JP7816581B2 - Extraction device, extraction method, and extraction program - Google Patents

Description

The disclosed technology relates to an extraction device, an extraction method, and an extraction program.

Conventionally, there is a technology for creating 3D models of outdoor structures using a vehicle-mounted 3D laser scanner (Mobile Mapping System: MMS) (Patent Document 1). This technology creates a point cloud and scan lines in a space where no point cloud exists, and then creates a 3D model, thereby achieving good results even when the point cloud is sparse (when the vehicle is traveling at high speed).

JP 2017-156179 A

It is conceivable to use point clouds acquired by 3D LiDAR (Light Detection and Ranging) to achieve real-time cylindrical object extraction to avoid contact with obstacles during construction. However, conventional technology extracts cylindrical objects by performing circle detection under the assumption that the cylindrical object is upright relative to the ground. In addition, creating scan lines and creating a 3D model requires time, making real-time object detection difficult. During construction, cylindrical objects are not limited to being upright; they may be tilted in the air or placed on the ground. Therefore, the point cloud may be arranged in a circular shape along a certain axis, or it may not be arranged in a circular shape due to obstacles. Even when circle detection is performed using conventional technology, as the tilt of the cylindrical object increases, the arrangement of the point cloud changes from that of an upright object, making circle detection impossible. Furthermore, the number of acquired point clouds may be small, and the object may not be recognized as a circle. As a result, real-time extraction of cylindrical objects in positions other than upright has been problematic.

The disclosed technology has been developed in consideration of the above points, and aims to provide an extraction device, extraction method, and extraction program that are capable of extracting cylindrical objects even when the cylindrical objects are tilted.

A first aspect of the present disclosure is an extraction device that includes an acquisition unit that acquires the tilt angle of cylindrical objects included in an input point cloud, a conversion unit that converts the coordinates of the point cloud using the acquired angle so that the tilt angle of the cylindrical objects becomes a reference angle, and an extraction unit that extracts cylindrical objects from the point cloud based on the converted coordinates of the point cloud.

A second aspect of the present disclosure is an extraction method in which an acquisition unit acquires the tilt angle of a cylindrical object included in an input point cloud, a conversion unit uses the acquired angle to convert the coordinates of the point cloud so that the tilt angle of the cylindrical object becomes a reference angle, and an extraction unit extracts the cylindrical object from the point cloud based on the converted coordinates of the point cloud.

A third aspect of the present disclosure is an extraction program that causes a computer to execute the following process: for an input point cloud, obtain the tilt angle of a cylindrical object contained in the point cloud; use the obtained angle to transform the coordinates of the point cloud so that the tilt angle of the cylindrical object becomes a reference angle; and extract the cylindrical object from the point cloud based on the transformed coordinates of the point cloud.

The disclosed technology makes it possible to extract cylindrical objects even if they are tilted.

FIG. 10 is a diagram illustrating an example of height groups of a point cloud. FIG. 10 is a diagram showing how circles are detected in a plane from a point cloud of height groups. 10 shows an example of a process for extracting a cylindrical object according to this embodiment. FIG. 2 is a block diagram showing a hardware configuration of the extraction device. 1 is a block diagram showing a configuration of an extraction device according to an embodiment of the present invention; FIG. 2 is a block diagram showing the configuration of an extraction unit of the extraction device of the present embodiment. 10 is a flowchart showing the flow of extraction processing by the extraction device. FIG. 2 is a diagram showing x-axis, y-axis, and z-axis based on the measurement points of the point cloud measuring instrument. 10 is a flowchart showing the flow of extraction processing by the extraction device. 10 is a flowchart showing the flow of extraction processing by the extraction device. 10 is a flowchart showing the flow of extraction processing by the extraction device. FIG. 10 is a diagram illustrating an example of extracting a diagonal utility pole.

An example of an embodiment of the disclosed technology will be described below with reference to the drawings. Note that identical or equivalent components and parts in each drawing are given the same reference numerals. Also, the dimensional proportions in the drawings have been exaggerated for the sake of explanation and may differ from the actual proportions.

First, we will explain the methodology of an embodiment of the present disclosure.

In an embodiment of the present disclosure, cylindrical objects (e.g., utility poles) are extracted from the point cloud and the point cloud is identified.

Specifically, as shown in Figure 1, the point cloud is divided into groups at a certain size along an arbitrary axis. Circles are detected in a planar manner from the grouped point cloud, as shown in Figure 2, and if the number of circle-detected groups exceeds a threshold, the object is deemed to be a cylindrical object. Figure 1 shows a utility pole, which is a cylindrical object, being held by heavy machinery. Figure 2 shows the results of detecting circles from the point cloud representing the cylindrical object, viewed from above.

If the target cylindrical object is, for example, a utility pole under construction, its direction and angle will change depending on the situation, such as whether it is placed on the ground, held or suspended, or freestanding, making the axis used to detect the circle unknown. Therefore, by inputting the angle based on 3D LiDAR in advance, extraction is possible in any state. For example, as a first input method, the angle is determined visually and input before calculation. At this time, it is assumed that the degree of inclination relative to the x-axis, y-axis, and z-axis is known to some extent. Alternatively, as a second input method, an object made of a highly reflective material or structure, or an object with an unusual shape, is attached to the utility pole or heavy equipment, and used to calculate the axis.

In the first input method, the degree of tilt relative to the x-axis, y-axis, and z-axis is input, and circle detection is performed along the axis of the input direction. Thresholds are also set for the distance from the 3D LiDAR, the size of the extracted point cloud, etc., and the utility pole point cloud is extracted.

In the second input method described above, for example, two highly reflective objects are attached to the grip of a heavy machine. The attached objects are extracted, and the axis connecting the two objects is used as the axis of the utility pole, and circle detection is performed.

In addition, when the object is nearly upright, it is possible to extract a cylindrical object by setting a large threshold for the deviation of the center point of the extracted circle without inputting the axis angle.

Also, the x-axis is the east-west direction, the y-axis is the north-south direction, and the z-axis is the height direction.

In addition, when extracting cylindrical objects, the detection is performed within a certain threshold value relative to the z-axis. This is because if the object is tilted too much, it will be impossible to detect the circle along the axis.

Also, after detecting the circles, the centers of the circles can be calculated and the central axis can be calculated by connecting the centers.

In addition, well-known RANSAC (Random Sample Consensus) processing can be used to detect circles.

The above makes it possible to extract points representing tilted cylindrical objects from a point cloud.

In addition, in an embodiment of the present disclosure, the point cloud of the cylindrical object is extracted from the point cloud by repeatedly cutting out the point cloud of the cylindrical object, tracking it, calculating the amount of movement, and moving the detection area, and tracking the point cloud of the cylindrical object.

Specifically, as shown in Figure 3, a detection area is set around the point cloud of the cylindrical object extracted in the initial state, and the detection area is then moved according to the amount of movement of the cylindrical object, thereby minimizing the detection area and extracting the point cloud of the cylindrical object within the point cloud while reducing calculation time.

Figure 3 shows an example of the process for extracting a cylindrical object in this embodiment. Figure 3 shows an example in which the cylindrical object is a utility pole. The process is indicated by A1 to A7. Note that A1, A2, A3, A5, and A6 represent cylindrical objects in three-dimensional space along the x, y, and z axes.

A1 shows the cylinder extracted in the initial state and the detection area. In A2, a cylinder is extracted from the point cloud one second after the initial state, in the same detection area as A1. In A3, the cylinders in A1 and A2 are tracked and the amount of movement is calculated. In A4, a new detection area is calculated from the amount of movement calculated in A3. In A5, a cylinder is extracted from the point cloud two seconds after the initial state, in the detection area calculated in A4. In A6, the cylinders in A2 and A5 are tracked and the amount of movement is calculated. In A7, a new detection area is calculated from the amount of movement calculated in A6.

To set the detection area, calculate the central axis of the extracted cylindrical object, and set the distance from the central axis to x cm in the x-axis direction, y cm in the y-axis direction, and z cm in the extension direction from the top and bottom ends. The detection area can also be set manually. In this case, the detection area can be set using the "position of the bottom end of the utility pole," "utility pole length," and "extension direction."

The method of this embodiment described above makes it possible to extract inclined cylindrical object point clouds from a point cloud in real time.

<Configuration>
The configuration of this embodiment will be described below.

Figure 4 is a block diagram showing the hardware configuration of the extraction device 100.

As shown in FIG. 4, the extraction device 100 has a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, storage 14, an input unit 15, a display unit 16, and a communication interface (I/F) 17. Each component is connected to each other via a bus 19 so that they can communicate with each other.

The CPU 11 is a central processing unit that executes various programs and controls each component. That is, the CPU 11 reads programs from the ROM 12 or storage 14 and executes the programs using the RAM 13 as a working area. The CPU 11 controls the above components and performs various calculations in accordance with the programs stored in the ROM 12 or storage 14. In this embodiment, the ROM 12 or storage 14 stores an extraction program.

ROM 12 stores various programs and data. RAM 13 temporarily stores programs or data as a working area. Storage 14 is composed of a storage device such as an HDD (Hard Disk Drive) or SSD (Solid State Drive), and stores various programs including the operating system, and various data.

The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to make various inputs, including point cloud data input as external data from an external point cloud measuring device 90.

The display unit 16 is, for example, a liquid crystal display, and displays various information. The display unit 16 may also function as the input unit 15 by adopting a touch panel system.

The communication interface (I/F) 17 is an interface for communicating with other devices such as terminals. For this communication, a wired communication standard such as Ethernet (registered trademark) or FDDI, or a wireless communication standard such as 4G, 5G, or Wi-Fi (registered trademark) is used.

Next, we will explain each functional configuration of the extraction device 100. Figure 5 is a block diagram showing the configuration of the extraction device 100 of this embodiment. Each functional configuration is realized when the CPU 11 reads out an extraction program stored in the ROM 12 or storage 14, expands it in the RAM 13, and executes it.

As shown in FIG. 5, the extraction device 100 is composed of processing units including a reading unit 110, an acquisition unit 112, a conversion unit 114, an extraction unit 116, a setting unit 118, a tracking unit 120, a resetting unit 122, and an alarm issuing unit 124. Note that each processing unit may be configured as a device.

The reading unit 110 reads point cloud data as external data from an external point cloud measuring instrument 90.

The acquisition unit 112 acquires the inclination angle of the cylindrical objects included in the point cloud represented by the point cloud data read by the reading unit 110.

Specifically, the degree of tilt with respect to the x-axis, y-axis, and z-axis is input via the input unit 15, and the tilt angle of the cylindrical object included in the point cloud is obtained. Alternatively, two highly reflective objects are attached to the cylindrical object, and the two attached objects are extracted from the point cloud data, and the tilt of the axis connecting the two objects is obtained as the tilt angle of the cylindrical object included in the point cloud.

The conversion unit 114 uses the acquired angle to convert the coordinates of the point cloud so that the angle of inclination of the cylindrical object becomes the reference angle.

The extraction unit 116 extracts cylindrical objects from the point cloud based on the coordinates of the transformed point cloud.

As shown in Figure 6, the extraction unit 116 is configured to include a division unit 132, a circle information calculation unit 134, and a cylindrical object determination unit 136. The configuration of each component of the extraction unit 116 will be described later.

The setting unit 118 sets a detection area around the cylindrical object based on the central axis of the cylindrical object extracted by the extraction unit 116, using pre-set conditions.

The tracking unit 120 tracks the cylindrical object at the next time from the point cloud within the detection area.

The resetting unit 122 calculates the amount of movement for the tracked cylindrical object and sets the detection area so that the detection area is moved by the amount of movement.

Tracking by the tracking unit 120 and calculation of the amount of movement and setting of the detection area by the resetting unit 122 are repeated.

The alarm unit 124 calculates the distance between the extracted cylindrical object and the point cloud surrounding the cylindrical object from the input point cloud, and issues an alarm if the distance is below a threshold value.

Next, we will explain each part of the extraction unit 116.

The dividing unit 132 divides the point cloud into N number of height groups in the height direction (z-axis direction) to generate height groups. N is an arbitrary number. Note that point clouds whose Z coordinates are equal to or less than an arbitrary height may be excluded from each of the point clouds included in the point cloud data read by the reading unit 110.

For each height group, the circle information calculation unit 134 projects the point cloud of that height group onto an XY plane including the x-axis and y-axis, and performs circle detection on a group-by-group basis. Circle detection uses well-known methods such as RANSAC (Random Sample Consensus) processing. Furthermore, if a circle is detected from the group, the circle information calculation unit 134 calculates the center coordinates and the circle radius.

The cylindrical object determination unit 136 determines whether the circle detected for each height group satisfies a predetermined condition. The predetermined condition may be, for example, at least one of the following conditions 1 to 3, or a combination of two or more of them. This determines whether the detected circle is a cylindrical object. Note that conditions 1 to 3 may be used alone or in combination.

1. Is the number of groups in which circles are detected equal to or greater than a given percentage of the N divided groups? 2. Are the detected center coordinates (x, y) within a given error range? 3. Are the detected circle radii within a given error range?

<Effect>
Next, the operation of the extraction device 100 will be described.

Figure 7 is a flowchart showing the flow of the extraction process performed by the extraction device 100. The extraction process is performed by the CPU 11 reading the extraction program from the ROM 12 or storage 14, expanding it into the RAM 13, and executing it. The extraction process is performed by the CPU 11 performing processing as each part of the extraction device 100. Note that the extraction process is an example of an extraction method.

In step S100, the CPU 11, as the reading unit 110, reads point cloud data as external data from the external point cloud measuring device 90.

In step S102, the CPU 11 deletes from each of the point clouds contained in the loaded point cloud data any point clouds whose distance from the measurement point of the point cloud measuring instrument 90 is greater than an arbitrary distance.

In step S103, the CPU 11, as the acquisition unit 112, acquires the angle a of the z-axis direction of the target cylindrical object relative to the x-y plane. For example, as shown in Figure 8, with the measurement point of the point cloud measuring instrument 90 as the reference, the x-axis is east-west, the y-axis is north-south, and the z-axis is altitude.

Here, angle a is input by the user. Alternatively, angle a can be estimated by performing circle detection from point cloud data for all axes, or by performing circle detection at each height along the z axis and estimating angle a that minimizes the deviation of the circle's center point.

In step S104, the CPU 11, as the conversion unit 114, converts the coordinates of the point cloud by correcting the point cloud so that the axis of the cylinder is the z-axis based on the angle a acquired in step S103. Specifically, the coordinates of the point cloud are converted so that the point cloud data is rotated around the axis by the amount of angle a.

In step S105, the CPU 11, as the division unit 132, divides and groups the point cloud data at any height in the Z-axis direction. Specifically, the point cloud is divided into N number of height groups in the height direction (z-axis direction) to generate height groups.

In step S106, the CPU 11, as the circle information calculation unit 134, projects the point cloud onto the xy plane for each group and performs circle detection. Specifically, for each height group, the point cloud for that height group is projected onto the xy plane and circle detection is performed on a group-by-group basis. For each height group, if a circle is detected from that group, the coordinates of the center point are calculated.

In step S107, the CPU 11, as the cylindrical object determination unit 136, determines whether the number of height groups for which circles were detected along the Z axis is equal to or greater than a threshold value and whether the deviation of the center points of the circles between the height groups for which circles were detected is within a threshold value. If the number of height groups for which circles were detected along the Z axis is less than the threshold value or if the deviation of the center points of the circles between the height groups for which circles were detected is greater than the threshold value, it determines that there is no cylindrical object candidate point group and returns to step S100 above.

On the other hand, if the number of height groups in which circles were detected along the Z axis is greater than or equal to the threshold value and the deviation of the center points of the circles between the height groups in which circles were detected is within the threshold value, proceed to step S108.

In step S108, the CPU 11 designates the point group of the height group in which a circle can be detected along the Z axis as the cylindrical object candidate cluster i.

Instead of the processes of steps S100 to S108 above, the processes of steps S100 to S216 shown in Figure 9 may be performed. Also, both the processes of steps S100 to S108 above and the processes of steps S100 to S216 may be performed.

In step S100 shown in Figure 9, the CPU 11, as the reading unit 110, reads point cloud data as external data from the external point cloud measuring device 90.

In step S102, the CPU 11 deletes from each of the point clouds contained in the loaded point cloud data any point clouds whose distance from the measurement point of the point cloud measuring instrument 90 is greater than an arbitrary distance.

In step S204, the CPU 11, functioning as the acquisition unit 112, obtains feature point cloud clusters from the point cloud data, connects two feature point cloud clusters, and calculates an axis. Here, a feature point cloud cluster is a cluster of point clouds of objects with characteristics such as high reflection intensity or a unique shape. Objects with characteristics such as high reflection intensity or a unique shape are preferably attached to or located close to utility poles, such as scaffolding bolts or heavy equipment grips.

In step S206, the CPU 11, as the conversion unit 114, converts the coordinates of the point cloud by rotating and correcting the point cloud data based on the axis calculated in step S204 so that the axis of the cylindrical object becomes the z-axis.

In step S208, the CPU 11, as the division unit 132, divides and groups the point cloud data at any height in the Z-axis direction. The point cloud is divided into N number of height groups in the height direction (z-axis direction) to generate height groups.

In step S210, the CPU 11, as the circle information calculation unit 134, projects the point cloud onto the xy plane for each group and performs circle detection. Specifically, for each height group, the point cloud for that height group is projected onto the XY plane and circle detection is performed on a group-by-group basis. For each height group, if a circle is detected from that group, the coordinates of the center point are calculated.

In step S212, the CPU 11, as the cylindrical object determination unit 136, determines whether the number of height groups for which circles were detected along the Z axis is equal to or greater than a threshold value and whether the deviation of the center points of the circles between the height groups for which circles were detected is within a threshold value. If the number of height groups for which circles were detected along the Z axis is less than the threshold value or if the deviation of the center points of the circles between the height groups for which circles were detected is greater than the threshold value, it determines that there is no cylindrical object candidate point group and returns to step S100 above.

On the other hand, if the number of height groups in which circles were detected along the Z axis is greater than or equal to the threshold value and the deviation of the center points of the circles between the height groups in which circles were detected is within the threshold value, proceed to step S216.

In step S216, the CPU 11 designates the point group of the height group in which a circle can be detected along the Z axis as the cylindrical object candidate cluster i.

Then, in step S110 shown in Figure 10, the CPU 11, as the setting unit 118, sets a detection area around the point cloud of cylindrical object candidate cluster i. For example, if the central axis of the cylindrical object is the z' axis, the detection area is set to a [cm] in the x' axis direction, b [cm] in the y' axis direction, and c [cm] in the z' axis direction.

In step S112, the CPU 11, as the reading unit 110, reads the point cloud data after t seconds from the external point cloud measuring device 90, and extracts the point cloud after t seconds using the same detection area as the initial state.

Then, for each cylindrical object candidate cluster i (i≦total number of cylindrical object candidate clusters), the processing of steps S114 to S120 described below is repeated.

In step S114, the CPU 11, as the tracking unit 120, tracks the point cloud of the cylindrical object candidate cluster and calculates the movement amount D.

In step S116, the CPU 11 determines whether the amount of movement D is greater than or equal to the lower threshold and less than or equal to the upper threshold. If the amount of movement D is less than the lower threshold, it is determined that the cylindrical object represented by the cylindrical object candidate cluster is not moving and therefore does not need to be tracked, and the process proceeds to step S118. If the amount of movement D is greater than the upper threshold, it is determined that the cylindrical object candidate cluster represents an object different from the target cylindrical object and therefore does not need to be tracked, and the process proceeds to step S118. On the other hand, if the amount of movement D is greater than or equal to the lower threshold and less than or equal to the upper threshold, the process proceeds to step S120.

In step S118, the CPU 11 does not select the cylindrical object candidate cluster i as the cylindrical object cluster to be tracked.

In step S120, the CPU 11 sets the cylindrical object candidate cluster i to the cylindrical object cluster i' to be tracked.

As described above, the cylindrical object to be tracked is extracted from the extracted cylindrical object candidate cluster.

Then, as shown in Figure 11, the processing of steps S122 to S136 described below is repeated every Δt seconds starting from the point cloud data after t' seconds.

In step S122, the CPU 11, as the resetting unit 122, sets the detection area so as to move the previous detection area by the movement amount D.

In step S124, the CPU 11, as the reading unit 110, reads the point cloud data after t' seconds from the external point cloud measuring device 90, and extracts the point cloud after t' seconds from the detection area set in step S122 above.

In step S126, the CPU 11, as the tracking unit 120, tracks the point cloud of the cylindrical object cluster i' to be tracked and calculates the movement amount D'.

In step S128, the CPU 11 determines whether the calculated movement amount D' is equal to or less than the threshold value. If the calculated movement amount D' is greater than the threshold value, the process proceeds to step S130. On the other hand, if the calculated movement amount D' is equal to or less than the threshold value, the process proceeds to step S134.

In step S130, the CPU 11 expands the detection area and extracts the cylindrical object again.

In step S132, the CPU 11 determines whether or not a cylindrical object was extracted in step S130. If a cylindrical object was not extracted in step S130, the process returns to step S100. On the other hand, if a cylindrical object was extracted in step S130, the process proceeds to step S134.

In step S134, the CPU 11, as the alarm unit 124, determines whether the distance to the surrounding point cloud is less than or equal to the threshold. If the distance to the surrounding point cloud is less than or equal to the threshold, the process proceeds to step S136.

On the other hand, if the distance to the surrounding point cloud is greater than the threshold, set t' = t' + Δt and return to step S122 above.

In step S136, the CPU 11, as the alarm issuing unit 124, issues an alarm and terminates the extraction process.

In this way, the distance between the tracked cylindrical cluster and the surrounding point cloud is calculated, and an alarm is issued if they come close. If an error is detected, such as the movement of the tracked cylindrical cluster being greater than the threshold, the process starts over.

<Example>
An example of extracting a diagonal utility pole as a cylindrical object, as shown in Fig. 12, will be described. First, the setting parameters for extracting the diagonal utility pole will be described.

When detecting circles, to enable clustering even when the density is low when viewed from above, the division height of the point cloud is narrowed to 0.4 m, and the division number N in the division unit 132 is set. Furthermore, because the division height is narrowed, the threshold used by the cylindrical object determination unit 136 to determine whether each detected center coordinate is within the error range is set to 0.6 m. This is because if this threshold were set to around 0.7 m to 0.8 m, circles detected from a worker near a utility pole being erected would be grouped with circles in the pole being erected.

Even if the initial tilt angle of the cylindrical object is not input, it is possible to extract the cylindrical object with an inclination of up to 45° using the above parameters. In this case, the conversion by the conversion unit 114 is also omitted. The dotted bubble in Figure 12 shows the results of circle detection for each height group of the point cloud of the diagonal utility pole.

In this embodiment, it is possible to detect in real time when a utility pole being erected is about to come into contact with a cable or wall, and to issue an alarm.

The most effective method for tracking is to combine the techniques "SHOT" and "ICP," but there is also a method that combines "Spiniimage" and "ICP." "Cropbox" creates a box (detection area) with specified XYZ widths, then adds translation and rotation to crop at any position.

The amount of movement after tracking must be within a distance of 100 cm and an inclination angle of 45°. Here, "SHOT", "ICP", "Spinimage", and "Cropbox" refer to libraries included in PCL (Point Cloud Library).

As explained above, the extraction device 100 of this embodiment obtains the tilt angle of cylinders contained in an input point cloud, uses the obtained angle to convert the coordinates of the point cloud so that the tilt angle of the cylinders becomes a reference angle, and extracts cylinders from the point cloud based on the converted coordinates of the point cloud. This makes it possible to extract cylinders with high accuracy even when the cylinders are tilted.

In addition, based on the central axis of the extracted cylindrical object, a detection area is set around the cylindrical object according to pre-set conditions, and the cylindrical object is tracked at the next time from the point cloud within the detection area. This makes it possible to track cylindrical objects in real time.

The system also calculates the amount of movement of the tracked cylindrical object, sets the detection area so that it moves by the amount of movement, and repeats tracking, calculating the amount of movement, and setting the detection area. This makes it possible to track cylindrical objects in real time.

The acquired point cloud data is input, and the tilt angle of the target cylindrical object is input, or the tilt angle of the target cylindrical object is detected based on the feature point cloud cluster, and the coordinates of the point cloud data are converted using this angle so that the central axis of the cylinder is aligned with the z-axis, and the cylinder is extracted from the point cloud data after coordinate conversion. This makes it possible to extract non-upright cylindrical objects from the acquired point cloud data in real time.

In the above embodiment, an example was described in which the point cloud was divided in the height direction, circle detection was performed for each height group, and cylindrical objects were extracted based on the results of the circle detection, but this is not limited to this. Cylinders may also be extracted from the results of converting the coordinates of the point cloud using other methods. Also, while an example was described in which the cylindrical objects were utility poles, this is not limited to this. Cylinders other than utility poles may also be targeted.

Furthermore, the extraction process executed by the CPU in the above embodiment by loading software (programs) may be executed by various processors other than the CPU. Examples of processors in this case include dedicated electrical circuits, such as PLDs (Programmable Logic Devices) whose circuit configuration can be changed after manufacture, such as FPGAs (Field-Programmable Gate Arrays), GPUs (Graphics Processing Units), and ASICs (Application Specific Integrated Circuits), which are processors with circuit configurations designed specifically to execute specific processes. Furthermore, the extraction process may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (e.g., multiple FPGAs, or a combination of a CPU and an FPGA, etc.). Furthermore, the hardware structure of these various processors is, more specifically, an electric circuit that combines circuit elements such as semiconductor elements.

In addition, while the above embodiment describes a configuration in which the extraction program is pre-stored (installed) in storage 14, this is not limited to this. The program may be provided in a form stored on a non-transitory storage medium such as a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), or a USB (Universal Serial Bus) memory. The program may also be downloaded from an external device via a network.

As mentioned above, the extraction device described in the embodiments of the present disclosure 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 following notes are further disclosed regarding the above embodiments.

(Additional note 1)
Memory and
at least one processor coupled to said memory;
Including,
The processor:
For the input point cloud, obtain the tilt angle of the cylindrical object included in the point cloud;
Using the acquired angle, convert the coordinates of the point cloud so that the angle of inclination of the cylindrical object becomes a reference angle;
an extraction device configured to extract a cylindrical object from the point cloud based on the transformed coordinates of the point cloud.

(Additional note 2)
A non-transitory storage medium storing a program executable by a computer to perform an extraction process,
The extraction process includes:
For the input point cloud, obtain the tilt angle of the cylindrical object included in the point cloud;
Using the acquired angle, convert the coordinates of the point cloud so that the angle of inclination of the cylindrical object becomes a reference angle;
extracting cylinders from the point cloud based on the transformed coordinates of the point cloud.

11 CPU
14 Storage 15 Input unit 90 Point cloud measuring instrument 100 Extraction device 110 Reading unit 112 Acquisition unit 114 Conversion unit 116 Extraction unit 118 Setting unit 120 Tracking unit 122 Resetting unit 124 Notification unit 132 Division unit 134 Circle information calculation unit 136 Cylindrical object determination unit

Claims (6)

an acquisition unit that acquires the inclination angle of a cylindrical object included in an input point cloud;
a conversion unit that converts the coordinates of the point cloud using the acquired angle so that the angle of inclination of the cylindrical object becomes a reference angle;
an extraction unit that extracts a cylindrical object from the point cloud based on the coordinates of the transformed point cloud;
Including,
The extraction unit
a division unit that divides the point cloud in a height direction and generates height groups;
a circle information calculation unit that projects the point cloud of each height group onto an XY plane including an x-axis and a y-axis and performs circle detection for each group of the divided point clouds;
a cylindrical object determination unit that extracts, from among the height groups, a point cloud of a height group in which a circle can be detected as a cylindrical object when the number of height groups in which a circle can be detected is equal to or greater than a threshold value and the deviation of the center points of the circles between the height groups in which a circle can be detected is within a threshold value;
An extraction device comprising : a setting unit that sets a detection area around the cylindrical object according to preset conditions based on the central axis of the cylindrical object extracted by the extraction unit;
The extraction device according to claim 1 , further comprising: a tracking unit that tracks the cylindrical object at a next time from the point cloud within the detection area. a resetting unit that calculates a movement amount of the tracked cylindrical object and resets the detection area so as to move the detection area by the movement amount;
3. The extraction device according to claim 2, wherein the tracking by the tracking unit and the calculation of the amount of movement and the setting of the detection area by the resetting unit are repeated. The extraction device of claim 3 further includes an alarm unit that calculates the distance between the extracted cylindrical object and a point cloud surrounding the cylindrical object from the input point cloud, and issues an alarm if the distance is equal to or less than a threshold value. an acquisition unit acquires, for the input point cloud, an inclination angle of a cylindrical object included in the point cloud;
a conversion unit converts the coordinates of the point cloud using the acquired angle so that the angle of inclination of the cylindrical object becomes a reference angle;
An extraction unit extracts a cylindrical object from the point cloud based on the coordinates of the transformed point cloud.
This includes:
In extracting the cylindrical object,
Divide the point cloud in the height direction and generate height groups.
For each of the divided point clouds, the point cloud of the height group is projected onto an XY plane including the x-axis and the y-axis, and circle detection is performed for each group.
If the number of height groups in which circles can be detected is equal to or greater than the threshold, and the deviation of the center points of the circles between the height groups in which circles can be detected is within the threshold, the point cloud of the height group in which circles can be detected is extracted as a cylindrical object.
Extraction method. For the input point cloud, obtain the tilt angle of the cylindrical object included in the point cloud;
Using the acquired angle, convert the coordinates of the point cloud so that the angle of inclination of the cylindrical object becomes a reference angle;
Extracting cylindrical objects from the point cloud based on the coordinates of the transformed point cloud.
This includes:
In extracting the cylindrical object,
Divide the point cloud in the height direction and generate height groups.
For each of the divided point clouds, the point cloud of the height group is projected onto an XY plane including the x-axis and the y-axis, and circle detection is performed for each group.
If the number of height groups in which circles can be detected is equal to or greater than the threshold, and the deviation of the center points of the circles between the height groups in which circles can be detected is within the threshold, the point cloud of the height group in which circles can be detected is extracted as a cylindrical object.
An extraction program that causes a computer to perform the processing.