Point group analysis processing device

Abstract

The present disclosure pertains to a point-group analysis processing device that: identifies an object point group representing three-dimensional coordinates of a monitored object, from among point groups which represent three-dimensional coordinates; groups points in the point groups, other than the object point group, as an obstacle point group; and determines that the monitored object is nearing an obstacle on the basis of the distance between the object point group and the obstacle point group.

Description

Point cloud analysis processing device

[0001] The present disclosure relates to an apparatus for determining proximity between structures based on point cloud data.

[0002] Techniques for creating a three-dimensional model of a structure using a three-dimensional laser scanner have been developed (see, for example, Patent Documents 1 and 2). In Patent Documents 1 and 2, a vehicle-mounted three-dimensional laser scanner (Mobile Mapping System: MMS) is used to scan the surface of the structure, thereby obtaining point cloud data representing the three-dimensional coordinates of points on the surface of the structure, and a three-dimensional model is created using the point cloud data.

[0003] The proximity and contact between structures can be detected by using the techniques of Patent Documents 1 and 2. However, when three-dimensional model data is created as in Patent Documents 1 and 2, it takes time to create the three-dimensional model data, making it difficult to detect the proximity and contact between structures in real time.

[0004] JP 2017-156179 A (NTT) JP 2018-195240 A (NTT)

[0005] The present disclosure aims to enable detection of the proximity of a monitoring target to any obstacle using point cloud data without creating three-dimensional model data.

[0006] In order to achieve the above objective, the system disclosed herein comprises a three-dimensional laser scanner that acquires point cloud data representing the three-dimensional coordinates of points on the surface of a structure, and a point cloud analysis processing device that uses the point cloud data to determine whether a monitored object is in proximity to an obstacle.

[0007] The point cloud analysis processing device disclosed herein identifies an object point cloud representing the three-dimensional coordinates of a monitored object from a point cloud representing three-dimensional coordinates, defines points in the point cloud excluding the object point cloud as an obstacle point cloud, and determines whether the monitored object is in proximity to an obstacle based on the distance between the object point cloud and the obstacle point cloud.

[0008] The point cloud analysis processing device of the present disclosure distinguishes between the object point cloud of the monitoring target and the obstacle point cloud, and determines the proximity of the obstacle based on the distance between the point clouds themselves. Therefore, the present disclosure makes it possible to detect the proximity of the monitoring target to any obstacle without creating three-dimensional model data.

[0009] The point cloud analysis processing device of the present disclosure may divide an area in which the obstacle point cloud is distributed into a plurality of partial areas, and determine an area in the area in which the obstacle point cloud is distributed that is close to the object point cloud by calculating a distance between a specific point in each of the plurality of partial areas generated by the division and the object point cloud. In this case, the point cloud analysis processing device of the present disclosure determines that the monitoring target is in proximity to an obstacle by calculating a distance between the obstacle point cloud included in the area close to the object point cloud and the object point cloud.

[0010] The point cloud analysis processing device according to the present disclosure may also reduce the number of at least one of the object point cloud and the obstacle point cloud, and calculate the distance between the reduced object point cloud and the obstacle point cloud.

[0011] The monitoring target may be a utility pole. In this case, the point cloud analysis processing device may identify a point cloud of a moving cylindrical structure from the point cloud, thereby identifying a target utility pole cluster indicating the point cloud of the monitoring target utility pole.

[0012] The above disclosures can be combined as much as possible.

[0013] According to the present disclosure, it is possible to detect the proximity of a monitoring target to any obstacle using point cloud data without creating three-dimensional model data.

[0014] 1 is an example of a system configuration of the present disclosure. FIG. 1 is an example of a point cloud and scan lines. FIG. 2 is an example of a three-dimensional model configured using scan lines. FIG. 1 is an example of a configuration of a point cloud analysis processing device of the present disclosure. FIG. 2 is an example of a proximity determination method executed by the point cloud analysis processing device of the present disclosure. FIG. 3 is an explanatory diagram of the operation of a target candidate extraction unit. FIG. 4 is an explanatory diagram showing an example of a marker. FIG. 5 is an explanatory diagram of the operation of the target candidate extraction unit. FIG. 6 is an explanatory diagram showing an example of circle detection. FIG. 7 is an explanatory diagram of the operation of a target object identification unit. FIG. 8 is an explanatory diagram showing an example of a calculation of the movement amount of a utility pole point cloud. FIG. 9 is an explanatory diagram of the operation of an object distance calculation unit and an alarm issuance unit. FIG. 10 is an explanatory diagram showing an example of an example of division of an obstacle point cloud. FIG. 11 is an explanatory diagram showing an example of a calculation of a distance H. FIG. 12 is an explanatory diagram showing an example of a process for reducing the point cloud of a target utility pole cluster i'. FIG. 13 is an explanatory diagram showing an example of a process for reducing the point cloud of a target utility pole cluster i'. Ｖ The following shows an example of calculating the distance between the center of gravity of the object point cloud and the center coordinate C of the voxel when the obstacle point cloud is clustered by point density. Ｖ An example of calculating the distance between is shown below.

[0015] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Note that the present disclosure is not limited to the embodiments shown below. These implementation examples are merely illustrative, and the present disclosure can be implemented in various forms with various modifications and improvements based on the knowledge of those skilled in the art. Note that components with the same reference numerals in this specification and drawings indicate the same components.

[0016] 1 shows an example of a system configuration according to the present disclosure. The system according to the present disclosure includes a point cloud analysis processing device 91 and a three-dimensional laser scanner 92. The point cloud analysis processing device 91 is a computer that acquires point cloud data from the three-dimensional laser scanner 92 and performs arbitrary processing using the point cloud data.

[0017] The three-dimensional laser scanner 92 is any device capable of acquiring point cloud data representing the three-dimensional coordinates of points on the surface of any structure 93, such as a utility pole 81 or a power line 82, by scanning the surface of the structure. The three-dimensional laser scanner 92 may be, for example, a three-dimensional LiDAR (Light Detection and Ranging) scanner, or may be an MMS.

[0018] The point cloud analysis processing device 91 may create a 3D model of a structure using point cloud data. For example, as shown in FIG. 2, the point cloud analysis processing device 91 can create a scan line 72 by connecting point clouds 71, and then create a 3D model 73 of a utility pole and a 3D model 74 of a power line based on the scan line 72, as shown in FIG. 3. Creating 3D model data in this manner requires time to create the model and calculate facility information. Therefore, the present disclosure uses point cloud data to calculate the distance between objects without creating a model.

[0019] The point cloud analysis processing device 91 of the present disclosure executes the proximity determination method of the present disclosure. Specifically, the point cloud analysis processing device 91 identifies an object point cloud representing the three-dimensional coordinates of a monitoring target from a point cloud representing three-dimensional coordinates, determines points in the point cloud excluding the object point cloud as an obstacle point cloud, and determines whether the monitoring target is in proximity to an obstacle based on the distance between the object point cloud and the obstacle point cloud.

[0020] In this embodiment, an example is shown in which a 3D laser scanner 92 is installed at a construction site where a utility pole is to be buried, point cloud data obtained by the 3D laser scanner 92 is directly input to a point cloud analysis processing device 91, and the point cloud analysis processing device 91 detects whether a utility pole 81 to be monitored is in proximity to any structure. Below, a specific description will be given, with the 3D laser scanner 92 as the reference, with east-west as the x-axis, north-south as the y-axis, and altitude as the z-axis.

[0021] 4 shows an example configuration of the point cloud analysis processing device 91. The point cloud analysis processing device 91 includes a structure detection unit 11 and a distance calculation unit 12. The structure detection unit 11 is a functional unit that identifies a monitoring target from point cloud data, and includes a point cloud file reading unit 111, a target candidate extraction unit 112, and a target identification unit 113. The distance calculation unit 12 is a functional unit that calculates the distance between the monitoring target and other structures, and includes a point cloud reduction unit 121, an object distance calculation unit 122, and an alarm issuance unit 123. The point cloud analysis processing device 91 can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided via a network.

[0022] FIG. 5 shows an example of the proximity determination method of the present disclosure. The point cloud analysis processing device 91 periodically executes steps S101 to S105 during construction work hours. In step S101, the point cloud file reading unit 111 of the point cloud analysis processing device 91 acquires point cloud data. In step S102, the point cloud analysis processing device 91 deletes point clouds in unnecessary ranges. In step S103, the target candidate extraction unit 112 of the point cloud analysis processing device 91 extracts a utility pole candidate point cloud. In step S104, the target identification unit 113 of the point cloud analysis processing device 91 extracts an object point cloud corresponding to the utility pole to be monitored from the utility pole candidate point cloud. In step S105, the point cloud analysis processing device 91 calculates the distance between the extracted object point cloud and the obstacle point cloud. This allows the point cloud analysis processing device 91 to determine whether the monitoring target is in proximity to an obstacle. In step S106, the point cloud analysis processing device 91 causes the alarm issuing unit 123 to issue an alarm based on the calculated distance.

[0023] The point cloud has information on three-dimensional coordinates based on the laser scanner 92. Therefore, based on the three-dimensional coordinates of the point cloud, it is possible to determine the inside and outside of the construction site where the utility pole will be buried, and further to determine the inside and outside of the movement range of the utility pole 81. Therefore, in step S102, based on the three-dimensional coordinates of the point cloud, point clouds other than those that may be close to the utility pole 81 to be monitored are excluded from the point cloud data. For example, point clouds that are more than a predetermined distance from the geographical position of the laser scanner 92 are deleted. This makes it possible to shorten the subsequent processing time in the point cloud analysis processing device 91.

[0024] In the present disclosure, in steps S103 to S104, the point cloud analysis processing device 91 identifies the target utility pole cluster I' indicating the point cloud of the utility pole 81 to be monitored by extracting the point cloud of a moving cylindrical structure from the point cloud.

[0025] 6 shows a first specific example of step S103. The direction of the utility pole 81 to be monitored is input (S311), the tilt angle a of the utility pole 81 is input (S312), the point cloud is corrected according to the tilt angle a so that the central axis of the utility pole 81 is aligned with the z-axis (S313), the point cloud data is grouped at any height in the z-axis direction (S314), the point cloud is projected onto the x-y plane for each group and circle detection is performed (S315), it is determined whether there are groups for which circles can be detected along the z-axis that are equal to or greater than a certain threshold and whether the deviation of the center points of the circles is within the threshold (S316), and the set of groups whose circle center points are within the threshold is designated as utility pole candidate cluster i (S317), and those outside the threshold are excluded from utility pole candidate cluster i (S318).

[0026] If the monitoring target is a utility pole under construction, depending on the construction status, the utility pole may be placed on the ground, held / suspended, or freestanding, and the central axis of the utility pole 81 to be circle-detected may be unknown. Therefore, by performing steps S311 and S312 and inputting the direction of the central axis and inclination angle a of the utility pole 81 to be monitored in advance, it becomes possible to quickly extract the object point cloud, which is the point cloud of the utility pole 81 to be monitored. Steps S311 to S313 are performed at the beginning of construction, and after construction has begun, steps S311 to S313 may be omitted by tracking utility pole candidate cluster i.

[0027] The direction of the utility pole 81 in step S311 is, for example, the angle between the central axis of the utility pole 81 projected onto the xy plane and the x-axis. The tilt angle a of the utility pole in step S312 is, for example, the angle between the central axis of the utility pole and the z-axis. The method of inputting the direction and angle in steps S311 and S312 is arbitrary, and the user may input them into the point cloud analysis processing device 91 on site, or a feature point cloud may be used.

[0028] When using the feature point group, for example, as shown in Fig. 7 , markers are attached to a utility pole 81, a gripping part 83 of heavy equipment, etc., and the direction and angle of the utility pole 81 are calculated using the markers. Here, the markers are objects made of a highly reflective material or structure, or objects with unique shapes, and any object that can be identified at a construction site, such as a scaffolding bolt or a gripping part 83 of heavy equipment, can be used. Furthermore, the markers may be attached only to the utility pole 81, only to the gripping part 83 of the heavy equipment, or to both. The shape of the gripping part 83 of the heavy equipment may also be used as the marker.

[0029] Note that steps S311 and S312 may be omitted, and any method may be used to detect a columnar structure having a central axis in any direction without narrowing down the angle, and extract the utility pole candidate cluster i. When a columnar structure having a central axis in any direction is detected, not only the utility pole 81 to be monitored but also other columnar structures may be included in the utility pole candidate cluster i. Therefore, the target candidate extraction unit 112 executes a process to exclude columnar structures other than the utility pole 81 from the utility pole candidate cluster i. For example, the user sets thresholds such as the distance from the 3D laser scanner 92 and the thickness of the cylindrical structure in the point cloud analysis processing device 91. Furthermore, the user may attach one or more markers to the utility pole 81 to be monitored, thereby enabling the target candidate extraction unit 112 to determine the utility pole 81 to be monitored based on the feature point cloud.

[0030] 8 shows a second specific example of step S103 using feature point clouds. An axis is calculated by connecting two feature point cloud clusters (S321), the point clouds are corrected so that the calculated axis becomes the z-axis (S323), the point cloud data is grouped at an arbitrary height in the z-axis direction (S324), the point cloud is projected onto the x-y plane for each group to detect circles (S325), it is determined whether there are groups for which circles can be detected along the z-axis that are equal to or greater than a certain threshold and whether the deviation of the center points of the circles is within a threshold (S326), the set of groups whose deviation of the center points of the circles is within the threshold is designated as utility pole candidate cluster i (S327), and those outside the threshold are excluded from utility pole candidate cluster i (S328).

[0031] Here, the correction in steps S313 and S323 can be exemplified by rotating the entire 3D point cloud data around an axis. The rotation angle is an angle equivalent to the tilt angle a of the utility pole 81 in step S313, and the angle between the axis calculated in step S321 and the z axis in step S323. Since this embodiment includes steps S311 to S313 and steps S321 to S323, the utility pole candidate cluster i can be generated quickly.

[0032] The grouping in steps S314 and S324 involves dividing the point cloud along the z-axis at predetermined widths, as indicated by dashed lines, and grouping the point cloud data contained within the ranges enclosed by the dashed lines, as shown in G1 to G7 in FIG. 9( a). As shown in FIG. 9( b), the grouped point cloud data is projected onto the xy plane, and circles 75 are detected using the point cloud data on the xy plane. For circle detection, for example, the well-known RANSAC (RANdom SAMple Consensus) process can be used. If a circle is detected in groups G1 to G6, it is determined whether the deviation of the center points of the circles in groups G1 to G6 is within a threshold value. If the deviation is within the threshold value, the set of point clouds contained in groups G1 to G6 is designated utility pole candidate cluster i.

[0033] After steps S315 and S325, the central axis of the cylindrical object may be calculated by connecting the center points of the circles of groups G1 to G6. Furthermore, after extracting utility pole candidate cluster i, the center of gravity or the center of the group of points included in utility pole candidate cluster i may be calculated.

[0034] When the monitored utility pole 81 is nearly upright, the center points of the circles of each group detected in S315 and S325 are aligned on the z-axis, and if the deviation of the center points is within a threshold, utility pole candidate cluster i can be extracted.

[0035] In step S103 of the present disclosure, the first specific example may be used, the second specific example may be used, or both the first and second specific examples may be used. The point cloud analysis processing device 91 may perform these selections in response to manual selection by the user.

[0036] As described above, by the point cloud analysis processing device 91 executing step S103, it becomes possible to extract the leaning utility pole 81 from the point cloud in real time.

[0037] (Operation of the Object Identification Unit 113) Fig. 10 shows a specific example of step S104. As shown in Fig. 11, a cutout range A is set around the utility pole candidate cluster i. ８１ -1 is set (S411), and the range A is the same as the initial state in the point cloud after t seconds. ８１ A point cloud is cut out by -2 (S412), a movement amount D of each utility pole candidate cluster i is calculated by tracking each utility pole candidate cluster i (S413), it is determined whether the movement amount D is equal to or greater than a first threshold and equal to or less than a second threshold (S414), the utility pole candidate cluster i whose movement amount D is equal to or greater than the first threshold and equal to or less than the second threshold is set as a target utility pole cluster i' (S415), and the utility pole candidate cluster i whose movement amount D is less than the first threshold or greater than the second threshold is excluded from the target utility pole cluster (S416).

[0038] Extraction range A in step S411 ８１ For example, if the central axis of the cylindrical shape is the z'-axis, then the cutout range for -1 can be a cm in the x'-axis direction, b cm in the y'-axis direction, and c cm in the z'-axis direction. Here, the cutout range is set to a range such that when utility pole 81 moves, utility pole candidate cluster I will fall within the same cutout range. If the object after movement is not within the cutout range, the amount of movement cannot be calculated, and therefore an error is generated and the object is not considered a utility pole candidate cluster.

[0039] The target utility pole cluster i' is identified in step S415, and all point clouds excluded from the target utility pole cluster i' in step S416 become obstacle point clouds. The target object identification unit 113 performs step S412 every t seconds to identify the target utility pole cluster i' every t seconds. Therefore, the target utility pole cluster i' and obstacle point cloud are identified every t seconds. As described above, the point cloud analysis processing device 91 executes step S104, making it possible to identify the utility pole 81 to be monitored from the point cloud in real time.

[0040] The movement amount D in step S412 is, for example, the center of gravity C of the point group included in the utility pole candidate cluster i. ８１The amount of movement of the electric pole candidate cluster i can be used. In steps S414 and S415, the electric pole candidate cluster i for which the amount of movement D is too small is an electric pole other than a monitoring target that has not moved. On the other hand, if the amount of movement D is too large, the electric pole candidate cluster i is a cylindrical object other than an electric pole, and there is a possibility that the electric pole that was originally a monitoring target has been replaced with another object such as a tree and recognized as such. By determining the amount of movement D using the first and second thresholds in step S414, it is possible to prevent these objects other than the monitoring target from being erroneously determined to be a monitoring target.

[0041] (Object distance calculation unit 122 and alarm issuing unit 123) In the present disclosure, in step S105, the point cloud analysis processing device 91 divides the area in which the obstacle point cloud is distributed into a plurality of partial areas, and calculates a distance between a specific point in each of the plurality of partial areas generated by the division and the center of gravity C of the object point cloud. ８１ The point cloud analysis processing device 91 then determines whether the monitoring target is close to an obstacle by calculating the distance between the obstacle point cloud and the object point cloud included in the area close to the object point cloud. In the present disclosure, an example is shown in which the center coordinates of each partial area are calculated as a "specific point."

[0042] 12 shows a specific example of steps S105 and S106. The object distance calculation unit 122 divides the obstacle point cloud into voxels, and calculates the center coordinates C of each voxel. Ｖ The object distance calculation unit 122 calculates the center of gravity C of the object point cloud included in the target utility pole cluster i′ (S512). ８１ and the center coordinate of the voxel C ＶThe object distance calculation unit 122 calculates the distance H between the obstacle point cloud and the object point cloud (S513). The object distance calculation unit 122 determines whether the distance H is within a threshold value (S611). If the distance H exceeds the threshold value (No in S611), the voxel is not designated as a nearby candidate voxel (S612). On the other hand, if the distance H is within the threshold value (Yes in S611), the voxel is designated as a nearby candidate voxel. The object distance calculation unit 122 calculates the distance H' between the obstacle point cloud and the object point cloud in the voxel whose distance H is within the threshold value (S613). If the minimum value of the distance H' is within the threshold value (Yes in S614), the alarm issuance unit 123 issues an alarm (S615).

[0043] 13 shows an example of voxelization of the obstacle point cloud in step S512. For ease of understanding, the figure shows an example in which the obstacle point cloud is divided into 16 voxels of 4×4 on a two-dimensional plane. In step S512, the center coordinates C of each of the voxels V1 to V16 are calculated. Ｖ If the voxels V9, V10, V11, V13, and V15 have a distance H equal to or less than the threshold, in steps S612 and S613, the object distance calculation unit 122 calculates the distance H' between the object point cloud included in the target utility pole cluster i' and the point clouds within the voxels V9, V10, V11, V13, and V15, and executes the determination in step S614.

[0044] After calculating the distance H, the voxels may be sorted in ascending order of distance H, and the distance H' to a voxel in any order may be calculated to calculate the shortest distance.

[0045] The distance H in step S513 is calculated by dividing all of the points included in the target utility pole cluster i′ by the center coordinates C Ｖ is within a threshold, the center of gravity C of the object point group included in the target utility pole cluster i' is not determined. ８１ and the center coordinate of the voxel C Ｖ Alternatively, after calculating the distance H, a method may be used in which the distance between each point group included in the voxel and each point group included in the target utility pole cluster i' is calculated in a brute force manner in order of decreasing distance H until the distance reaches a certain percentage of the total number of voxels. For example, if there are 100 voxels, and it is desired to calculate up to the order that accounts for 50%, calculations are performed up to the 50th voxel.

[0046] In the present disclosure, the number of object point clouds may be reduced, and the distance between the reduced object point cloud and the obstacle point cloud may be calculated. For example, the object distance calculation unit 122 may execute step S511 of reducing the number of point clouds included in the target utility pole cluster i' before step S512. Point cloud data reduction can be performed by, for example, uniform sampling or thinning by the number of steps, or both. Uniform sampling is performed by thinning a specified range A as shown in FIG. Ｒ This is a method of thinning out points so that there is only one point in the target utility pole cluster i'. Thinning out by step number is a method of assigning a number to each point and thinning out points with the set number, as shown in FIG. 16. The set number is, for example, an even number. Reducing the points included in the target utility pole cluster i' by at least one of these methods can improve the calculation speed.

[0047] In this embodiment, the central coordinate of the voxel is calculated as a "specific point" in step S513, but the "specific point" of each partial region is not limited to the central coordinate of the voxel, but may be the center of gravity of the obstacle point group contained in the voxel, or any point on the edge of the voxel.

[0048] In step S512, the method of dividing the area where the obstacle point cloud is distributed into multiple parts is not limited to voxelization, which divides the area by a predetermined volume, but objects may also be separated by clustering. In step S512, since the division of the entire point cloud only requires the creation of point clusters of a certain size, methods such as k-means, DBSCAN, or HDBSCAN may be used to divide the area by regarding point densities above a certain threshold as one cluster.

[0049] Furthermore, in step S512, after voxelization, the point cloud may be reduced using the same method as in step S511 to improve calculation speed. In this way, when the entire 3D point cloud data is voxelized after point cloud reduction, the accuracy of distance calculation may be improved by calculating the distance H after restoring the reduced data. On the other hand, if the reduced data is not restored, the calculation speed can be improved.

[0050] (Calculation result) FIG. 17 shows the center of gravity C of the object point cloud. ８１ and the center coordinate of the voxel C Ｖ An example of calculating the distance between the center of gravity C of the object point cloud is shown below. ８１ and the center coordinate of the voxel C Ｖ The distance between the center of gravity C and the target point cloud was calculated, and the cumulative processing time was measured. ８１ and the center coordinate of the voxel C Ｖ The position coordinates of the voxel having a small distance from the target point group C ８１ and the center coordinate of the voxel C Ｖ By calculating the distance between the pole 81 and the target pole, the voxel of the obstacle closest to the pole 81 can be identified in a short time.

[0051] Furthermore, by appropriately setting the threshold value of the distance H in step S611, it is possible to significantly reduce the calculation time required to find the closest distance to the utility pole 81 that is the monitoring target.

[0052] 18 shows an example of the processing time when clustering is performed by point density in step S512. The center coordinates of the cluster obtained by clustering the obstacle point cloud by point density are used to calculate the center of gravity C of the object point cloud. ８１ and the center coordinate of the cluster C Ｖ As a result, it was found that by clustering the obstacle point cloud based on point density, it is possible to quickly identify a partial area made up of clusters of the obstacle point cloud that is close to the utility pole 81 to be monitored.

[0053] As described above, the present disclosure uses a 3D laser scanner 92 capable of real-time processing, identifies an object point cloud corresponding to a utility pole 81 to be monitored from among the acquired point clouds, and determines whether the utility pole 81 to be monitored is in proximity to an obstacle based on the distance H' between at least one of the object point clouds and at least one of the obstacle point clouds. In this way, the system of this embodiment can detect the proximity of the monitored object to any obstacle using point cloud data, without creating 3D model data.

[0054] Here, the present disclosure utilizes a 3D laser scanner 92 capable of real-time processing, and executes processes from identifying the object point cloud to calculating the distance H' using a method capable of real-time processing in a point cloud analysis processing device 91. Furthermore, the present disclosure enables accurate identification and proximity detection of monitored objects from the point cloud in real time, regardless of the direction and angle of the object, without post-processing. Therefore, the present disclosure uses point cloud data to accurately calculate the proximity between monitored objects in real time, enabling proximity detection.

[0055] In the above embodiment, an example was shown in which the monitoring target was a utility pole, but the present disclosure is not limited to this and can be applied to any structure for which a point cloud can be acquired by the 3D laser scanner 92. In addition, in the above embodiment, all points other than the object point cloud are set as the obstacle point cloud, but a specific structure can be set as the obstacle point cloud by using a method similar to that for the object point cloud.

[0056] 11: Structure detection unit 12: Distance calculation unit 71: Point cloud 72: Scan line 73: 3D model of utility pole 74: 3D model of power wire 75: Circle 81: Utility pole 82: Power wire 83: Heavy equipment gripping unit 91: Point cloud analysis processing device 92: 3D laser scanner 93: Structure 111: Point cloud file reading unit 112: Target candidate extraction unit 113: Target identification unit 121: Point cloud reduction unit 122: Object distance calculation unit 123: Alarm issuing unit

Claims

1. A point cloud analysis processing device which identifies an object point cloud representing the three-dimensional coordinates of a monitoring target from a point cloud representing three-dimensional coordinates, defines points in the point cloud excluding the object point cloud as an obstacle point cloud, and determines whether the monitoring target is in proximity to an obstacle based on the distance between the object point cloud and the obstacle point cloud.

2. The point cloud analysis processing device according to claim 1, which divides an area in which the obstacle point cloud is distributed into multiple parts, and determines an area in the area in which the obstacle point cloud is distributed that is close to the object point cloud by calculating the distance between a specific point in each of the multiple partial areas generated by the division and the object point cloud, and determines that the monitored object is in proximity to an obstacle by calculating the distance between the obstacle point cloud and the object point cloud included in the area close to the object point cloud.

3. The point cloud analysis processing device according to claim 1, further comprising: reducing the number of at least one of the object point cloud and the obstacle point cloud; and calculating a distance between the object point cloud and the obstacle point cloud after the reduction.

4. The point cloud analysis processing device according to claim 1, wherein the monitored object is a utility pole, and a target utility pole cluster indicating the point cloud of the monitored utility pole is identified by identifying a point cloud of a moving cylindrical structure from the point cloud.

Images