Power transmission line channel hidden danger target positioning method and system based on image segmentation

Abstract

The application belongs to the technical field of image processing, and provides a power transmission line channel hidden danger target positioning method and system based on image segmentation. The method comprises the following steps: acquiring two-dimensional image data and three-dimensional point cloud data of a power transmission line channel scene; extracting a multi-scale global feature map of the two-dimensional image data of the power transmission line channel scene, fusing all different scale global feature maps based on a coordinate attention mechanism, and obtaining all hidden danger target segmentation regions in the two-dimensional image data of the power transmission line channel scene; screening out three-dimensional point cloud data projected on each hidden danger target segmentation region based on a calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the power transmission line channel scene; classifying the three-dimensional point cloud data of each hidden danger target segmentation region, obtaining point cloud data corresponding to each hidden danger target, and finally realizing positioning of each hidden danger target. The application can realize hidden danger targets of different scales in a complex background, and significantly improve hidden danger target segmentation performance.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of image processing, and particularly relates to a transmission line passage hidden danger target positioning method and system based on image segmentation. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] In such a complex scene as engineering operation near the transmission line, the engineering vehicle may touch the transmission line due to improper operation, which may cause damage to the transmission line and personnel casualties, and seriously affect the normal operation of the power grid.

[0004] The prior art provides a method of using two-dimensional image data of a scene, based on a segmentation neural network to perform target segmentation processing on the scene image, but for complex scenes such as construction near the transmission line, the hidden danger target size varies greatly, the background environment is complex, and other objects are shielded. The existing image segmentation method is not suitable for complex scenes of the transmission line, resulting in poor segmentation accuracy and difficulty in accurately positioning the hidden danger target. SUMMARY

[0005] To solve the above technical problems, the present application provides a transmission line passage hidden danger target positioning method and system based on image segmentation, which can realize segmentation of hidden danger targets with different scales in complex backgrounds and significantly improve the performance of hidden danger target segmentation.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] The first aspect of the present application provides a transmission line passage hidden danger target positioning method based on image segmentation.

[0008] In one or more embodiments, a transmission line passage hidden danger target positioning method based on image segmentation is provided, comprising:

[0009] Obtaining two-dimensional image data and three-dimensional point cloud data of the transmission line passage scene;

[0010] Extracting a multi-scale global feature map of the two-dimensional image data of the transmission line passage scene, fusing all different scale global feature maps based on a coordinate attention mechanism, and obtaining all hidden danger target segmentation regions in the two-dimensional image data of the transmission line passage scene based on the mapping relationship between the fused feature map and the hidden danger target segmentation region;

[0011] Based on the calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the transmission line passage scene, the three-dimensional point cloud data projected on each hidden danger target segmentation region is screened out;

[0012] The three-dimensional point cloud data of each hidden danger target segmentation area is classified to obtain the point cloud data corresponding to each hidden danger target, and finally the positioning of each hidden danger target is realized.

[0013] As an implementation form, the process of fusing all different scale global feature maps based on the coordinate attention mechanism comprises:

[0014] Each different scale global feature map is correspondingly assigned a channel;

[0015] For each channel, the corresponding scale global feature map is encoded along the horizontal coordinate and the vertical coordinate respectively to obtain the corresponding horizontal coordinate and vertical coordinate encoded features;

[0016] For each channel, the corresponding horizontal coordinate and vertical coordinate encoded features are spliced along the spatial dimension, the spliced features are normalized and activated by a nonlinear activation function, and then the fusion feature map of each channel is obtained through activation function and normalization weighting.

[0017] As an implementation form, the expression of the fusion feature map of each channel is:

[0018]

[0019] Wherein, δ is the regularization and nonlinear activation operation, σ represents the activation function operation, Conv is the convolution operation, Y v (i,j) is the fusion feature map of the vth channel; (i,j) is the horizontal coordinate point and the vertical coordinate point respectively; is the horizontal coordinate encoded feature, is the vertical coordinate encoded feature; m is the maximum coordinate value of the fusion feature map along the horizontal direction; n is the maximum coordinate value of the fusion feature map along the vertical direction; x and y are the horizontal coordinate and the vertical coordinate respectively.

[0020] As an implementation form, for each channel, the process of encoding the corresponding scale global feature map along the horizontal coordinate is:

[0021] One-dimensional mean pooling is performed on the corresponding scale global feature map along the horizontal coordinate direction, starting from the leftmost side of the corresponding scale global feature map, sliding along the width direction with a set window size, and calculating the average of the pixel values covered in the window to obtain the horizontal coordinate encoded feature.

[0022] As an implementation form, for each channel, the process of encoding the corresponding scale global feature map along the vertical coordinate is:

[0023] A one-dimensional mean pooling is performed on the corresponding scale global feature map along the vertical coordinate direction, starting from the lowermost side of the corresponding scale global feature map, sliding in sequence along the height direction with a set window size, and performing average calculation on the pixel values covered in the window to obtain a vertical coordinate encoding feature.

[0024] As an implementation, a two-dimensional image data of a transmission line passage scene is processed by using an image segmentation model to obtain all hidden danger target segmentation regions therein.

[0025] The image segmentation model comprises a feature extraction network, a cross-level feature fusion network and a small target segmentation head; the feature extraction network is used to extract multi-scale global feature maps of the two-dimensional image data of the transmission line passage scene; the cross-level feature fusion network is used to fuse all different scale global feature maps based on a coordinate attention mechanism to obtain a fusion feature map; and the small target segmentation head is used to obtain all hidden danger target segmentation regions in the two-dimensional image data of the transmission line passage scene based on a mapping relationship between the fusion feature map and the hidden danger target segmentation region.

[0026] The second aspect of the application provides a transmission line passage hidden danger target positioning system based on image segmentation.

[0027] In one or more embodiments, a transmission line passage hidden danger target positioning system based on image segmentation comprises:

[0028] A data acquisition module is configured to acquire two-dimensional image data and three-dimensional point cloud data of a transmission line passage scene;

[0029] An image segmentation module is configured to extract multi-scale global feature maps of the two-dimensional image data of the transmission line passage scene, fuse all different scale global feature maps based on a coordinate attention mechanism, and obtain all hidden danger target segmentation regions in the two-dimensional image data of the transmission line passage scene based on a mapping relationship between the fusion feature map and the hidden danger target segmentation region.

[0030] A point cloud projection module is configured to filter out three-dimensional point cloud data projected on each hidden danger target segmentation region based on a calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the transmission line passage scene.

[0031] A target positioning module is configured to classify the three-dimensional point cloud data of each hidden danger target segmentation region to obtain point cloud data corresponding to each hidden danger target, and finally realize positioning of each hidden danger target.

[0032] The third aspect of the application provides a computer readable storage medium.

[0033] A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the image segmentation based transmission line channel hidden danger target positioning method as described above.

[0034] A fourth aspect of the present application provides a computer program product.

[0035] A computer program product comprising computer programs / instructions which, when executed by a processor, implement the steps of the image segmentation based transmission line channel hidden danger target positioning method as described above.

[0036] A fifth aspect of the present application provides an electronic device.

[0037] An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the image segmentation based transmission line channel hidden danger target positioning method as described above.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] (1) The present application determines all hidden danger target segmentation regions by segmenting the two-dimensional image data of the transmission line channel scene, and then screens out the three-dimensional point cloud data projected on each hidden danger target segmentation region according to the calibration relationship between the two-dimensional image data and the three-dimensional point cloud data, and further classifies the three-dimensional point cloud data of each hidden danger target segmentation region, thereby realizing accurate positioning of hidden danger targets in the transmission line channel with different scales under complex background, and significantly improving the hidden danger target segmentation performance.

[0040] (2) The present application fuses features based on the coordinate attention mechanism for the multi-scale global feature map of the two-dimensional image data of the transmission line channel scene, introduces grid coordinate information, extends the attention mechanism from the pixel level to the image coordinate level, can capture the global spatial structure of the image, pays more attention to the position information of the image, and enhances the understanding and representation ability of the semantic information of the image; the coordinate attention mechanism models the correlation between features in the channel dimension, and captures the relationship between different positions in the spatial dimension, thereby improving the understanding and modeling ability of image features; the coordinate attention mechanism has good scalability when processing large-size images, and can effectively handle complex scenes and image contexts. BRIEF DESCRIPTION OF DRAWINGS

[0041] The drawings accompanying the specification of the present application form a part thereof, serve to provide further understanding of the present application, and together with the description of the exemplary embodiments of the present application and the explanation thereof, explain the present application, and do not constitute an improper limitation of the present application.

[0042] Figure 1is a flowchart of a power transmission line channel hidden danger target positioning method based on image segmentation of an embodiment of the present application.

[0043] Figure 2 is a structural diagram of a power transmission line channel hidden danger target positioning system based on image segmentation of an embodiment of the present application.

[0044] Figure 3 is a schematic diagram of an electronic device of an embodiment of the present application. DETAILED DESCRIPTION

[0045] The present application will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0047] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, there is a presence of a feature, step, operation, device, component, and / or combinations thereof.

[0048] Figure 1 is a flowchart of a power transmission line channel hidden danger target positioning method based on image segmentation of an embodiment of the present application, as Figure 1 The power transmission line channel hidden danger target positioning method based on image segmentation in the present embodiment can include:

[0049] S101, acquiring two-dimensional image data and three-dimensional point cloud data of a power transmission line channel scene;

[0050] S102, extracting a multi-scale global feature map of the two-dimensional image data of the power transmission line channel scene, fusing all different scale global feature maps based on a coordinate attention mechanism, and obtaining all hidden danger target segmentation regions in the two-dimensional image data of the power transmission line channel scene based on a mapping relationship between the fused feature map and the hidden danger target segmentation region;

[0051] S103, based on the calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the power transmission line channel scene, screening out three-dimensional point cloud data projected on each hidden danger target segmentation region;

[0052] S104, classify the three-dimensional point cloud data of each hidden danger target segmentation region to obtain the point cloud data corresponding to each hidden danger target, and finally realize the positioning of each hidden danger target.

[0053] The embodiment determines all hidden danger target segmentation regions by segmenting the two-dimensional image data of the power line channel scene, and then screens the three-dimensional point cloud data projected on each hidden danger target segmentation region according to the calibration relationship between the two-dimensional image data and the three-dimensional point cloud data, and classifies the three-dimensional point cloud data of each hidden danger target segmentation region, thereby realizing accurate positioning of hidden danger targets in the power line channel with different scales under complex background, and significantly improving the hidden danger target segmentation performance.

[0054] In step S101, the two-dimensional image data of the power line channel scene can be obtained by an image acquisition device, such as a camera mounted on a drone.

[0055] The three-dimensional point cloud data of the power line channel scene can be obtained by a laser radar device.

[0056] In step S102, the multi-scale global feature map of the extracted two-dimensional image data of the power line channel scene includes but is not limited to shape features, color features, and spatial position features.

[0057] In step S102, the process of fusing all different scale global feature maps based on the coordinate attention mechanism includes:

[0058] Step S1021: Assign a channel to each different scale global feature map;

[0059] Step S1022: For each channel, encode the corresponding scale global feature map along the horizontal and vertical coordinates to obtain the corresponding horizontal and vertical coordinate encoding features.

[0060] In step S1022, for each channel, the process of encoding the corresponding scale global feature map along the horizontal coordinate is as follows:

[0061] Perform one-dimensional mean pooling on the corresponding scale global feature map along the horizontal coordinate direction. Starting from the leftmost side of the corresponding scale global feature map, slide along the width direction with a set window size (for example, the window size is 3), and calculate the average of the pixel values covered in the window to obtain the horizontal coordinate encoding feature.

[0062] The embodiment integrates the pixel information scattered in the horizontal direction through the mean pooling operation. After the average calculation of the pixels at different positions, the pixels are connected with each other, so that the network can capture the long-range dependency relationship in the horizontal direction.

[0063] In the embodiment, for each channel, the process of encoding the corresponding scale global feature map along the vertical coordinate is specifically as follows:

[0064] The one-dimensional mean pooling is performed on the corresponding scale global feature map along the vertical coordinate direction, starting from the lowermost side of the corresponding scale global feature map, sliding along the height direction in a set window size, and calculating the average of the pixel values covered in the window to obtain the vertical coordinate encoding feature.

[0065] In the embodiment, the pixel information dispersed in the vertical direction is integrated through the mean pooling operation. After the average calculation, the pixels at different positions are associated with each other, so that the network can capture the long-range dependence in the vertical direction.

[0066] Step S1023: For each channel, the corresponding horizontal coordinate and vertical coordinate encoding features are spliced along the spatial dimension, the spliced features are regularized and activated through normalization and a nonlinear activation function, and then the fusion feature map of each channel is obtained through the activation function and normalization weighting.

[0067] Specifically, the expression of the fusion feature map of each channel in step S1023 is as follows:

[0068]

[0069] wherein δ is the regularization and nonlinear activation operation, σ represents the activation function operation, Conv is the convolution operation, Y v (i,j) is the fusion feature map of the vth channel; (i,j) are the horizontal coordinate point and the vertical coordinate point, respectively; is the horizontal coordinate encoding feature, is the vertical coordinate encoding feature; m is the maximum coordinate value of the fusion feature map along the horizontal direction; n is the maximum coordinate value of the fusion feature map along the vertical direction; x and y are the horizontal coordinate and the vertical coordinate, respectively.

[0070] In the embodiment, the multi-scale global feature map of the two-dimensional image data of the power line channel scene is fused based on the coordinate attention mechanism, the grid coordinate information is introduced, the attention mechanism is extended from the pixel level to the image coordinate level, the global spatial structure of the image can be captured, the position information of the image is paid more attention to, and the understanding and representation ability of the semantic information of the image are enhanced; the coordinate attention mechanism models the correlation between features in the channel dimension, and captures the relationship between different positions in the spatial dimension, thereby improving the understanding and modeling ability of the image features; when processing large-size images, the coordinate attention mechanism has good scalability and can effectively process complex scenes and image contexts.

[0071] In step S102, the two-dimensional image data of the transmission line channel scene is processed by using an image segmentation model to obtain all hidden danger target segmentation regions therein;

[0072] The image segmentation model comprises a feature extraction network, a cross-level feature fusion network and a small target segmentation head; the feature extraction network is used to extract multi-scale global feature maps of the two-dimensional image data of the transmission line channel scene; the cross-level feature fusion network is used to fuse all different scale global feature maps based on a coordinate attention mechanism to obtain a fusion feature map; and the small target segmentation head is used to obtain all hidden danger target segmentation regions in the two-dimensional image data of the transmission line channel scene based on a mapping relationship between the fusion feature map and the hidden danger target segmentation region.

[0073] For example, the image segmentation model can be implemented by using a YOLOv series algorithm, so that the C2f_Faster module is used in the feature extraction network and the cross-level feature fusion network to improve the performance of the model while focusing on the simplicity and interpretability of the model, so that the model is more suitable for hidden danger target image segmentation and positioning tasks in complex scenes.

[0074] It can be understood here that in other embodiments, the image segmentation model can also be implemented by using other segmentation algorithms, which will not be described here in detail.

[0075] In step S103, the two-dimensional image data and the three-dimensional point cloud data of the transmission line channel scene can be pre-calibrated according to the coordinate relationship of the corresponding acquisition device.

[0076] In step S104, the classification algorithm for the three-dimensional point cloud data of each hidden danger target segmentation region can use the SVM algorithm and the like. It should be noted here that the classification algorithm is used for point cloud fine segmentation, and the type of the classification algorithm does not affect the essence of the present scheme.

[0077] The types of the three-dimensional point cloud data here include but are not limited to transmission line point cloud data, ground point cloud data, vegetation point cloud data, building point cloud data and hidden danger target point cloud data and the like.

[0078] Figure 2 is a structure diagram of a hidden danger target positioning system for a transmission line channel based on image segmentation in an embodiment of the present application, and corresponds to the hidden danger target positioning method for a transmission line channel based on image segmentation in Figure 1 , as shown in Figure 2 , the hidden danger target positioning system for a transmission line channel based on image segmentation in the present embodiment can comprise:

[0079] A data acquisition module 201 is configured to acquire two-dimensional image data and three-dimensional point cloud data of a transmission line channel scene.

[0080] an image segmentation module 202, configured to extract multi-scale global feature maps of the two-dimensional image data of the transmission line channel scene, fuse all global feature maps of different scales based on a coordinate attention mechanism, and obtain all hidden danger target segmentation regions in the two-dimensional image data of the transmission line channel scene based on a mapping relationship between the fused feature maps and the hidden danger target segmentation regions;

[0081] a point cloud projection module 203, configured to filter out three-dimensional point cloud data projected on each hidden danger target segmentation region based on a calibration relationship between the two-dimensional image data of the transmission line channel scene and the three-dimensional point cloud data;

[0082] a target positioning module 204, configured to classify the three-dimensional point cloud data of each hidden danger target segmentation region to obtain point cloud data corresponding to each hidden danger target, and finally realize positioning of each hidden danger target.

[0083] It should be noted that, Figure 2 the various modules in the image segmentation-based transmission line channel hidden danger target positioning system in the first aspect of the present application and the various steps in the image segmentation-based transmission line channel hidden danger target positioning method in the second aspect of the present application correspond one-to-one, and have the same specific implementation process, which will not be repeated here. Figure 1

[0084] The embodiment determines all hidden danger target segmentation regions by segmenting the two-dimensional image data of the transmission line channel scene, filters out three-dimensional point cloud data projected on each hidden danger target segmentation region according to a calibration relationship between the two-dimensional image data and the three-dimensional point cloud data, and then classifies the three-dimensional point cloud data of each hidden danger target segmentation region, thereby realizing accurate positioning of hidden danger targets in a transmission line channel with different scales in a complex background and significantly improving hidden danger target segmentation performance.

[0085] Referring to Figure 3 , a schematic diagram of an electronic device is given. It should be noted that, Figure 3 the electronic device 300 shown is only an example and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0086] As shown in Figure 3 , the electronic device 300 includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 302 or programs loaded from a storage portion 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for system operation are also stored. The central processing unit 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0087] ​The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including a display such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a local area network (LAN) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as necessary. A removable recording medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 310 as necessary, so that a computer program read out from it is installed in the storage section 308 as necessary.

[0088] The central processing unit 301 in the electronic device of the present embodiment, when executing the program, realizes the steps in the method for locating a target of a transmission line passage hidden danger based on image segmentation as shown in Figure 1

[0089] In particular, according to the embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the method as shown in Figure 1 In such embodiments, the computer program can be downloaded and installed from a network by the communication section 309, and / or installed from the removable recording medium 311. When the computer program is executed by the central processing unit 301, various functions defined in the apparatus of the present application are performed.

[0090] The computer program instructions corresponding to the method as shown in Figure 1 may also be stored in a computer readable storage medium that can direct the computer or other programmable data processing equipment to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction apparatus, which realizes the functions specified in the flow Figure 1 charts or multiple flowcharts and / or the functions specified in one block or multiple blocks. Figure 1 charts or multiple flowcharts and / or the functions specified in one block or multiple blocks.

[0091] It can be understood by those skilled in the art that all or part of the flowcharts in the above-mentioned embodiments can be implemented by a computer program to instruct the relevant hardware, and the program can be stored in a computer readable storage medium, and when the program is executed, it can include the flowcharts of the above-mentioned embodiments of the method. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc. ​

[0092] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. The present application can have various changes and modifications for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for locating hidden danger targets in a transmission line corridor based on image segmentation, characterized in that, The method comprises the following steps: acquiring two-dimensional image data and three-dimensional point cloud data of a power transmission line channel scene; extracting multi-scale global feature maps of the two-dimensional image data of the power transmission line channel scene, fusing all different-scale global feature maps based on a coordinate attention mechanism, and obtaining all hidden danger target segmentation regions in the two-dimensional image data of the power transmission line channel scene based on a mapping relationship between the fused feature maps and the hidden danger target segmentation regions; screening three-dimensional point cloud data projected on each hidden danger target segmentation region based on a calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the power transmission line channel scene; classifying the three-dimensional point cloud data of each hidden danger target segmentation region to obtain point cloud data corresponding to each hidden danger target, and finally realizing positioning of each hidden danger target; the process of fusing all different-scale global feature maps based on the coordinate attention mechanism comprises the following steps: allocating a channel to each different-scale global feature map; for each channel, encoding the corresponding scale global feature map along the horizontal coordinate and the vertical coordinate respectively to obtain corresponding horizontal coordinate and vertical coordinate encoded features; for each channel, splicing the corresponding horizontal coordinate and vertical coordinate encoded features along the spatial dimension, performing regularization and activation processing on the spliced features through normalization and a nonlinear activation function, and then obtaining the fusion feature map of each channel through activation function and normalization weighting; the expression of the fusion feature map of each channel is as follows: wherein, is a regularization and nonlinear activation operation, represents an activation function operation, is a convolution operation, is a fused feature map of the v th channel; respectively represent horizontal and vertical coordinate points; is a horizontal coordinate encoding feature, is a vertical coordinate encoding feature; is a maximum coordinate value of the fused feature map along the horizontal direction; is a maximum coordinate value of the fused feature map along the vertical direction; respectively represent horizontal and vertical coordinates; processing the two-dimensional image data of the power transmission line channel scene by using an image segmentation model to obtain all hidden danger target segmentation regions therein; the image segmentation model comprises a feature extraction network, a cross-level feature fusion network and a small target segmentation head; the feature extraction network is used to extract multi-scale global feature maps of the two-dimensional image data of the power transmission line channel scene; the cross-level feature fusion network is used to fuse all different-scale global feature maps based on a coordinate attention mechanism to obtain a fusion feature map; and the small target segmentation head is used to obtain all hidden danger target segmentation regions in the two-dimensional image data of the power transmission line channel scene based on a mapping relationship between the fusion feature map and the hidden danger target segmentation regions.

2. The image segmentation based transmission line corridor hidden danger object positioning method of claim 1, wherein, for each channel, the process of encoding the corresponding scale global feature map along the horizontal coordinate is as follows: performing one-dimensional mean pooling on the corresponding scale global feature map along the horizontal coordinate direction, starting from the leftmost side of the corresponding scale global feature map, sliding along the width direction with a set window size, and performing average calculation on the pixel values covered in the window to obtain horizontal coordinate encoded features.

3. The image segmentation based transmission line corridor hidden danger object positioning method of claim 1, wherein, for each channel, the process of encoding the corresponding scale global feature map along the vertical coordinate is as follows: performing one-dimensional mean pooling on the corresponding scale global feature map along the vertical coordinate direction, starting from the lowermost side of the corresponding scale global feature map, sliding along the height direction with a set window size, and performing average calculation on the pixel values covered in the window to obtain vertical coordinate encoded features.

4. An image segmentation based power line corridor hidden danger object positioning system, characterized in that, The method is implemented by using the image segmentation-based power transmission line channel hidden danger target positioning method according to any one of claims 1-3, and comprises the following steps: a data acquisition module is configured to acquire two-dimensional image data and three-dimensional point cloud data of a power transmission line channel scene; An image segmentation module is configured to extract multi-scale global feature maps of two-dimensional image data of a transmission line passage scene, fuse all global feature maps of different scales based on a coordinate attention mechanism, and obtain all hidden danger target segmentation regions in the two-dimensional image data of the transmission line passage scene based on a mapping relationship between the fused feature maps and the hidden danger target segmentation regions. A point cloud projection module is configured to filter out three-dimensional point cloud data projected on each hidden danger target segmentation region based on a calibration relationship between the two-dimensional image data and the three-dimensional point cloud data of the transmission line passage scene. A target positioning module is configured to classify the three-dimensional point cloud data of each hidden danger target segmentation region to obtain point cloud data corresponding to each hidden danger target, and finally realize positioning of each hidden danger target.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps in the image segmentation-based transmission line passage hidden danger target positioning method according to any one of claims 1-3.

6. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to implement the steps in the image segmentation-based transmission line passage hidden danger target positioning method according to any one of claims 1-3.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps in the image segmentation-based transmission line passage hidden danger target positioning method according to any one of claims 1-3.

Images