Point cloud processing device and point cloud processing method

Abstract

The present disclosure involves a point cloud processing device that: creates a box on the basis of a point cloud cluster; determines a point cloud of a moving object on the basis of a temporal change in the box; rotates the point cloud of the moving object by rotating the box of the point cloud of the moving object; and extracts the point cloud of the moving object by using the post-rotation point cloud of the moving object.

Description

Point cloud processing device and point cloud processing method

[0001] The present disclosure relates to real-time tracking technology for moving structures.

[0002] A technology for three-dimensionally modeling outdoor structures using an in-vehicle three-dimensional laser scanner (Mobile Mapping System: MMS) has been developed (see, for example, Patent Documents 1 and 2). This technology creates point clouds and scan lines on a space where there are no point clouds, and then creates a three-dimensional model, so that even when the point clouds are sparse (when the vehicle speed is high), good results can be obtained for the reproduction rate and the matching rate.

[0003] It is desired to realize a system that monitors the behavior of objects during construction work, detects proximity and contact between objects, and notifies workers. In the analysis of point clouds, the point cloud of a cylindrical object cannot be correctly extracted unless the central axis of the cylindrical object is set. Automatically setting the central axis of a cylindrical object requires time for arithmetic processing, so it is not suitable for real-time monitoring. Therefore, in Patent Documents 1 and 2, when extracting a cylindrical object, the state of the central axis of the cylindrical object was measured and set in the device.

[0004] Japanese Patent Application Laid-Open No. 2017-156179 International Publication No. 2024 / 023900

[0005] In construction work for moving an elongated structure such as a utility pole, the direction of the central axis of the structure is likely to change. In such an environment, measuring the state of the central axis of the structure and setting it in the device places a burden on the worker.

[0006] Therefore, an object of the present disclosure is to enable extraction of point clouds of a moving structure without setting the central axis of the structure.

[0007] The measurement system of the present disclosure includes a three-dimensional measurement device that measures the shape of a structure, and a point cloud processing device that acquires point cloud data indicating the shape of the structure from the three-dimensional measurement device.

[0008] The point cloud processing device of the present disclosure executes the point cloud processing method of the present disclosure. In the point cloud processing method of the present disclosure, the point cloud processing device creates a box based on the clusters of the point cloud, determines the point cloud of the moving object based on the temporal change of the box, rotates the box of the point cloud of the moving object to rotate the point cloud of the moving object, and extracts the point cloud of the moving object using the rotated point cloud of the moving object.

[0009] The point cloud processing device may rotate the point cloud within the box of the point cloud of the moving object so that the axis of the moving object is in a predetermined direction, using the information of the box of the point cloud of the moving object.

[0010] Note that the above disclosures can be combined as much as possible.

[0011] According to the present disclosure, it is possible to extract the point cloud of a moving structure without setting the central axis of the structure.

[0012] Examples of the configuration of the measurement system of the present disclosure are shown. An example of a scan line is shown. An example of a three-dimensional model of a utility pole and electric wires is shown. An example of a method for extracting a cylindrical object is shown. It is an explanatory diagram of problems in the extraction of a cylindrical object. It is an explanatory diagram of the outline of the present disclosure. An example of the proximity detection method of the present embodiment is shown. An example of the point cloud processing method of the present embodiment is shown. It is an explanatory diagram showing an example of creating a box. It is an explanatory diagram showing an example of the temporal change of a box. It is an explanatory diagram showing an example of the degree of overlap of boxes. It is an explanatory diagram showing an example of the rotation of a box. It is an explanatory diagram showing an example of extracting the box of a utility pole to be monitored. It is an explanatory diagram showing an example of estimating the box at the destination. It is an explanatory diagram showing an example of the movement of a utility pole to be monitored. It is an explanatory diagram showing an example of the rotation of a box. It is an explanatory diagram showing an example of the rotation of a box when the object to be monitored is lost. It is an explanatory diagram showing an example of correcting the rotation angle.

[0013] 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 examples are merely illustrative, and the present disclosure can be implemented in various modified forms based on the knowledge of those skilled in the art. In this specification and the drawings, components having the same reference numerals indicate the same components.

[0014] (Measurement System) Figure 1 shows an example of the configuration of the measurement system of the present disclosure. The measurement system of the present disclosure comprises a point cloud processing device 91 and a three-dimensional measuring device 92. The three-dimensional measuring device 92 measures the shape of a structure. The point cloud processing device 91 is a computer that acquires point cloud data indicating the shape of the structure from the three-dimensional measuring device 92 and performs arbitrary processing using the point cloud data.

[0015] The 3D measuring device 92 is any device capable of acquiring point cloud data representing the 3D coordinates of the surface shape of any structure 93, such as a utility pole 81 or power line 82, by scanning the surface of the structure. For example, a 3D laser scanner can be used. The 3D laser scanner can be, for example, a 3D LiDAR (Light Detection and Ranging) or an MMS. In this embodiment, point cloud data acquired from a 3D LiDAR is used as input as an example, but any point cloud data with 3D information can be used. Point cloud data from a stereo camera or manually created can also be used.

[0016] The point cloud processing unit 91 acquires and updates point cloud data from the 3D measuring device 92. The point cloud data may not be generated by the 3D measuring device 92, such as a 3D laser scanner, but may also be generated from a stereo camera or images. The point cloud processing unit 91 may superimpose point cloud data with a time width of several frames.

[0017] The point cloud processing device 91 may create a three-dimensional model of a structure using the point cloud data. For example, the point cloud processing device 91 extracts an object, creates a three-dimensional model of the object using the extracted point cloud data, and saves the created model information. In creating the three-dimensional model, for example, as shown in Figure 2, the point cloud processing device 91 can create scan lines 72 by connecting the point cloud 71, and based on the scan lines 72, it can create a three-dimensional model 73 of a utility pole and a three-dimensional model 74 of a power line as shown in Figure 3.

[0018] (Method for extracting cylindrical objects) When the direction of the central axis of the utility pole 81 is the z-axis, as shown in Figure 4, the point cloud processing device 91 divides the point cloud along the z-axis at a certain height to extract the point cloud, and detects circles in a planar manner from the extracted point cloud 71. The detected circles are grouped based on their center and size, and based on information such as the height and inclination of the circle group, it is determined whether the point cloud 71 is that of the target cylindrical object (utility pole), and the point cloud 71 corresponding to the utility pole 81 is extracted. By using this cylindrical object extraction method, it becomes possible to extract the point cloud 71 of the monitored utility pole 81 in real time without using a dictionary.

[0019] Here, known methods such as RANSAC processing can be used for circle detection. Furthermore, the parameters of the extraction algorithm can be set to arbitrary values. After circle detection, the center of each circle may be calculated, and the central axis and the diameter of the cylindrical object may be calculated by connecting the centers. Alternatively, cylindrical objects may be extracted from the point cloud, and the identified cylindrical objects may be modeled for each target object.

[0020] (Problems with conventional extraction of cylindrical objects) There is a need to realize a system that monitors the behavior of objects during construction work, detects proximity and contact between objects, and notifies the workers. However, in construction work to install and remove utility poles 81, the central axis of the utility pole 81 changes, such as changing from a lying position to an upright position, or from an upright position to a lying position. In such an environment, measuring the position of the central axis of the utility pole 81 and setting it in the device becomes a burden on the workers.

[0021] Furthermore, if the central axis of the utility pole 81 is misaligned, multiple arcs will be detected from the extracted point cloud 71, as shown in Figure 5, making it impossible to correctly extract cylindrical objects. Therefore, when attempting to apply this to a real-time monitoring of the behavior of the utility pole 81, a problem arises where the utility pole 81 cannot be correctly extracted.

[0022] (Summary of this disclosure) When the central axis of the utility pole 81 changes, the distribution of the point cloud 71 of the utility pole 81 changes in the height direction. Since the point cloud data is the three-dimensional position information itself, the process of extracting the three-dimensional space including the point cloud 71 of the utility pole 81 is very easy. Therefore, in this disclosure, the point cloud 71 of the utility pole 81 is processed using a box which is the three-dimensional space including the point cloud 71 of the utility pole 81.

[0023] As shown in Figure 6, the point cloud processing device 91 is in box B, which contains the point cloud 71 of the utility pole 81 in the Nth frame. Ｎ Create a box containing the point cloud 71 of the utility pole 81. When the direction of the central axis of the utility pole 81 changes in the Nth frame, in the N+1th frame the box containing the point cloud 71 of the utility pole 81 becomes B Ｎ＋１ It changes as shown above. Therefore, box B changes shape at the same position. Ｎ and B Ｎ＋１ By making this determination, it is easy to determine that it is the utility pole 81 that is being monitored.

[0024] Furthermore, as shown in Figure 6, Box B Ｎ and B Ｎ＋１ The inclination of the diagonal corresponds to the inclination of the central axis of the monitored utility pole 81. Therefore, in this disclosure, instead of the central axis of the utility pole 81, a line or diagonal passing through the centroid of the box is used to rotate the point cloud together with the box so that the central axis of the utility pole 81 coincides with the z-axis. Since the process of rotating the point cloud together with the box can be done at very high speed, the point cloud 71 of the monitored utility pole 81 can be extracted at very high speed by extracting cylindrical objects from this rotated point cloud.

[0025] In other words, the point cloud processing device 91 of the present disclosure executes the point cloud processing method of the present disclosure. In the point cloud processing method of the present disclosure, the point cloud processing device 91 creates a box based on the cluster of points (procedure S211 described later), determines the point cloud of the moving object based on the time change of the box (procedure S213 described later), rotates the point cloud of the moving object by rotating the box of the point cloud of the moving object (procedure S215 described later), and extracts the point cloud of the moving object using the rotated point cloud of the moving object (procedure S216 described later).

[0026] Here, the point cloud processing device 91 uses the information of the box of the point cloud of the moving object to rotate the point cloud within the box of the point cloud of the moving object so that the axis of the moving object is in a predetermined direction. In this embodiment, for ease of understanding, an example is shown where the predetermined direction is the z-axis as shown in Figure 4.

[0027] Therefore, this disclosure makes it possible to extract a point cloud of the monitored utility pole 81 in real time without setting the central axis of the cylindrical object. Below, an example of its application to construction work to erect a utility pole 81 that is lying horizontally on the ground will be described in detail.

[0028] (First Embodiment) Figure 7 shows an example of the proximity detection method of this embodiment. The point cloud processing device 91 executes the object extraction procedure S1 and the proximity detection procedure S2. The object extraction procedure S1 includes procedures S11, S12, S13, and S14. The proximity detection procedure S2 includes procedures S21 and S22.

[0029] Procedure S11: The point cloud processing device 91 acquires point cloud data measured by the 3D measuring device 92 in real time at regular intervals. Procedure S12: The point cloud processing device 91 extracts the point cloud 71 of the monitored utility pole 81 and creates a 3D model. Procedure S13: The point cloud processing device 91 saves the information of the created 3D model.

[0030] Procedure S21: Calculate the distance between the saved 3D model and the obstacle and detect proximity. Procedure S22: The alarm generator 94 issues an alarm according to the conditions.

[0031] In step S11, the time interval at which the point cloud processing device 91 acquires data may be every frame, but it can be set to a time that corresponds to the time required to extract the utility poles 81 to be monitored. For example, even if point cloud data is input every 100 ms, if it takes 300 ms to extract the utility poles 81 to be monitored, the point cloud processing device 91 may acquire point cloud data every 4 frames.

[0032] In step S12, the point cloud processing device 91 executes the point cloud processing method of the present disclosure when extracting the point cloud 71 of the monitored utility pole 81. The point cloud processing method of the present disclosure generates a box containing the object extracted by clustering in each frame of the 3D point cloud data, determines that it is the same moving object based on the change of the box between frames, adjusts the angle of the box, extracts a cylindrical object from the point cloud contained in the box, and tracks the extracted cylindrical object in real time in each frame.

[0033] Furthermore, after step S13, the point cloud processing device 91 may track the utility pole 81 to be monitored using the created 3D model.

[0034] According to this embodiment, a moving cylindrical object can be extracted in real time, and an alarm can be triggered according to the distance to an obstacle. Although a cylindrical object is described as an example in this embodiment, it is applicable to all methods of detecting and tracking point clouds of objects along an axis.

[0035] (Second Embodiment) In this embodiment, the details of procedure S12 will be described with reference to Figure 8. The point cloud processing device 91 first executes procedures S211 to S216.

[0036] Procedure S211: Clustering is performed on the point cloud 71 to create clusters. Procedure S212: A box is created for each cluster. Procedure S213: It is determined whether the boxes in the previous frame and the current frame are the same object based on the amount of movement and degree of overlap. This procedure is performed on the two frames that are closest in time. Procedure S214: It is determined that the amount of movement of the center of gravity of the boxes of the same object exceeds a threshold. This makes it possible to determine cylindrical objects whose central axis orientation has changed. Procedure S215: The point cloud processing device 91 rotates the box of the cylindrical object whose central axis orientation has changed so that the central axis of the cylindrical object determined in procedure S214 is perpendicular to the ground. Procedure S216: The point cloud of the cylindrical object is extracted from the point cloud within the rotated box. In this embodiment, since the central axis of the cylindrical object is perpendicular to the ground, the cylindrical object can be extracted with high accuracy using the method described in Figure 4.

[0037] When the point cloud processing device 91 extracts a cylindrical object in step S216, it starts tracking the utility pole 81 as the object to be monitored. Specifically, it executes the following steps S222 to S226, steps S231 to S233, and steps S241 to S242.

[0038] Procedure S221: The point cloud processing device 91 acquires the point cloud for the next frame from the 3D measuring device 92. In this embodiment, the frame in which procedures S211 to S216 are performed is referred to as the Nth frame, and the next frame is referred to as the N+1th frame.

[0039] Procedure S222: The point cloud processing device 91 extracts a box representing the utility pole 81 to be monitored. Procedure S225: Once the point cloud processing device 91 has extracted the box representing the utility pole 81 to be monitored, it rotates the extracted box so that the utility pole 81 is perpendicular to the ground. Procedure S226: The point cloud processing device 91 extracts a point cloud of cylindrical objects from the point cloud within the rotated box. In this embodiment, since the central axis of the cylindrical object is perpendicular to the ground, the cylindrical object can be extracted with high accuracy using the method described in Figure 4.

[0040] Procedure S231: When the point cloud 71 of the cylindrical object was extracted in procedure S226, the point cloud processing device 91 determined that it had been able to track the utility pole 81 and calculated the difference with the previous frame. For example, it calculated the angle difference between the diagonal vector obtained in the Nth frame and the diagonal vector obtained in the (N+1)th frame, and stored the calculation result in the storage medium 91M.

[0041] In this embodiment, when extracting the box of the utility pole 81 to be monitored in step S222, an example is shown in which the destination position of the utility pole 81 to be monitored is estimated. However, a step S232 for performing this estimation may be provided after step S231. In this case, the estimated position of the next cylindrical object is updated in step S232.

[0042] Step S241: When the point cloud of the cylindrical object cannot be extracted in step S226, the point cloud processing device 91 rotates the box at an angle different from that in step S225, and performs the extraction process of the cylindrical object from the rotated point cloud. Step S242: Until the cylindrical object can be extracted, step S241 is repeated a predetermined number of times. When the cylindrical object cannot be extracted even after being repeated the predetermined number of times, the angle φ at which the cylindrical object was last extracted in step S241 is recorded in the storage medium 91M.

[0043] When step S232 or step S242 ends, the point cloud processing device 91 returns to step S221. In step S221, the frame of the point cloud 71 newly acquired from the three-dimensional measurement device 92 becomes the (N + 1)-th frame, and the frames processed in the previous steps S222 to S226, steps S231 to S233, and steps S241 to S242 become the N-th frame.

[0044] (Identification of the moving object in step S212) For example, as shown in FIG. 9, the point cloud processing device 91 creates a box for each cluster using the minimum value C ＭＩＮ and the maximum value C ＭＡＸ of the cluster, and calculates the center of gravity G Ｂ of the box, the diagonal dimension D Ｄ and the inclination A Ｄ of the box. The point cloud processing device 91 may calculate the diagonal vector. Further, the point cloud processing device 91 may calculate the center-of-gravity vector from the end of the box B ＭＩＮ such as the minimum value C ＭＡＸ or the maximum value C Ｎ of the cluster to the center of gravity G Ｎ of the box B Ｎ .

[0045] (Amount of movement and degree of overlap in step S213) For example, as shown in FIG. 10, the point cloud processing device 91 calculates the movement distance D Ｎ from the center of gravity G ＢＮ of the box B Ｎ＋１ in the N-th frame to the center of gravity G ＢＮ＋１ of the box B Ｇ in the (N + 1)-th frame. Further, the box B Ｎ and the box B Ｎ＋１The degree of overlap is calculated. The degree of overlap can be calculated using the ratio of how much the boxes overlap. For example, IoU (Intersection over Union), which shows the ratio of the overlapping area to the joined area, can be used, as shown in Figure 11.

[0046] The point cloud processing device 91 uses this information to determine, for example, the travel distance D Ｇ If the threshold is within the threshold and the IoU is above the threshold, they are determined to be the same. Note that, as a result of the search, previously detected box B Ｎ And then there's box B, which was detected this time and is similar to it. Ｎ＋１ You may treat them as boxes of the same object.

[0047] (Rotation of the box in step S215) When the point cloud processing device 91 determines that an object is moving, it rotates the box B so that the utility pole 81 stands vertically to the ground, as shown in Figure 12. Ｎ Rotate the point cloud contained within it. If the cylindrical object (utility pole) that you want to extract is placed along the X-axis of the point cloud data, rotate it 90° around the Y-axis so that the X-axis points in the same direction as the original Z-axis (perpendicular to the ground). If the cylindrical object (utility pole) is placed along the Y-axis, rotate it 90° around the X-axis. In other words, rotate it so that the axis on which the object is placed is the same as the Z-axis. This process is for when the object is lying on flat ground, and rotating it 90° makes the cylindrical object vertical. There may also be sites on slopes, but the initial box rotation of 90° makes the extraction process of cylindrical objects applicable.

[0048] (Box extraction in procedure S222) In this procedure, the point cloud processing device 91 extracts the boxes of the monitored utility pole 81. At this time, the point cloud processing device 91 may use the box information of the monitored utility pole 81 up to that point to estimate the box of the destination of the monitored utility pole 81. For example, as shown in Figure 10, box B Ｎ and Box B Ｎ＋１ When an overlap occurs, the point cloud processing device 91 processes box B Ｎ Box B of the point cloud 71 that overlaps with this Ｎ＋１This allows for extraction of the point cloud 71 of the monitored utility pole 81 with high accuracy while reducing the extraction time.

[0049] Also, Box B Ｎ Center of gravity G Ｎ Using box B Ｎ＋１ It may be estimated that, for example, the point cloud processing device 91 is located in box B, as shown in Figure 13. Ｎ Center of gravity G Ｎ Using the (N+1)th frame, find the centroid G of the box. Ｆ Estimate the position of the center of gravity G. Ｆ The position estimation is for box B Ｎ Center of gravity G Ｎ The amount of displacement can be calculated by applying it to a Kalman filter.

[0050] Then, as shown in Figure 14, the point cloud processing device 91 estimates the centroid G Ｆ Based on the position of box B, Ｎ Using the height, width, and depth, box B of the point cloud of the monitored utility pole 81 in the N+1th frame. Ｆ It creates a point cloud. The point cloud processing device 91 then processes box B Ｆ Assuming that a point cloud 71 of the utility pole 81 to be monitored exists in box B Ｆ Identify the clusters in the point cloud 71, and mark the boxes of the identified point cloud clusters as B Ｎ＋１ Extract it as follows.

[0051] (Rotation of the box in procedure S225) From the second time onward, the utility pole 81 is tilted as shown in Figure 15. For this reason, the point cloud processing device 91 uses the angle difference of the central axis of the utility pole 81 between the previous Nth frame and the N+1th frame, and the box B of the Nth frame. Ｎ Box B of the N+1th frame for Ｎ＋１ The angle difference θ is calculated using the change and stored for each frame. Here, the angle difference θ is, for example, the centroid vector V of the Nth frame. Ｎ and the centroid vector V of the N+1th frame Ｎ＋１ The angle difference can be used.

[0052] As shown in Figure 16, the point cloud processing device 91 calculates the cumulative value of the angle difference θ θＮ Using this, the rotation angle such that the utility pole 81 is perpendicular to the ground is calculated, and box B Ｎ＋１ Rotate it. For example, the point cloud processing device 91 rotates at an angle (90 - θ). Ｎ ) calculate, box B Ｎ＋１ to (90-θ Ｎ Rotate it by ) degrees. Box B after rotation Ｒ The central axis of the utility pole 81 is approximately parallel to the z-axis. Therefore, box B Ｎ＋１ The point cloud of the utility pole 81 inside can be extracted.

[0053] (Rotation of the box in step S241) When the point cloud 71 of the utility pole 81 is lost in step S226, the point cloud processing device 91 stores the cumulative value of the angle difference θ in the previous step S225. Ｎ Using this, the cumulative value of the angle difference θ in step S225 θ Ｎ Set to a different angle. For example, as shown in Figure 17, the point cloud processing device 91 stores the cumulative value of the angle difference θ up to the M frame in step S225. Ｍ Using this, the rotation angle (90 - θ) Ｍ ) is calculated. Then, the point cloud processing device 91 calculates box B Ｎ＋１ to (90-θ Ｍ Rotate it by ) degrees, and then box B after rotation. Ｒ This determines whether or not a cylindrical object can be extracted.

[0054] This (90-θ) Ｍ If a cylindrical object cannot be extracted due to a rotation of ) degrees, the point cloud processing device 91 changes the value of X and stores the cumulative value of the angle difference θ up to the M+X frame in step S225. Ｍ＋Ｘ Using the rotation angle (90 - θ) Ｍ＋Ｘ ) calculate, box B Ｎ＋Ｘ to (90-θ Ｍ＋Ｘ Rotate it by ) degrees, and then box B after rotation. Ｒ This determines whether a cylindrical object can be extracted. In this way, the point cloud processing device 91 changes the value of X and determines the box B after rotation. Ｒ This determines whether or not a cylindrical object can be extracted.

[0055] As shown in Figure 17, in procedure S241, the cumulative value of the angle difference θ up to the M+X frame is θ Ｍ＋Ｘ Box B after rotation using Ｒ Once a cylindrical object is extracted, the point cloud processing device 91 generates a centroid vector V with respect to the z-axis. Ｒ The angle φ is calculated. The point cloud processing device 91 calculates the rotation angle (90 - θ). Ｍ＋１ This angle φ is corrected. For example, the point cloud processing device 91 corrects the box B after rotation as shown in Figure 18. Ｒ Rotate it by an angle φ, and this angle φ is the result of the rotation of box B. Ａ A cylindrical object is extracted from the point cloud 71. This results in box B of the N+1th frame. Ｎ＋１ Because it can be rotated perpendicular to the ground, the point cloud 71 of the monitored utility pole 81 can be extracted with high accuracy.

[0056] (Example where the point cloud of a cylindrical object cannot be extracted in step S242) In step S241, the point cloud processing device 91 changes the value of X a predetermined number of times, and the rotated box B Ｒ This determines whether a cylindrical object can be extracted. If a point cloud of a cylindrical object cannot be extracted even then, the cumulative value of the angle difference θ used last in step S241 is used. Ｍ＋Ｘ This is stored in the storage medium 91M. This angle φ will be used in step S241 of the next frame.

[0057] Because this embodiment includes steps S241 and S242, it is possible to extract the monitored utility pole 81 even if the monitored utility pole 81 moves more than predicted. Furthermore, even if an obstacle is placed between the monitored utility pole 81 and the 3D measuring device 92, and the point cloud of the monitored utility pole 81 is not included in the data frame from the 3D measuring device 92, it is possible to continue tracking the utility pole 81 without stopping due to an error.

[0058] (Other Embodiments) The point cloud processing device 91 of the present invention 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. The program of this disclosure is a program that causes a computer to realize each of the functions provided in the point cloud processing device 91 according to this disclosure, and is a program that causes a computer to execute each of the procedures provided in the point cloud processing method executed by the point cloud processing device 91 according to this disclosure.

[0059] (Operation and Effects of the Disclosure) As described above, the Disclosure generates boxes containing objects extracted by clustering in each frame of three-dimensional point cloud data continuously acquired by the three-dimensional measuring device 92, determines whether they are the same moving object based on the distance traveled and the degree of overlap of the centroid positions of the boxes between frames, adjusts the angle of the box, and extracts cylindrical objects from the point cloud contained in the box.

[0060] Therefore, this disclosure makes it possible to extract point clouds of cylindrical objects without setting the central axis of the cylindrical object. Here, since the point cloud data itself indicates position coordinates, a box can be calculated in real time. As a result, it becomes possible to extract the cylindrical object to be monitored in each frame in real time, thereby enabling real-time tracking of the cylindrical object.

[0061] This disclosure makes it possible to track and trace a cylindrical object from a lying position to an upright position (and conversely, from an upright position to a lying position) using point cloud information acquired by a real-time processing 3D measuring device 92.

[0062] Furthermore, this disclosure is applicable not only to cylindrical utility poles, but also to point clouds of any object with an axial shape, such as building materials like reinforcing bars and beams.

[0063] 71: Point cloud 72: Scanline 73: 3D model of utility pole 74: 3D model of power line 81: Utility pole 82: Power line 91: Obstacle proximity detection device 91M: Storage medium 92: 3D measuring device 93: Structure 94: Alarm triggering device

Claims

Create a box based on the clusters in the point cloud. The point cloud of the moving object is determined based on the time change of the box. By rotating the box of the point cloud of the moving object, the point cloud of the moving object is rotated. The point cloud of the moving object after rotation is used to extract the point cloud of the moving object. Point cloud processing device. 　 Using the box information of the point cloud of the moving object, the point cloud within the box of the moving object is rotated so that the axis of the moving object is in a predetermined direction. The point cloud processing device according to claim 1. 　 A three-dimensional measuring device for measuring the shape of a structure, A point cloud processing device according to claim 1 or 2, which acquires point cloud data showing the shape of the structure from the three-dimensional measuring device, A measurement system equipped with the following features. 　 Create a box based on the clusters in the point cloud. The point cloud of the moving object is determined based on the time change of the box. By rotating the box of the point cloud of the moving object, the point cloud of the moving object is rotated. The point cloud of the moving object after rotation is used to extract the point cloud of the moving object. Point cloud processing method.

Images