Distance estimation device

Abstract

This invention provides a simple separation distance estimation device that can determine the separation distance between a power transmission line and surrounding objects with a minimal number of processing steps. [Solution] The separation distance estimation device 1 includes a model generation unit 6 that generates a model by machine learning from training data in which a camera mounted on an aircraft takes images of a first power line and structures surrounding the first power line as input data, and a known first distance from the camera to the first power line as input data, and a known first separation distance from the first power line to the structures as output data; and an estimation unit 7 that estimates a second separation distance from the second power line to the objects based on the model, using estimation input data which includes estimation images taken by the camera to a second power line and objects surrounding the second power line, and a known second distance from the camera to the second power line.

Description

Technical Field

[0001] The present invention relates to a separation distance estimation device that estimates the separation distance between a power transmission line and an object existing around it.

Background Art

[0002] Conventionally, the separation distance of a power transmission line has been measured by aerial survey using, for example, a helicopter equipped with a laser measurement device. Also, since the aerial survey by helicopter is conducted on a multi-year cycle, for example, there were cases where the state where the upper end of a growing tree approached the power transmission line could not be grasped immediately. In this case, the safety separation distance between the power transmission line and the objects existing around it is not ensured, and there is a risk of non-compliance with safety standards in the electrical equipment installation standards. To solve such problems, in recent years, technologies for obtaining the distance between a power transmission line and a structure have been developed, and inventions related thereto have already been disclosed.

[0003] Patent Document 1 discloses an invention related to an information processing program, an information processing device, an information processing method, and a model generation method, which is an information processing program for deriving the distance between power transmission and distribution facilities and surrounding objects. The invention disclosed in Patent Document 1 is photographed by a stereo camera, a plurality of images including power transmission and distribution facilities are acquired for a subject, when the plurality of images are input, the acquired plurality of images are input to a first model learned to generate depth information to generate depth information, and when an image is input, the acquired image is input to a second model learned to classify the power transmission and distribution facilities and surrounding objects to obtain the classification of the power transmission and distribution facilities and surrounding objects, and based on the obtained classification and depth information, a process of deriving the distance between the power transmission and distribution facilities and the surrounding objects is executed by a computer.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

[0005] However, the invention disclosed in Patent Document 1 creates a disparity map as depth information from stereo images captured by a stereo camera, and derives the distance between the power transmission and distribution equipment and surrounding objects based on this depth information. Therefore, the invention disclosed in Patent Document 1 imposes the burden of introducing a new stereo camera, and the processing steps involved in deriving the distance between the power transmission and distribution equipment and surrounding objects become numerous, potentially complicating the configuration of the information processing device.

[0006] This invention addresses the aforementioned conventional circumstances and aims to provide a simple separation distance estimation device that can determine the separation distance between a power transmission line and surrounding objects with minimal processing steps and without requiring any new equipment. [Means for solving the problem]

[0007] To achieve the above objective, the first invention is characterized by comprising: a model generation unit that generates a model by machine learning from training data in which a photographing means mounted on an aerial means takes as input data images of a first power line and structures present around the first power line and a known first distance from the photographing means to the first power line, detects the first power line and structures from the images and outputs a known first separation distance from the first power line to the structures as output data; and an estimation unit that takes as estimation input data images of a second power line and objects present around the second power line and a known second distance from the photographing means to the second power line, detects the second power line and objects from the estimation images and estimates and outputs a second separation distance from the second power line to the objects based on the model.

[0008] In this invention, the flight means is capable of flying along a predetermined flight path (for example, one described by three-dimensional coordinates of latitude, longitude, and altitude), and a drone is used as an example. The photographic means is a monocular camera, for example. Furthermore, the structures and objects are, for example, trees and buildings surrounding the power lines, other power lines strung alongside the first power line, heavy machinery such as cranes, etc.

[0009] In the invention with the above configuration, the model generation unit detects the first power transmission line and structures present around the first power transmission line on images taken of the first power transmission line and structures present around the first power transmission line, and generates a model that learns the geometric relationship between a known first distance and a first separation distance. Furthermore, the estimation unit detects the second power line and objects present around the second power line from estimation images taken of the second power line and the objects present around the second power line, and estimates and outputs a second separation distance corresponding to a known second distance based on the model. Therefore, without determining the reduction ratio of the estimation image, an absolute value (e.g., meters) corresponding to the second distance is estimated as the second separation distance.

[0010] The second invention is that, in the first invention, the image shows the first power line above the structure, and the estimation image shows the second power line above the object. The model generation unit detects the center of the image, the first power line intersection where the structural line connecting the center and the structure intersects the first power line, and the structural intersection where the structural line intersects the structure, from the image, and the first power line intersection and structure corresponding to the first distance and the distance between the center and the first power line intersection. The method generates a model that learns the distance to the intersection point of the structure as the first separation distance. The estimation unit detects, from the estimation image, the estimation center of the estimation image, the second power line intersection point where the object line connecting the estimation center and the object intersects the second power line, and the object intersection point where the object line intersects the object. The estimation unit then estimates the distance between the second power line intersection point and the object intersection point, which corresponds to the second distance and the distance between the estimation center and the second power line intersection point, as the second separation distance.

[0011] In this configuration, the structural line rises or falls along the vertical direction of the image, from the center of the image to the structure. Therefore, the center of the image, the first power line intersection, and the structure intersection are all located on the structural line. Similarly, the object line rises or falls along the vertical direction of the estimation image, from the estimation center of the estimation image to the object. Therefore, the estimation center of the estimation image, the second power line intersection, and the object intersection are all located on the object line.

[0012] In the invention with the above configuration, in addition to the operation of the first invention, the model generation unit generates a model that learns the relationship between the first distance and the first separation distance, etc., using an image, and the estimation unit estimates the second separation distance corresponding to the known second distance using this model and an estimation image. Furthermore, since the center of the image corresponds to the optical axis of the camera equipped with the shooting means and does not move within the image, the points at a first distance and points at a first separation distance on the structural line descending from the center of the image to the structure are accurately detected on the image by the model generation unit.

[0013] Furthermore, the estimation center of the estimation image, like the center of the image, corresponds to the optical axis of the camera and does not move within the image. Therefore, the estimation unit accurately detects the second distance point and the first separation distance point on the object line descending from the estimation center to the object, respectively, on the estimation image.

[0014] The third invention is characterized in that, in the first or second invention, it is further equipped with a display unit that displays an estimation result based on a second separation distance, wherein the estimation result is a second separation distance relating to the distance between one tower supporting one end of the second power transmission line and the other tower supporting the other end of the second power transmission line. In an invention with such a configuration, the second separation distance between transmission towers is obtained as estimation input data from a plurality of estimation images taken by the imaging means at predetermined timings while the flight means flies between one transmission tower and the other transmission tower, maintaining a known second distance from the imaging means to the second power transmission line. In the invention with the above configuration, in addition to the effects of the first or second invention, a second separation distance is obtained with respect to the distance between transmission towers, so it becomes clear whether or not a safe separation distance is secured between the transmission line and the object with respect to the distance between transmission towers. [Effects of the Invention]

[0015] According to the first invention, the absolute value of the second separation distance can be estimated without determining the reduction ratio of the estimation image, so the second separation distance between the power line and the object present around it can be determined with fewer processing steps.

[0016] According to the second invention, in addition to the effects of the first invention, the estimation unit can accurately estimate a second separation distance corresponding to a known second distance using an estimation image based on a model generated by the model generation unit.

[0017] Furthermore, according to the second invention, a second separation distance corresponding to two types of estimation input data, an estimation image and a second distance, can be estimated based on a model generated from three types of training data: an image, a first distance, and a first separation distance. Therefore, because the number of types of training data input during model generation and the number of estimation data input during estimation are small, the separation distance estimation device is easy to handle and user-friendly.

[0018] According to the third invention, in addition to the effects of the first or second invention, it becomes clear whether or not a safe separation distance is secured between the second transmission line and the object with respect to the distance between transmission towers, so that locations that do not conform to the safety standards in the electrical equipment installation standards can be identified. [Brief explanation of the drawing]

[0019] [Figure 1] This is a diagram showing the configuration of a separation distance estimation device according to an embodiment. [Figure 2] This is a process diagram of the model generation method executed by the model generation unit of the separation distance estimation device according to the embodiment. [Figure 3]This is an example of a layout diagram when taking an image which is input data of teacher data. [Figure 4] This is an example of an image which is input data of teacher data corresponding to FIG. 3. [Figure 5] This is an example of a layout diagram when taking an image which is input data of teacher data. [Figure 6] This is an example of an image which is input data of teacher data corresponding to FIG. 5. [Figure 7] This is an example of a layout diagram when taking an image which is input data of teacher data. [Figure 8] This is an example of an image which is input data of teacher data corresponding to FIG. 7. [Figure 9] This is a process diagram of a separation distance estimation method executed by the separation distance estimation device according to the embodiment. [Figure 10] This is a layout diagram when taking an estimation image which is estimation input data. [Figure 11] This is an image which is estimation input data. [Figure 12] This is an estimation result based on the second separation distance.

Embodiments of the Invention

Examples

[0020] The separation distance estimation device according to the embodiment of the present invention will be described in detail with reference to FIGS. 1 to 12. FIG. 1 is a configuration diagram of the separation distance estimation device according to the example. As shown in FIG. 1, the separation distance estimation device 1 according to the example is a computer including a communication unit 2, an input unit 3, an output unit 4, a control unit 5, a display unit 9, and a storage unit 10. Among these, the control unit 5 is a central processing unit that controls the operation of the separation distance estimation device 1, and includes a model generation unit 6, an estimation unit 7, and a calculation unit 8. The storage unit 10 includes a model storage unit 10a and an output storage unit 10b. Furthermore, the training data storage unit 11 is a server that stores training data for the model generation unit 6 to generate a model. The estimation data storage unit 12 is a server that stores estimation input data, etc., for the estimation unit 7 to estimate the second separation distance. The training data storage unit 11 and the estimation data storage unit 12 can communicate with the separation distance estimation device 1 via the communication unit 2 and the network 60, respectively.

[0021] Next, the communication unit 2 to the storage unit 10 that constitute the separation distance estimation device 1 will be described. The communication unit 2 is a means by which the distance estimation device 1 communicates with the training data storage unit 11 and the estimation data storage unit 12 via the network 60. The input unit 3 acquires the training data and estimation input data received by the communication unit 2 and transmits them to the control unit 5, while the output unit 4 is a means for transmitting the second separation distance and the like output by the control unit 5 to the communication unit 2.

[0022] The model generation unit 6 generates a model from the input and output data of the training data using machine learning. This machine learning is performed using a known method, and the generated model is stored in the model storage unit 10a. Furthermore, the input data for the training data consists of images taken by a camera mounted on a drone (the means of flight) as it flies along the first power line and photographs the first power line and structures surrounding it, and a known first distance from the camera to the first power line. Since the camera is a monocular camera and is integrated with the flight means, the first distance can be the distance from the flight means to the first power line.

[0023] The estimation unit 7 estimates the second separation distance from the estimation input data, based on the model generated by the model generation unit 6 and stored in the model storage unit 10a. The estimated second separation distance is temporarily stored in the output storage unit 10b. The input data for estimation consists of estimation images taken by a camera mounted on the flight device, which photographs the second power line and objects surrounding the second power line, and a known second distance from the camera to the second power line. The estimation unit 7 then detects the second power line and the objects from the estimation images and estimates a second separation distance from the second power line to the objects.

[0024] The calculation unit 8 calculates the estimation result based on the second separation distance. The estimation result is the second separation distance for the distance between transmission towers, and is stored in the output storage unit 10b. Such a second separation distance is estimated for each of the multiple estimation images taken by the flight means while flying along the second power line while maintaining a constant distance from the second power line. This estimation result will be explained with reference to Figure 12. The display unit 9 is specifically a monitor that displays the estimation results calculated by the calculation unit 8 as a graph or table.

[0025] Next, the model generation method 20 executed by the model generation unit 6 of the distance estimation device 1 will be explained using Figures 2 to 4. Figure 2 is a process diagram of the model generation method executed by the model generation unit of the distance estimation device according to the embodiment. Figure 3 is an example of a layout diagram when taking images which are input data for training data. Figure 4 is an example of an image which is input data for training data, corresponding to Figure 3. As shown in Figure 2, the model generation method 20 comprises the training data acquisition step S21 and the model generation step S22. The model generation step S22 further comprises the detection step S22-1 and the interval detection step S22-2. The S21 training data acquisition process is a process in which the model generation unit 6 acquires training data from the training data storage unit 11 via the network 60, the communication unit 2, and the input unit 3. Here, the method for capturing images included in the training data will be explained using Figure 3.

[0026] As shown in Figure 3, the images that serve as the input data for the training data are obtained when the imaging means 42 photographs the first power transmission line 40 and the structures 41 surrounding the first power transmission line 40. These images are detected by the image sensor (not shown) of the imaging means 42, which detects the images projected onto a plane S perpendicular to the optical axis 42a of the camera of the imaging means 42, where the first power transmission line 40 and the structures 41 are located. In Figure 3, the first power transmission line 40 and the imaging means 42 are positioned by a flight means (not shown) at an altitude higher than the altitude of any vertical cross-section 40a of the first power transmission line 40 in the vertical direction V. Therefore, the optical axis 42a is also positioned at an altitude higher than the altitude of the vertical cross-section 40a.

[0027] Furthermore, the first distance D1 is the distance of the line segment 48 that connects the imaging means 42 and the center point of the vertical section 40a, and is inclined at an angle α with respect to the horizontal direction H. Furthermore, the distance from the center point of an arbitrary vertical section 40a to the upper end 41a of the structure 41 is the first separation distance R1. In the training data, both the first distance D1 and the first separation distance R1 are known. Also, the first separation distance R1 is a value with respect to the vertical direction V. However, if the structure 41 is located diagonally below the vertical direction V of the first power transmission line 40, as shown by the dashed line, the first separation distance R1' is a value in the direction inclined with respect to the vertical direction V.

[0028] Next, the model generation process of S22 will be explained using Figures 2 and 4. In the model generation process of S22, the detection process of S22-1 is performed by the model generation unit 6, as shown in Figure 4, on the image 43 captured by the imaging means 42, detecting the center 44 of the image 43, the first power transmission line 40, the structure 41, the first power transmission line intersection 46 where the structure line 45 connecting the center 44 and the structure 41 intersects the first power transmission line 40, and the structure intersection 47 where the structure line 45 intersects the upper end 41a of the structure 41. In this way, if there are multiple structure intersections 47 due to the shape of the structure 41 (for example, an irregular shape), the structure intersection 47 closest to the first power transmission line intersection 46 is selected. The above detection is performed using known detection methods such as edge enhancement and AI-based object recognition technology. Note that the structural line 45 is a straight line parallel to the plane S (see Figure 3) perpendicular to the optical axis 42a.

[0029] Furthermore, in image 43, the optical axis 42a of the imaging means 42 (see Figure 3) coincides with the center 44 of image 43. Also, the distance K1 between the center 44 of image 43 and the first power line intersection 46 is above the first power line 40 (in the Y1 direction, which is parallel to the plane S (see Figure 3) perpendicular to the optical axis 42a). Therefore, the distance L1 between the first power line intersection 46 and the structure intersection 47 and the first separation distance R1 are not geometrically identical. Furthermore, the structural line 45 is a virtual line, and when the optical axis 42a is positioned at an altitude higher than the altitude of the vertical cross-section 40a of the first power transmission line 40, it is conceived as a straight line extending downward from the center 44 of the image 43 (in the Y2 direction, which is the opposite direction to the Y1 direction).

[0030] In the subsequent interval detection step S22-2, as shown in Figure 4, the model generation unit 6 detects the first distance D1 (see Figure 3), the interval K1 between the center 44 and the first power line intersection 46, and the interval L1 between the first power line intersection 46 and the structure intersection 47. Furthermore, the model generation unit 6 learns the relationship between the first distance D1, the interval K1, the interval L1, and the first separation distance R1 (see Figure 3), and generates a model. This learning is performed using a known machine learning method. Since the structure intersection 47 closest to the first power line intersection 46 is selected, the interval L1 between the first power line intersection 46 and the structure intersection 47 is the shortest distance. The model generation process in S22 is completed by executing the interval detection process in S22-2.

[0031] Next, we will explain the input and output data for training data using images taken in a different configuration than those shown in Figures 3 and 4, with reference to Figures 5 through 8. Figure 5 is an example of a configuration diagram when taking images that serve as input data for training data. Figure 6 is an example of an image that serves as input data for training data, corresponding to Figure 5. Figure 7 is another example of a configuration diagram when taking images that serve as input data for training data. Figure 8 is an example of an image that serves as input data for training data, corresponding to Figure 7. Furthermore, the same reference numerals are used for the components shown in Figures 1 to 4 in Figures 5 to 8, and their explanations are omitted. As shown in Figure 5, the imaging means 42 is positioned at an altitude lower than the altitude of any vertical cross-section 40a of the first power transmission line 40 in the vertical direction V. At this time, the optical axis 42a is also positioned at an altitude lower than the altitude of the vertical cross-section 40a. Furthermore, the first distance D1 is the distance of the line segment 48 that is inclined at an angle α with respect to the horizontal direction H.

[0032] In this case, as shown in Figure 6, the distance K1 between the center 44 of image 43 and the intersection 46 of the first power line 43 lies below the first power line 40 (in the Y2 direction, which is parallel to the plane S (see Figure 3) perpendicular to the optical axis 42a). The structural line 45 is conceivable as straight lines extending from the center 44 of image 43 in the Y1 direction and the Y2 direction, respectively, when the optical axis 42a is positioned at an altitude lower than the altitude of the vertical cross-section 40a of the first power line 40. Furthermore, the distance L1 between the intersection 46 of the first power line 43 and the structural intersection 47 and the first separation distance R1 are not geometrically identical.

[0033] Next, as shown in Figure 7, the imaging means 42 is positioned at an altitude equivalent to the altitude of any vertical cross-section 40a of the first power transmission line 40 in the vertical direction V. At this time, the optical axis 42a of the camera of the imaging means 42 is also positioned at an altitude equivalent to the altitude of the vertical cross-section 40a. Furthermore, the first distance D1 is the distance of line segment 48 along the horizontal direction H. In addition, the optical axis 42a is also parallel to the horizontal direction H. In this case, as shown in Figure 8, the distance K1 between the center 44 of image 43 and the first power line intersection 46 is zero. Furthermore, the structural line 45 is conceived as a straight line extending in the Y2 direction from the center 44 of image 43 when the optical axis 42a is positioned at an altitude equivalent to the altitude of the vertical section 40a. Also, the distance L1 between the first power line intersection 46 and the structural intersection 47 and the first separation distance R1 are geometrically identical. Figures 5 to 8 show the case where the structure 41 is located vertically below the first power transmission line 40, but as in Figure 3, the structure 41 may be located diagonally below the vertical direction V of the first power transmission line 40.

[0034] Next, the distance estimation method 30 performed by the distance estimation device 1 will be explained using Figures 9 to 11. Figure 9 is a process diagram of the distance estimation method performed by the distance estimation device according to the embodiment. Figure 10 is a layout diagram when capturing the estimation image, which is the input data for estimation. Figure 11 is the image, which is the input data for estimation. As shown in Figure 9, the separation distance estimation method 30 comprises the estimation data acquisition step S31, the estimation step S32, the calculation step S33, and the display step S34. The estimation step S32 further comprises the detection step S32-1 and the second separation distance estimation step S32-2. The estimation data acquisition step in S31 is a step in which the estimation unit 7 acquires the estimation input data stored in the estimation data storage unit 12 via the network 60, the communication unit 2, and the input unit 3. Here, the method for capturing the estimation image will be explained using Figure 10.

[0035] As shown in Figure 10, the estimation image 52 (see Figure 11), which is the input data for estimation, is obtained when the imaging means 42 photographs the second power line 50 and the object 51 present around the second power line 50. This estimation image 52 is the image of the second power line 50 and the object 51 projected onto a plane S perpendicular to the optical axis 42a, which is detected by the image sensor (not shown) of the imaging means 42. Furthermore, the known second distance D2 is the distance of the line segment 57 that connects the imaging means 42 and the center point of any vertical cross-section 50a of the second power transmission line 50, and is inclined at an angle α with respect to the horizontal direction H. Furthermore, the distance from the center of any vertical cross-section 50a to the upper end 51a of the object 51 is the second separation distance R2 to be estimated. Since the object 51 is located below the second power transmission line 50 along the vertical direction V, the second separation distance R2 is a value with respect to the vertical direction V.

[0036] Therefore, as shown in Figure 11, the estimation image 52, which is the input data for estimation, is displayed with the optical axis 42a of the shooting means 42 (see Figure 10) coinciding with the estimation center 53 of the estimation image 52, and the second power line 50 is displayed in the Y1 direction above the object 51.

[0037] Next, the estimation process in S32 will be explained using Figures 9 and 11. In the estimation process in S32, the detection process in S32-1 is performed by the estimation unit 7, as shown in Figure 11, on the estimation image 52, detecting the estimation center 53 of the estimation image 52, the second power line intersection 55 where the object line 54 connecting the estimation center 53 and the object 51 intersects with the second power line 50, and the object intersection 56 where the object line 54 intersects with the upper end 51a of the object 51. This detection is performed using the same method as the detection method performed in the detection process in S22-1. Also, similar to the case of the structure intersection 47, the object intersection 56 closest to the second power line intersection 55 is selected.

[0038] In the subsequent second separation distance estimation step S32-2, the estimation unit 7 estimates a second separation distance R2 (see Figure 10) based on the model generated by the model generation unit 6, which corresponds to the known second distance D2 (see Figure 10), the distance K2 between the estimation center 53 and the second power line intersection 55, and the distance L2 between the second power line intersection 55 and the object intersection 56. Furthermore, since the object intersection 56 closest to the second power line intersection 55 is adopted, the distance L2 between the second power line intersection 55 and the object intersection 56 becomes the shortest distance.

[0039] Furthermore, the calculation step in S33 is a step in which the calculation unit 8 calculates an estimation result based on a second separation distance R2 for the distance between transmission towers. The distance between transmission towers is the distance between one transmission tower that supports one end of the second transmission line 50 and the other transmission tower that supports the other end of the second transmission line 50. This distance between transmission towers is described, for example, in kilometers. This description in kilometers can be calculated from three-dimensional coordinate values ​​(latitude, longitude, altitude), which are the flight routes of flight means such as drones, using a known method, since these three-dimensional coordinate values ​​are pre-set. Therefore, the distance between transmission towers described in kilometers is created in a table format, associated with the three-dimensional coordinate values ​​for each transmission tower supporting the second power transmission line 50, and stored in the estimated data storage unit 12 or the output storage unit 10b. In addition, the transmission tower numbers that identify each transmission tower and their three-dimensional coordinate values ​​are also stored in the table. Furthermore, the second separation distance R2 is described, for example, in kilometers. This second separation distance R2 in kilometers can be obtained by expressing the first separation distance R1 in kilometers.

[0040] Furthermore, the second separation distance R2 is estimated using multiple estimation images taken while the flight device is flying between one tower and the other as estimation input data, and can therefore be associated with three-dimensional coordinate values. Thus, the second separation distance R2, described in kilometers, can be stored in the table described above. This allows for the estimation result of the second separation distance R2 for the distance between towers to be obtained. This estimation result is transmitted to the estimation data storage unit 12 via the output unit 4, the communication unit 2, and the network 60, and stored therein.

[0041] The display step in S34 is a step in which the control unit 5 displays the estimation result calculated by the calculation unit 8 on the display unit 9. As shown in Figure 12, the estimation results are displayed as a graph with the horizontal axis representing the distance between transmission towers (km) and the vertical axis representing the relative distance (m) between the second transmission line 50 and the object 51. The safety separation distance (dashed line) according to the electrical equipment installation standards is also displayed. With this display method, the second separation distance R2 for the distance between transmission towers can be understood as the difference in the vertical axis direction between the second transmission line 50 and the object 51. The separation distance estimation method 30 is terminated by the execution of the display step in S34.

[0042] As explained above, the separation distance estimation device 1 can estimate the absolute value of the second separation distance R2 without determining the reduction ratio of the estimation image 52. Therefore, the second separation distance R2 can be determined with fewer processing steps and without the need for new equipment. Furthermore, the estimation unit 7 can accurately obtain the second separation distance R2 by estimating the second separation distance R2 corresponding to the known second distance D2 using the estimation image 52 based on the model generated by the model generation unit 6.

[0043] Furthermore, as explained in Figures 3 to 8, as long as the imaging means 42 captures images 43 or estimation images 52 such that the first power line 40 and the structure 41, or the second power line 50 and the object 51, do not overlap, the second separation distance R2 can be accurately determined regardless of whether the altitude of the imaging means 42 is higher, lower, or equal to the altitude of the vertical cross-section 40a of the first power line 40 or the vertical cross-section 50a of the second power line 50. Furthermore, since the distance L2 between the second power line intersection 55 and the object intersection 56 is the shortest distance, the estimated second separation distance R2 is also the shortest distance. Therefore, the reliability of the second separation distance R2 is good.

[0044] Furthermore, the estimation results displayed by the display unit 9 clearly indicate whether or not a safe clearance distance is secured between the power lines and the object in the direction of the distance between transmission towers, making it possible to identify locations that do not comply with the safety standards for electrical equipment installation. Furthermore, since the second separation distance R2 between the transmission towers can be understood as the difference in the vertical axis direction between the second transmission line 50 and the object 51, it is possible to immediately determine whether the second separation distance R2 exceeds the safe separation distance and is approaching the second transmission line 50. In addition, by displaying the tower numbers that identify one and the other transmission towers, it is possible to correctly understand which transmission towers the estimation result for is being displayed. In addition, by searching for estimation images in which a safe clearance distance has not been ensured, it is possible to visually identify the type of object that caused the safety non-compliance.

[0045] It should be noted that the separation distance estimation device according to the present invention is not limited to those shown in the embodiments. For example, the calculation step in S33 and the display step in S34 may be omitted. Also, in the display step in S34, the estimation results may be displayed on the display unit 9 as a table, in addition to the graph shown in Figure 12. [Industrial applicability]

[0046] This invention can be used as a distance estimation device for estimating the distance between a power transmission line and an object present in its vicinity. [Explanation of symbols]

[0047] 1…Separation distance estimation device 2…Communication unit 3…Input unit 4…Output unit 5…Control unit 6…Model generation unit 7…Estimation unit 8…Calculation unit 9…Display unit 10…Storage unit 10a…Model storage unit 10b…Output storage unit 11…Training data storage unit 12…Estimation data storage unit 20…Model generation method 30…Separation distance estimation method 40…First power transmission line 40a…Longitudinal section 41…Structure 41a…Upper end 42…Shooting means 42a…Optical axis 43…Image 44…Center 45…Structural line 46…Intersection of first power transmission line 47…Intersection of structure 48…Line segment D1…First distance R1, R1'…First separation distance 50…Second power transmission line 50a…Longitudinal section 51…Object 51a…Upper end 52…Estimation image 53... Estimation center 54... Object line 55... Second transmission line intersection 56... Object intersection 57... Line segment D2... Second distance R2... Second separation distance 60... Network

Claims

1. A model generation unit generates a model using machine learning from training data, which is a camera mounted on an aerial means that takes images of a first power line and structures surrounding the first power line as input data, and a known first distance from the camera to the first power line as input data, and detects the first power line and the structures from the images, and outputs a known first separation distance from the first power line to the structures as output data. Separation distance estimation device characterized in that the imaging means takes estimation images of a second power line and objects present around the second power line, and a known second distance from the imaging means to the second power line as estimation input data, and includes an estimation unit that detects the second power line and the objects from the estimation images and estimates a second separation distance from the second power line to the objects based on the model.

2. The aforementioned image shows the first power transmission line positioned above the structure. The aforementioned estimation image shows the second power line above the object, The model generation unit detects, from the image, the center of the image, the first power line intersection point where the structural line connecting the center and the structure intersects the first power line, and the structural intersection point where the structural line intersects the structure, and generates the model that learns the relationship between the first distance, the distance between the center and the first power line intersection point, the distance between the first power line intersection point and the structural intersection point, and the first separation distance. The separation distance estimation device according to claim 1, characterized in that the estimation unit detects, from the estimation image in addition to the second power line and the object, the estimation center of the estimation image, the second power line intersection point where the object line connecting the estimation center and the object intersects the second power line, and the object intersection point where the object line intersects the object, and estimates the second separation distance corresponding to the second distance, the distance between the estimation center and the second power line intersection point, and the distance between the second power line intersection point and the object intersection point.

3. The system includes a display unit that displays the estimation result based on the second separation distance, The separation distance estimation device according to claim 1 or 2, characterized in that the estimation result is the second separation distance between one tower supporting one end of the second power transmission line and the other tower supporting the other end of the second power transmission line.

Images