced.

[0068] The insulator key point detection results are subjected to cross arm division processing to obtain cross arm fitting confidence and classification type probability, and the specific process includes: clustering the insulator key point detection results to obtain a plurality of cross arm groups; using the least square method to perform linear fitting processing on the plurality of cross arm groups to obtain cross arm group standard deviation; judging the plurality of cross arm groups based on the cross arm group standard deviation to obtain effective cross arm groups; and performing cross arm division processing on the effective cross arm groups to obtain cross arm fitting confidence and classification type probability.

[0069] Specifically, the process of performing cross arm division processing on the insulator key point detection results to obtain the tower type classification results is a logic chain of gradually extracting the cross arm structural features and mapping to the tower type through the spatial distribution analysis of the key points.

[0070] First, the key point detection results of the insulator are clustered to obtain a plurality of cross arm groups. The two ends of the insulator are usually fixed on the cross arm, and the key points of the insulators on the same cross arm have strong correlation in spatial coordinates. Based on the mounting relationship between the insulator and the cross arm, a clustering algorithm such as DBSCAN or K-means is used to group the key points.

[0071] The cross arm itself is a rigid straight structure, and the key points of the insulators on the same cross arm should theoretically be distributed on the same straight line. Therefore, the key points of each cross arm group are fitted with a straight line, and the least squares method is used to calculate the fitting straight line to minimize the sum of the squared distances of all key points in the group to the straight line, that is, the optimal fitting. The standard deviation of the distance of all key points in the group to the fitted straight line is calculated, which reflects the dispersion degree of the key points deviating from the fitted straight line.

[0072] A standard deviation threshold is set. If the standard deviation of a cross arm group is less than the threshold, it is determined to be an "effective cross arm group", and if the standard deviation is greater than the threshold, it is determined to be an "invalid cross arm group" and is rejected.

[0073] Based on the number, spatial distribution relationship, such as the height difference in the vertical direction and the position in the horizontal direction, of the effective cross arm groups, the cross arm fitting confidence and the classification type probability are determined.

[0074] The classification type probability is a probability vector of straight line tower data, corner tower data and terminal tower data.

[0075] Based on the straight line tower data and the cross arm fitting confidence, a straight line tower angle is obtained, specifically including: performing cross arm direction fitting and confidence weighting processing on the straight line tower data and the cross arm fitting confidence to obtain a weighted direction vector; and processing the weighted direction vector using an angle conversion model to obtain the straight line tower angle .

[0076] Specifically, two endpoints X and Y are taken in the order of x coordinates, the vector L between the two endpoints is calculated, the weighted direction vector is processed using an angle conversion model to obtain the straight line tower angle , wherein the angle conversion model converts the vector into an absolute angle, and preferably uses an arctan function to avoid direction ambiguity.

[0077] After the straight-line tower angle is obtained, the straight-line tower angle is corrected. Specifically, historical straight-line tower data is obtained; the historical straight-line tower data and the cross arm group standard deviation are input into a horizontal detection model constructed by a neural network algorithm to obtain a confidence influence factor and a mean offset; the confidence influence factor, the mean offset, and the cross arm fitting confidence are input into a correction coefficient relationship to obtain a correction coefficient; and the straight-line tower angle is corrected based on the correction coefficient to obtain a corrected straight-line tower angle.

[0078] Specifically, historical straight-line tower image data that has been labeled in historical inspection is collected, and the standard deviation of a cross arm group in each historical data is calculated. A horizontal detection model is constructed based on a neural network (such as LSTM or CNN), and the historical straight-line tower data and the cross arm group standard deviation are input into the horizontal detection model to obtain a confidence influence factor and a mean offset. The confidence influence factor reflects the matching degree of the current cross arm group and the historical effective cross arm mode, and the mean offset represents the offset of the current cross arm group standard deviation relative to the mean of the historical effective cross arm standard deviation.

[0079] The confidence influence factor, the mean offset, and the cross arm fitting confidence are input into a correction coefficient relationship to obtain a correction coefficient. The correction coefficient relationship is as follows:

[0080]

[0081] wherein, is the mean offset, is the confidence influence factor, is the cross arm fitting confidence, is the correction coefficient, and the straight-line tower angle is corrected based on the correction coefficient to obtain a corrected straight-line tower angle.

[0082] The corner tower data is input into a turning angle relationship to obtain a corner tower angle. Specifically, the corner tower has two pairs of cross arms, and the corner tower data (two pairs of cross arms) is , , , The cross arm vector is , .

[0083] The angle between the two cross arms is .

[0084]

[0085] The direction of the corner tower is determined as dir=sign( ). If clockwise rotation (right turn), dir is a rotation symbol, left turn takes , right turn takes . .

[0086] The corner tower angle is:

[0087]

[0088] The terminal tower data is subjected to key judgment processing to obtain a terminal tower angle, specifically including: the terminal tower data is subjected to image recognition judgment processing to obtain outgoing terminal data and cross arm midpoint data; the outgoing terminal data and the cross arm midpoint data are input into a terminal tower angle relationship formula for processing to obtain the terminal tower angle .

[0089] The outgoing terminal is the farthest point from the midpoint, and the coordinates are , and the coordinates of the cross arm ends are: , The cross arm midpoint is:

[0090]

[0091]

[0092]

[0093] The terminal tower angle is , is the outgoing terminal vector, is the horizontal axis vector, , is the included angle between the outgoing terminal direction and the horizontal axis direction.

[0094] The classification type probability, the straight line tower angle, the corner tower angle, and the terminal tower angle are subjected to weighted processing to obtain a directional absolute angle, including:

[0095] The classification type probability and the cross arm fitting confidence are subjected to probability normalization processing to respectively obtain a weight of the straight line tower angle, a weight of the corner tower angle, and a weight of the terminal tower angle; the straight line tower angle, the corner tower angle, the terminal tower angle, and the corresponding weights are subjected to weighted processing to obtain the directional absolute angle.

[0096] The next tower direction of the target tower is determined based on the directional absolute angle.

[0097] Specifically, the directional absolute angle is sent to the flight control system, the attitude of the unmanned aerial vehicle is adjusted, and it is ensured that the next tower is in the frame of the unmanned aerial vehicle holder.

[0098] Another embodiment of the present application provides a tower orientation device, specifically, please refer to Figure 2 , Figure 2 The structural diagram of the tower orientation device in one of the embodiments of the present application is shown, and the system comprises:

[0099] The acquisition module 11 is configured to acquire a tower head image of a target tower.

[0100] The detection module 12 is configured to input the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight tower data, corner tower data and terminal tower data.

[0101] The division module 13 is configured to perform cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability.

[0102] The processing module 14 is configured to obtain a straight tower angle based on the straight tower data and the cross arm fitting confidence, input the corner tower data into a turning angle relationship to obtain a corner tower angle, and perform end key judgment processing on the terminal tower data to obtain a terminal tower angle.

[0103] The weighting module 15 is configured to perform weighting processing on the classification type probability, the straight tower angle, the corner tower angle and the terminal tower angle to obtain a directional absolute angle.

[0104] The determination module 16 is configured to determine a next tower direction of the target tower based on the directional absolute angle.

[0105] Referring to Figure 3 , which is a structural diagram of the tower orientation device provided by the embodiment of the present application, the tower orientation device 20 provided by the embodiment of the present application comprises a processor 21, a memory 22, and a computer program stored in the memory 22 and configured to be executed by the processor 21, wherein the processor 21 implements the steps in the above tower orientation method embodiments when executing the computer program, such as steps S1-S6 in Figure 1 ; or, the processor 21 implements the functions of each module in the above device embodiments when executing the computer program, such as the acquisition module 11.

[0106] For example, the computer program can be divided into one or more modules stored in the memory 22 and executed by the processor 21 to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the tower orientation device 20. For example, the computer program can be divided into an acquisition module 11, a detection module 12, a division module 13, and the like, and the specific functions of each module are as follows:

[0107] The acquisition module 11 is configured to acquire a tower head image of a target tower.

[0108] The detection module 12 is configured to input the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight line tower data, corner tower data, and terminal tower data.

[0109] The division module 13 is configured to perform cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability.

[0110] The processing module 14 is configured to obtain a straight line tower angle based on the straight line tower data and the cross arm fitting confidence, input the corner tower data into a turning angle relationship to obtain a corner tower angle, and perform end key judgment processing on the terminal tower data to obtain a terminal tower angle.

[0111] The weighting module 15 is configured to perform weighting processing on the classification type probability, the straight line tower angle, the corner tower angle, and the terminal tower angle to obtain an orientation absolute angle.

[0112] The determination module 16 is configured to determine a next tower direction of the target tower based on the orientation absolute angle.

[0113] The tower orientation device 20 can include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art can understand that the schematic diagram is only an example of the tower orientation device and does not constitute a limitation on the tower orientation device 20, which can include more or fewer components than the diagram, or combine certain components, or different components, for example, the tower orientation device 20 can also include an input / output device, a network access device, a bus, and the like.

[0114] The processor 21 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The processor 21 is a control center of the tower orientation device 20, and is connected with various parts of the tower orientation device 20 through various interfaces and lines.

[0115] The memory 22 can be used to store computer programs and / or modules. The processor 21 realizes various functions of the tower orientation device 20 by running or executing the computer programs and / or modules stored in the memory 22, and calling data stored in the memory 22. The memory 22 can mainly include a program storage area and a data storage area. The program storage area can store an operating system, at least one application program required for a function (such as a sound playing function, an image playing function, etc.), etc. The data storage area can store data created according to use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 22 can include a high-speed random access memory, and can also include a nonvolatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0116] The modules integrated in the tower orientation device 20 can be stored in a computer readable storage medium if they are implemented in the form of software function units and sold or used as independent products. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware. The computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0117] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The program can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned various method embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0118] Correspondingly, the embodiment of the present application provides a computer readable storage medium including a stored computer program. When the computer program runs, it controls the device where the computer readable storage medium is located to perform the steps in the tower orientation method of the above-mentioned embodiment, such as the steps S1-S6 in the method described in the above-mentioned embodiment. Figure 1

[0119] Compared with the prior art, the embodiment of the present application has at least one of the following advantages:

[0120] ​The application obtains a tower head image of a target tower; inputs the tower head image into a constructed target detection model to obtain insulator key point detection results and tower type classification results, wherein the tower type classification results at least include straight line tower data, corner tower data and terminal tower data; performs cross arm division processing on the insulator key point detection results to obtain cross arm fitting confidence and classification type probability; obtains a straight line tower angle based on the straight line tower data and the cross arm fitting confidence; inputs the corner tower data into a turning angle relationship to obtain a corner tower angle; performs end key judgment processing on the terminal tower data to obtain a terminal tower angle; performs weighted processing on the classification type probability, the straight line tower angle, the corner tower angle and the terminal tower angle to obtain a directional absolute angle; and determines a next tower direction of the target tower based on the directional absolute angle.

[0121] Compared with the prior art, the application constructs a special processing method for different types of towers, fully considers the structural differences of straight line towers, corner towers and terminal towers, makes the directional angle calculation more targeted, effectively improves the directional accuracy under different tower types, combines the tower type prediction probability and the cross arm fitting confidence through a multi-type weighted fusion strategy, enhances the adaptability in complex scenes, avoids the single model misjudgment problem, improves the stability of the directional unmanned aerial vehicle when facing different types of towers, and realizes the unmanned aerial vehicle full autonomous adaptive distribution network tower inspection task by combining the target detection model and the multi-type tower weighted processing to obtain the next tower direction in the unmanned aerial vehicle adaptive inspection.

[0122] The above-described embodiments only express several embodiments of the application, and the description is more specific and detailed, but it cannot be understood as a limitation on the scope of the patent of the application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the application, a number of modifications and improvements can be made, which belong to the protection scope of the application. Therefore, the protection scope of the patent of the application should be subject to the appended claims.

Claims

1. A method for orienting towers, characterized in that, include: Obtain the tower head image of the target tower; The tower head image is input into the constructed target detection model to obtain the insulator key point detection results and tower type classification results. The tower type classification results include at least straight tower data, corner tower data and terminal tower data. The detection results of the key points of the insulator are processed by crossarm segmentation to obtain the crossarm fitting confidence and classification type probability; Based on the straight tower data and the crossarm fitting confidence level, the straight tower angle is obtained; the angle tower angle is obtained by inputting the angle tower data into the turning angle relationship; the terminal tower angle is obtained by performing end-key judgment processing on the terminal tower data. The classification type probability, the straight tower angle, the corner tower angle, and the terminal tower angle are weighted to obtain the absolute orientation angle, including: The probability of the classification type and the confidence of the crossarm fitting are normalized to obtain the weights of the straight tower angle, the angle tower angle, and the terminal tower angle, respectively. The directional absolute angle is obtained by weighting the straight tower angle, the corner tower angle, the terminal tower angle, and the corresponding weights. The direction of the next tower of the target tower is determined based on the absolute directional angle.

2. The tower orientation method as described in claim 1, characterized in that, The step of inputting the tower head image into the constructed target detection model to obtain insulator key point detection results and tower type classification results includes: The tower head image is preprocessed to obtain a processed tower head image, wherein the preprocessing process includes at least Gaussian denoising and perspective correction; The processed tower head image is input into the target detection model constructed by the improved YOLOv8 algorithm for processing, to obtain the insulator key point detection results and the tower type classification results.

3. The tower orientation method as described in claim 1, characterized in that, The process of performing crossarm segmentation on the detection results of the key points of the insulator to obtain the crossarm fitting confidence and classification type probability includes: Clustering the detection results of the key points of the insulator yields several crossarm groups; The standard deviation of the crossarm group is obtained by performing linear fitting on several crossarm groups using the least squares method. Based on the standard deviation of the crossarm group, several crossarm groups are judged to obtain the effective crossarm group; The effective crossarm groups are subjected to crossarm segmentation processing to obtain the crossarm fitting confidence and classification type probability.

4. The tower orientation method as described in claim 1, characterized in that, The process of obtaining the straight tower angle based on the straight tower data and the crossarm fitting confidence level includes: The straight tower data and the crossarm fitting confidence score are subjected to crossarm direction fitting and confidence weighting to obtain a weighted direction vector. The angle of the straight tower is obtained by processing the weighted direction vector using an angle transformation model.

5. The tower orientation method as described in claim 1, characterized in that, The step of performing end-key judgment processing on the terminal tower data to obtain the terminal tower angle includes: The terminal tower data is processed by image recognition to obtain the outgoing end data and the crossarm midpoint data; The data from the outgoing end and the data from the midpoint of the crossarm are input into the formula for the angle of the terminal tower for processing, and the angle of the terminal tower is obtained.

6. The tower orientation method as described in claim 3, characterized in that, After obtaining the angle of the straight tower, the tower orientation method further includes: Obtain historical straight-line tower data; The historical straight tower data and the standard deviation of the crossarm group are input into the horizontal detection model constructed by the neural network algorithm to obtain the confidence influence factor and mean offset. The confidence level influence factor, the mean offset, and the crossarm fitting confidence level are input into the correction coefficient formula to obtain the correction coefficient; Based on the correction coefficient, the angle of the straight tower is corrected to obtain the corrected angle of the straight tower.

7. A pole orientation device, characterized in that, include: The acquisition module is used to acquire images of the tower head of the target tower. The detection module is used to input the tower head image into the constructed target detection model to obtain the insulator key point detection results and tower type classification results, wherein the tower type classification results include at least straight tower data, corner tower data and terminal tower data; The segmentation module is used to perform crossarm segmentation processing on the detection results of key points of the insulator, and obtain the crossarm fitting confidence and classification type probability. The processing module is used to obtain the straight tower angle based on the straight tower data and the crossarm fitting confidence; input the angle tower data into the turning angle relationship to obtain the angle tower angle; and perform end-key judgment processing on the terminal tower data to obtain the terminal tower angle. The weighting module is used to weight the classification type probability, the straight tower angle, the corner tower angle, and the terminal tower angle to obtain the absolute orientation angle, including: The probability of the classification type and the confidence of the crossarm fitting are normalized to obtain the weights of the straight tower angle, the angle tower angle, and the terminal tower angle, respectively. The directional absolute angle is obtained by weighting the straight tower angle, the corner tower angle, the terminal tower angle, and the corresponding weights. The determination module is used to determine the direction of the next tower of the target tower based on the absolute orientation angle.

8. A pole orientation device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the tower orientation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the tower orientation method as described in any one of claims 1 to 6.

Images