JP5577312B2 - Data analysis apparatus, data analysis method, and program - Google Patents

Description

  The present invention relates to a data analysis apparatus, a data analysis method, and a program for detecting a cylindrical feature based on point cloud data representing a three-dimensional shape of the feature surface.

  Patent Document 1 discloses a technique for acquiring three-dimensional point cloud data representing the shape of a feature using a laser scanner. For example, in a mobile mapping system, a laser scanner is mounted on the top of an automobile and irradiates the surrounding area with a laser. The optical axis of the laser is scanned in the vertical and horizontal directions by changing the elevation angle and azimuth angle, and laser pulses are emitted every minute angle within the scanning range. The distance is measured based on the time from the laser emission to the reception of the reflected light, and at that time, the laser emission direction, time, and the position / posture of the vehicle body are measured. The measurement of the position and orientation of the vehicle body is performed using, for example, a GPS / IMU (Global Positioning System: Inertial Measurement Unit). From the measurement data obtained in this way, point cloud data representing the three-dimensional coordinates of the point reflecting the laser pulse is obtained.

  Simultaneously with the acquisition of the point cloud data, a video is taken using the camera. The image can be used when the user designates a measurement target part in data analysis.

JP 2009-204615 A

  Conventionally, it has been necessary to manually read columnar features such as utility poles and other features based on point cloud data, and manually using 3D CAD editing tools and the like. Extraction work was being done.

  An object of the present invention is to provide a data analysis apparatus, a data analysis method, and a program for automatically detecting a cylindrical feature based on point cloud data representing a three-dimensional shape of the feature surface.

  A data analysis apparatus according to the present invention detects a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in a target space, and the target space is a vertical plane. Point space is distributed to a partial space setting means that divides and sets a plurality of columnar partial spaces, and a height that is equal to or higher than a threshold value that is preset in accordance with the height of the target feature in the partial spaces. A subspace selection means for selecting an object as a target subspace, and a point cloud concentration line in which the point group projected on the horizontal plane of the target subspace in the horizontal plane gathers in the vicinity of a predetermined reference or more Searching for minutes, and finding the remaining point cloud excluding the point cloud corresponding to the wall surface, and a wall surface searching means for estimating the point cloud concentration line segment as a position in the horizontal plane of the wall surface of the feature, In the distribution of the residual points in the horizontal plane Zui and has a cylindrical surface searching means for searching the position of the cylindrical surface of the object feature.

  Another data analysis apparatus according to the present invention includes block space setting means for dividing the partial space of interest and setting a plurality of block spaces stacked vertically, and the wall surface search means includes a wall surface for each block space. The cylindrical surface search means searches for the position of the cylindrical surface using the point group in the block space where the wall surface position is not detected as the remaining point group.

  In another data analysis apparatus according to the present invention, the wall surface search means is the point cloud concentration line segment in a space adjacent to the block space, and the point cloud concentration line segment detected in the block space. The connected line segment connected to is searched, and when the connected line segment exists, the point group concentrated line segment in the block space is estimated as the wall surface position.

  In another data analysis apparatus according to the present invention, when the position of the cylindrical surface is detected in the block space, the data analysis device is positioned at a height corresponding to the block space, and a horizontal section of the target feature A re-search space setting means for setting a columnar re-search space having a horizontal cross section of a size including the center axis of the cylindrical surface, and the point group in the re-search space in the horizontal plane Cylindrical surface re-searching means for obtaining a position of the cylindrical surface based on the distribution and extracting a cylindrical surface point group existing on the surface of the cylindrical surface from the point group.

  The data analysis apparatus according to another aspect of the present invention further includes a cylindrical shape determining unit that obtains a three-dimensional shape of the cylindrical surface based on the cylindrical surface point group extracted corresponding to the block space.

  A data analysis method according to the present invention is a method for detecting a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in the target space, wherein the target space is a vertical plane. A point space is distributed to a partial space setting step for dividing and setting a plurality of columnar partial spaces, and to a height equal to or higher than a threshold value set in advance according to the height of the target feature in the partial spaces. A subspace selection step for selecting an object as a target subspace, and a point cloud concentration line in which the point cloud projected onto the horizontal plane of the target subspace is gathered in the vicinity of a predetermined reference or more in the horizontal plane Searching for a minute, a wall surface searching step for estimating the point cloud concentration line segment as a position in the horizontal plane of the wall surface of the feature, and obtaining a residual point cloud excluding the point cloud corresponding to the wall surface, The remainder in the horizontal plane Based on the distribution of the group have, and cylindrical surface searching step of searching for a position of the cylindrical surface of the object feature.

  A program according to the present invention is a program for causing a computer to perform processing for detecting a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in a target space, The computer includes a partial space setting unit that divides the target space along a vertical plane and sets a plurality of columnar partial spaces, a threshold value that is set in advance according to the height of the target feature in the partial spaces A subspace selecting means for selecting as a target subspace a point cloud distributed to a height of the reference, a reference set in advance by the point group projected onto the horizontal plane among the line segments in the horizontal plane of the target subspace Search for point cloud concentration line segments gathered in the vicinity, wall surface search means for estimating the point cloud concentration line segments as the position of the wall surface of the feature in the horizontal plane, and the point cloud corresponding to the wall surface Calculated excluding residual point group, based on the distribution of the residual point group in the horizontal plane, the object cylindrical surface searching means for searching the position of the cylindrical surface of the feature, function as.

  According to the present invention, a cylindrical feature can be automatically detected based on point cloud data representing the three-dimensional shape of the feature surface.

It is a block diagram which shows the structure of the outline of the utility pole extraction system which concerns on embodiment of this invention. It is a general | schematic flowchart of the utility pole extraction process by the utility pole extraction system which concerns on embodiment of this invention. It is a general | schematic flowchart of a telephone pole search process. It is a general | schematic flowchart of a telephone pole search process. It is a schematic flowchart of a wall surface determination process. It is a schematic flowchart of a wall surface determination process. It is a schematic diagram explaining the process which detects the wall surface in a block space. It is a general | schematic flowchart of an electric pole extraction process. It is a schematic diagram which shows an example of a re-search space. It is a general | schematic flowchart of a telephone pole shape registration process.

  Hereinafter, an electric pole extraction system 2 according to an embodiment of the present invention (hereinafter referred to as an embodiment) will be described with reference to the drawings. This system is a data analysis device that detects a utility pole, which is a cylindrical feature, based on point cloud data representing the three-dimensional shape of the feature surface. This system extracts a power pole by separating a cylindrical surface having a radius corresponding to a power pole on a feature surface substantially perpendicular to the ground surface and a wall surface that is flat compared to the cylindrical surface. The point cloud data is acquired by, for example, a laser scanner mounted on a vehicle traveling on the ground like the above-described mobile mapping system. In addition, measurement may be performed by installing a laser scanner on the ground. In order for the point cloud data to represent the three-dimensional shape of the feature surface, it is necessary to perform laser scanning at a point density corresponding to the scale of the irregularities on the feature surface. In this respect, laser scanning performed from the ground such as a road using a vehicle or a tripod can realize, for example, a scanning density that can capture the shape of a cylindrical surface of a utility pole and a wall surface of a building that are built in the vicinity of the road. .

  FIG. 1 is a block diagram showing a schematic configuration of the utility pole extraction system 2. The system includes an arithmetic processing device 4, a storage device 6, an input device 8, and an output device 10. As the arithmetic processing device 4, it is possible to make dedicated hardware for performing various arithmetic processing of this system, but in this embodiment, the arithmetic processing device 4 uses a computer and a program executed on the computer. Built.

  A CPU (Central Processing Unit) of the computer constitutes the arithmetic processing unit 4, and a partial space setting unit 20, a partial space selection unit 22, a block space setting unit 24, a wall surface search unit 26, a cylindrical surface search unit 28, and a re-described later. It functions as a search space setting means 30, a cylindrical surface re-search means 32, a cylindrical shape determination means 34, and a cylindrical feature registration means 36.

  The storage device 6 is composed of a hard disk or the like built in the computer. The storage device 6 replaces the arithmetic processing unit 4 with the partial space setting means 20, the partial space selection means 22, the block space setting means 24, the wall surface searching means 26, the cylindrical surface searching means 28, the re-search space setting means 30, and the cylindrical surface re-searching means. 32, the program for functioning as the cylindrical shape determining means 34 and the cylindrical feature registering means 36, other programs, and various data necessary for the processing of this system are stored. For example, the storage device 6 stores point cloud data of the analysis target space as the processing target data. For example, a road and its vicinity area are set as the analysis target space. In-vehicle GPS / IMU measurement data obtained by measuring the position and orientation of a vehicle equipped with a laser scanner is also stored in the storage device 6. The storage device 6 stores in advance information related to the thickness (radius) and height of the utility pole to be detected.

  The input device 8 is a keyboard, a mouse, or the like, and is used for a user to operate the system.

  The output device 10 is a display, a printer, or the like, and is used for indicating the position of the utility pole in the target space obtained by the present system to the user by screen display, printing, or the like.

  FIG. 2 is a schematic flowchart of the utility pole extraction process performed by the utility pole extraction system 2. Each means of the arithmetic processing unit 4 will be described with reference to FIG.

  The partial space setting means 20 sets a plurality of columnar partial spaces obtained by dividing the target space along the vertical plane as initial unit spaces for wall surface extraction processing (S40). In the present embodiment, the partial space setting means 20 divides the target space into a two-dimensional orthogonal lattice (mesh) shape in a horizontal plane (the XY plane in the XYZ orthogonal coordinate system) and forms a quadrangular prism-shaped partial space. Is generated. For example, the horizontal cross section of the partial space can be a square having a width W and a depth D of 50 cm. The height H of the partial space can be made to coincide with the height of the target space (dimension in the Z-axis direction), and the height is set larger than the height of the utility pole for detection purposes. In the program constituting the partial space setting means 20, the width W, the depth D, and the height H are parameterized and can be changed by the user using the input device 8, for example.

  The partial space selection means 22 selects a partial space including a point group distributed up to a height equal to or higher than a preset threshold as a target partial space. It is preferable to set the threshold value so that a partial space including the utility pole is reliably selected on the basis of the height of the utility pole to be detected, while a partial space not including the utility pole is selected as much as possible.

  Specifically, the partial space setting unit 20 sets a plurality of partial spaces in a predetermined order, and the partial space selection unit 22 determines whether or not the partial space is the target partial space. As described above, point cloud data is acquired by laser irradiation from a laser scanner located on a road or the like, and features such as utility poles and buildings visible from the laser scanner are captured on a three-dimensional shape based on the point cloud data. Therefore, for example, the partial space setting means 20 determines a reference plane based on the portion of the plane corresponding to the road or the like in the three-dimensional shape, and horizontally in the road width along the road. A line is set (S42). Then, the starting point for setting the partial space is set in order along the center line, and the partial space is set in the order closer to the starting point in both directions along the line orthogonal to the center line (S44). If the height of the point group in the partial space is equal to or greater than the threshold value, the partial space selection unit 22 determines the partial space as the target partial space (S46).

The block space setting means 24 divides the target partial space by a horizontal plane and sets a plurality of block spaces that are stacked vertically (S48). For example, it is possible to set the width W, comparable to the depth D of the height H _B subspace of the block space.

  For each block space, the wall surface searching means 26 searches for a line segment (point cloud concentration line segment) where the point group projected on the horizontal plane in the horizontal plane gathers in the vicinity of a predetermined reference or more, and the line segment Is estimated as the position of the wall surface in the horizontal plane (horizontal wall surface position) (S50). By searching for the position of the wall surface in the horizontal plane for each block space, it is possible to separate a plurality of wall surfaces with different positions and orientations that may exist at different heights in the subspace that is elongated in the vertical direction. Here, the wall surface search means 26 may be configured to determine that a wall surface exists at the position of the point cloud concentration line segment by detecting the point cloud concentration line segment in a certain block space. In the embodiment, a point cloud concentration line segment (connected line segment) connected to the point cloud concentrated line segment is further searched in a space adjacent to the block space, and when there is a connected line segment, a point in the block space is present. The group concentration line segment is configured to determine the wall surface position. This has the effect of improving the reliability of the detected wall surface, particularly when the horizontal cross-section size (width W and depth D) of the block space (or partial space) is set to be relatively small. Less.

The cylindrical surface search means 28 obtains the remaining point group (residual point group) excluding the point group existing in the block space corresponding to the wall surface, and based on the distribution of the residual point group in the horizontal plane, The position of the cylindrical surface is searched (S52). Specifically, the cylindrical surface search means 28 calculates the center position C _α and the radius R _α of the cylindrical surface in the XY plane.

Here, it is assumed that there is only a part of the cylindrical surface in the block space, and the cylindrical surface (center coordinate C _α and radius R determined by the cylindrical surface search means 28 based on the remaining point group extracted for each block space. The reliability and accuracy of _α ) can be low.

Therefore, when the position of the cylindrical surface is detected by the cylindrical surface search means 28 in the block space, a re-search process is performed to recalculate the position of the cylindrical surface based on the position (S54). The re-search space setting means 30 has a size including a circle whose horizontal cross section is the horizontal cross section of the utility pole as a space for performing re-search (re-search space), the center of which is the cylindrical surface obtained by the cylindrical surface search means 28 of coincides with the center C _alpha, also the position in the vertical direction (the lower end and the upper end position) is cylindrical surface to set the space is block space and essentially the same detection. In this embodiment, the re-search space is a rectangular parallelepiped with a horizontal cross section. For example, the horizontal cross section can be set larger than that of the block space, and in this embodiment, the width and the depth are each set to be about 10 cm larger.

The cylindrical surface re-search unit 32 obtains the position (center coordinate C _β and radius R _β ) of the cylindrical surface based on the distribution of the point group in the re-search space in the horizontal plane, and among the point group in the re-search space, A point cloud (cylindrical surface point cloud) existing on the surface of the surface is extracted. Here, the cylindrical surface point group, for example, a radius _{[R β, R β + ΔR} ] can extract those present in the region of a ring shape that is in scope. ΔR is an assumed point group width of the surface of the utility pole, and is set to a size according to the measurement accuracy of the unevenness of the surface of the utility pole and the position of the point group, for example. Further, many utility poles have a taper on the side surface, and become thinner at higher positions. In consideration of this, ΔR can be set. Note that the re-search is preferably performed using the remaining point group extracted for each block space and existing in the re-search space.

  The cylindrical surface search by the cylindrical surface search means 28 and the cylindrical surface re-search means 32 described above was performed for each height corresponding to the block space set as the target partial space. This search can be performed in order in the direction from the lower stage to the upper stage of a plurality of block spaces set in the target subspace, for example.

The cylindrical shape determining unit 34 obtains the three-dimensional shape of the cylindrical surface based on the cylindrical surface point group extracted by the re-search for each height corresponding to the block space set as the target partial space (S56). Specifically, the cylindrical shape determining unit 34 projects the cylindrical surface point group obtained corresponding to the target partial space onto a horizontal plane, and obtains the center position C and the radius R of the cylindrical surface based on the distribution. Further, the cylindrical shape determining unit 34 selects a point group existing on the cylindrical surface defined by the center C and the radius R from the cylindrical surface point group obtained by the re-search, and the cylindrical surface point group by the cylindrical surface re-search unit 32. Extract in the same way as Then, the maximum height A _H and the minimum altitude A _L of the point group. Three-dimensional shape of a cylindrical surface is identified by the central C, the radius R, the maximum altitude A _H and the minimum altitude A _L, the shape is registered in the storage device 6 by cylindrical feature registration unit 36. In this embodiment, the cylindrical shape determining unit 34 generates a wire frame model representing the three-dimensional shape of the cylindrical surface, and the cylindrical feature registering unit 36 stores the lines constituting the model in the storage device 6.

  Hereinafter, the process performed by the utility pole extraction system 2 will be described in more detail. The utility pole extraction system 2 initializes search information at the start of processing. By this initialization, for example, the storage contents of various work areas provided in the storage device 6 and RAM (Random Access Memory) are erased. Various parameters used for processing such as the size of the partial space (width W, depth D, height H) are set by user input, or a parameter set prepared in advance in the storage device 6 is stored in a work area such as a RAM. Is read.

  For example, when the partial space setting means 20 divides the target space into a mesh shape along the X-axis direction and the Y-axis direction, if the width W and the depth D of the partial space are set as parameters, these are the intervals between the meshes. (Partial space size), and the number of subspaces arranged in the X and Y directions is determined according to the size of the target space. The range of the index indicating the position of the partial space in each of the X and Y directions is determined based on the number of arrays. The partial space setting means 20 changes the index representing the two-dimensional mesh-like arrangement of the partial space in a predetermined order, and is defined by a position (coordinate range in each direction of X and Y) corresponding to the set index. Set the space.

  Here, when the point cloud data is acquired by performing laser irradiation from the vehicle traveling on the road, as described above, the reference plane is based on the portion of the plane corresponding to the road or the like in the three-dimensional shape represented by the point cloud data. Is set, and the center line of the three-dimensional shape is set along the road. In order to simplify the explanation, the center line is assumed to be the X axis (ie, Y = 0), the index in the X direction increases in the direction in which the vehicle travels, and the Y direction corresponding to the road width direction leaves the center line. It is assumed that the absolute value of the index increases according to For example, the index in the Y direction is set to 0 at the position of the center line, set to 1, 2, 3,... To the left with respect to the traveling direction, and set to −1, −2, −3,.・ Can be set. In this case, for example, the subspace setting unit 20 sequentially increments the index in the X direction, and changes the index in the Y direction in both the positive direction and the negative direction of the Y axis at each value of the index in the X direction. . That is, the partial space setting means 20 sets the partial space by increasing the index in the Y direction from 0 in order of a certain index value in the X direction, and sets the partial space by decreasing in order from -1. To do.

  For each partial space (primary partial space) set in this way, a utility pole search process described later is performed. Further, for each subspace, a point of the maximum elevation in the point cloud is searched, a subspace (secondary subspace) centered in the horizontal plane is set, and the primary subspace and A similar utility pole search process is performed.

3 and 4 are schematic flowcharts of the utility pole search process. The partial space selection means 22 takes in the point group included in the partial spaces that are sequentially set, for example, from the storage device 6 such as a hard disk into a work area such as a RAM (S60). There are a number of point clouds equal to or greater than the predetermined threshold Th1 in the point cloud captured in the partial space (in the case of “Yes” in S62), and the maximum altitude A _PMAX of the distribution range of the point cloud is equal to or greater than the predetermined threshold Th2. In some cases (in the case of “Yes” in S64), the partial space selection unit 22 determines that the partial space is the target partial space. On the other hand, when the number of point groups in the subspace is less than the threshold Th1 (in the case of “No” in S62), and when the maximum altitude A _{PMAX of the} point group is less than the threshold Th2 (“No in S64” ”), The subspace is not determined as the target subspace, and the subsequent processing for the subspace is not performed.

The block space setting means 24 sequentially sets a plurality of block spaces that are stacked in the height direction of the target partial space. For example, the block space setting means 24 sequentially sets the block space from the lower position of the target partial space upward, and when the height of the upper end of the block space exceeds the maximum altitude A _PMAX of the point group of the _target partial space, The setting of the block space in the target partial space is finished. The check target altitude h _C is a variable for controlling this operation, and the search start altitude (the height obtained by adding an offset value to the height of the lower end of the block space to be set first) is set as an initial value of h _C. (S66). Since there are many features other than detection objects such as guardrails, fences, and planting in the block space close to the reference plane, it is preferable to provide an offset value of about 3 to 5 m. The offset value can be adjusted by the user operating the input device 8. h _C is updated by adding the height H _B of the block space every time the block space is set (S 70), and the block space setting means 24 has a height H _B with the updated h _C as the upper end height. A block space is generated and a point cloud is taken into the block space (S72).

  If the point space does not exist in the generated block space (in the case of “No” in S74), the block space setting means 24 returns to the processing S70 and generates the next block space.

  On the other hand, when a point cloud exists in the generated block space (in the case of “Yes” in S74), the wall surface search means 26, the cylindrical surface search means 28, the re-search space setting means 30 and the cylindrical surface re-establishment in the block space. Processing by the search means 32 is started (processing proceeds to node A in FIG. 4).

  If the wall surface determination is not performed for the block space (in the case of “No” in S76), the wall surface determination by the wall surface searching means 26 is performed (S78). The determination result of the wall surface determination process S78 or the determination result when the wall surface determination has already been performed (in the case of “Yes” in S76) indicates that there is something other than the wall surface in the block space. If there is (in the case of “No” in S80), the utility pole extraction process in the block space is performed (S82). On the other hand, when the result of the wall surface determination indicates that only the wall surface exists in the block space (in the case of “Yes” in S80), the utility pole extraction process S82 is not performed.

Above processing S70~S82 Between checked altitude _{h C} is the maximum altitude _{A PMAX} smaller, repeated (S68, S84).

As will be described later, in the utility pole extraction process S82, the cylindrical surface search means 28 and the cylindrical surface re-search means 32 search the cylindrical surface twice, and a cylindrical surface point group is extracted from the block space where the cylindrical surface is detected. Is done. The cylindrical shape determining means 34 is the same method as the cylindrical surface searching means 28 described later, and based on the distribution on the horizontal plane of the cylindrical surface point group extracted corresponding to the target partial space, the center position C and radius of the cylindrical surface R is calculated (S86). Further, the cylindrical shape determining means 34 defines a columnar rectangular parallelepiped space having a horizontal cross section including a circle defined by the calculated center C and radius R, and takes in a point cloud existing in the space (S88). The vertical center axis of the rectangular parallelepiped is located at the center C of the circle, and one side is set to a square horizontal section having a size corresponding to the diameter 2R. Corresponding to the search processing so far performed by the cylindrical surface search means 28 and the like, for example, the heights of the upper and lower ends of the rectangular parallelepiped are set to the maximum altitude A _PMAX and the minimum altitude A _PMIN of the point group in the target subspace. Can do.

The cylindrical shape determining means 34 extracts a point group in the space set in step S88 that has a position in the horizontal plane on the circumference defined by the center C and the radius R, and the height of the extracted point group. range of the maximum value and the minimum value respectively a maximum altitude a _H and the minimum altitude a _L telephone poles shaped (S90). Then, the center C, the radius R, a wire frame model representing the shape of a cylindrical surface which is specified by the maximum altitude A _H and the minimum altitude A _L is generated (S92), a cylindrical feature registration unit 36 lines constituting the model Is registered in the storage device 6 (S94).

  Next, the wall surface determination process S78 and the utility pole extraction process S82 will be described. 5 and 6 are schematic flowcharts of the wall surface determination process S78. FIGS. 5 and 6 show a series of processes connected to the node B. FIG. FIG. 7 is a schematic diagram for explaining the processes S100 to S104 for detecting the wall surface in the block space in the wall surface determination process S78. FIG. 7 is a top view of the block space, and shows an example of the arrangement of the rectangle 50 defined by the width W and the depth D of the block space and the point group in the XY plane in the block space. Has been. The wall surface search means 26 selects arbitrary two data points PA and PB from the point group in the block space, and uses the line segment L0 in the XY plane having these two points as both edges as the edge (horizontal plane inner wall surface position). ) As a candidate line (S100). Then, strip-like areas EA each having a width w (total width 2w) are set on both sides of the line segment L0. A rectangular parallelepiped wall surface determination space having a height range matching the block space and having the strip-shaped area EA as a horizontal cross section is set, and a point group existing in the space is captured (S102). The wall surface search means 26 determines whether the point group of the space can be separated as a wall surface (S104). For example, in the determination process, the wall surface search means 26 counts the number of data points whose coordinates on the XY plane are located in the area EA among the point group in the block space. Whether or not the data point is in the area EA can be determined by whether or not the length of the perpendicular line from the data point to the line segment L0 on the XY plane is w or less, for example. In addition, the edge condition imposes that the three-dimensional shape represented by the point group in the area EA (or the wall surface determination space) is not flat. Specifically, the wall surface search means 26 determines that the data point included in the area EA is not flat if the height difference (difference between the maximum elevation and the minimum elevation) ΔZ is equal to or greater than a preset step threshold γ. To do.

  Here, the width w is a parameter and absorbs the variation in the position in the direction orthogonal to the edge of the data point corresponding to the wall surface. For example, w can be about 3 cm. The height difference threshold γ can be set to about 2 cm, for example.

  The wall surface search means 26 sets candidate lines for all combinations of two data points in the block space, performs the above-described determination, gathers in the area EA above a predetermined reference, and ΔZ is a threshold γ or more. The segment L0 is searched in the block space, and the segment L0 is set as a point group concentrated segment. The reference can be, for example, the number of point groups assumed at the position of the wall surface, and the number is set according to the point density of the laser scan, the size of the wall surface determination space, or the like. The reference can also be defined by the ratio of the number of point groups in the wall determination space to the number of point groups in the block space. For example, the wall surface searching means 26 determines a line segment L0 where the most data points are collected in the area EA as a point group concentrated line segment.

  When the point group concentration line segment is not detected, it is determined that the point group in the block space to be processed cannot be separated as a wall surface (in the case of “No” in S106), and is determined to be due to features other than the wall surface. (S108).

  On the other hand, when a point group concentrated line segment is detected, the line segment is an edge of the wall surface, and the point group in the wall surface determination space can be separated as a wall surface (in the case of “Yes” in S106). However, in the present embodiment, as already described, it is further determined whether or not it is a wall surface after searching for a connected line segment with respect to the point cloud concentration line segment. The search process for the connected line segment will be described with reference to FIGS. The wall surface search means 26 changes the angle of the connected line segment connected to one end and the connected line segment connected to the other end of the point group concentrated line segment detected in the block space with respect to the extension line of the point group concentrated line segment. Explore.

  For example, at one end of the point cloud concentration line segment, the direction of the extension line of the line segment is set to a search angle θ = 0 °, and θ is changed to 90 ° by 2 ° on the right side with respect to the extension line, It is examined whether or not a connected line segment exists at each set search angle θ (S110). At each search angle θ, it is checked whether or not there is a point group concentrated line segment that becomes a connected line segment by basically the same processing as the search of the point group concentrated line segment in the block space. Specifically, a wall surface determination space connected to one end of the point cloud concentration line segment detected in steps S100 to S106 is set. The wall surface determination space is set to a rectangular planar shape similar to the area EA. The rectangle is arranged such that the midpoint of the short side is connected to one end of the point cloud concentration line segment, and the center line along the long side coincides with the search angle θ. A point cloud existing in the set wall surface determination space is taken in the same manner as in step S102, and it is determined whether the point group can be separated as a wall surface in the same manner as in step S104 (S112).

  When a point cloud that can be separated as a wall surface is detected in the search (in the case of “Yes” in S114), the point cloud that the center line segment L0 of the area EA of the wall surface determination space at that time constitutes a connected line segment Middle line segment. When a connected line segment is detected, the point cloud concentration line segment detected in the original block space is an edge of the wall surface, and it is determined that the point cloud in the wall surface determination space can be separated as a wall surface. (S116).

  Similar to the search processing in the right angle range at one end of this point cloud concentration line segment (S110 to S118), the search processing in the left angle range at one end of the point cloud concentration line segment (S120 to S128), the point cloud collection Search processing in the right angle range at the other end of the middle line segment (S130 to S138) and search processing in the left angle range at the other end of the point group concentrated line segment (S140 to S148) are performed. If the connected line segment is not detected in any of the search processes in these four angle ranges, it is determined that the point cloud concentration line segment detected in the original block space is not an edge of the wall surface (S150). ).

FIG. 8 is a schematic flowchart of the utility pole extraction process S82. The cylindrical surface search means 28 takes in a point group to be processed in the block space (S200). Here, when there is a point group separated as a wall surface in the wall surface determination processing S78 in the block space, the cylindrical surface search means 28 leaves the point group excluded from the point group existing in the block space. Are set as processing targets of the cylindrical surface search. When there is a point group to be processed in the block space (in the case of “Yes” in S202), a substantial cylindrical surface search process is started, and the cylindrical surface search means 28 performs processing target points in the block space. Based on the group, a process of obtaining the center C _α and the radius R _α of the cylindrical surface is performed (S204).

Processing for calculating the center C _α [X _Cα , Y _Cα ] and the radius R _α will be described. The cylindrical surface search means 28 selects arbitrary three points (P1 [X _P1 , Y _P1 ], P2 [X _P2 , Y _P2 ], P3 [X _P3 , Y _P3 ]) from the point group. When the circle defined by the center C _α and the radius R _α passes through these three points, the following equations (1) to (3) are obtained.
(X _Cα -X _P1 ) ² + (Y _Cα -Y _P1 ) ² = R _α ² (1)
(X _Cα -X _P2 ) ² + (Y _Cα -Y _P2 ) ² = R _α ² (2)
(X _Cα -X _P3 ) ² + (Y _Cα -Y _P3 ) ² = R _α ² (3)

X _Cα , Y _Cα and R _α are calculated by solving them together. Here, the magnitude of the radius can be assumed according to the cylindrical feature to be detected. In the present system for extracting the utility pole, the diameter is roughly about 20 cm, so the radius can be assumed to be in the range of about 10 cm. Therefore, if the calculated radius R _alpha does not correspond to the range, three P1~P3 selected is determined not to a point on the cylindrical surface of the utility pole.

When the radius R _α falls within the range, the calculated number N _ON of points on the cylindrical surface and the number N _IN of points inside the cylindrical surface are counted. Specifically, an arbitrary point is sequentially selected from a point group in the block space, and a distance r from the center C _α of the position in the XY plane of the point is calculated. When r> R _α + ΔR, the selected point is not counted as being located outside the point group width on the assumed surface of the utility pole. On the other hand, if _{R α ≦ r ≦ R α +} ΔR _counts up _{N ON.} Also, if a r <R _alpha, counts up the _{N IN.}

Results for all the point group in the block space was determined of the r, if the internal telephone poles point also exists at one point, that is, when N _IN is not zero, three P1~P3 that the selected utility poles cylinder Judge that it is not a point on the surface. On the other hand, when N _IN is 0, the number N _ON of points on the circumference is recorded for a circle (center C _α and radius R _α ) obtained based on the three points P1 to P3.

The cylindrical surface search means 28 performs the above calculation for all combinations of arbitrary three points P1 to P3 selected from the remaining point group in the block space, and determines the cylindrical surface having the largest number of points N _ON on the circumference of the block. A candidate for a cylindrical surface of a utility pole in space.

When a candidate cylindrical surface is specified, that is, when the center C _α can be calculated (in the case of “Yes” in S206), a re-search process is performed. On the other hand, (the case of "No" in S206) If the center C _alpha can not be calculated, and ends the utility pole extraction processing S82.

FIG. 9 is a schematic diagram illustrating an example of a re-search space in which the re-search process is performed. FIG. 9 is a view of the target space from above, and a point group obtained from the arrangement of a certain block space (rectangle 50), the arrangement of a re-search space (rectangle 60), and the cylindrical surface of the utility pole in the horizontal plane. 62 distributions are shown. The re-search space is set so that the center of the horizontal cross section coincides with the center C _α specified by the cylindrical surface search means 28 and one side of the rectangle 60 is larger than the rectangle 50 by about 10 cm. As a result, the re-search space captures not only the point group existing in the block space but also the point group existing in the adjacent space as the point group existing on the surface of the cylindrical surface candidate detected in the block space. be able to.

The re-search process following the initial search process (S200 to S206) by the cylindrical surface search means 28 will be described with reference to FIG. The re-search space setting means 30 sets a re-search space, and a point group existing in the space is read into a work area such as a RAM as a processing target point group (S208). The cylindrical surface re-search unit 32 performs processing for obtaining the center C _β and the radius R _β of the cylindrical surface based on the processing target point group in the re-search space (S210). This process can be performed basically in the same manner as the first search process S204. Using a re-search space that can encompass the entire circumference and using a point cloud where candidates for the cylindrical surface were detected in the initial search, the re-search process can detect the cylindrical surface with better accuracy than the initial search. High nature. Moreover, in this embodiment, in order to improve accuracy, four points P1 to P4 are selected from the point group, and a cylindrical surface whose cross section is a circle passing through these four points is searched. When the cylindrical surface passing through the four points is specified, that is, when the center C _β can be calculated (“Yes” in S212), the center of the re-search space is shifted from C _α to C _β ( S214), thereby present on the surface of the cylindrical surface defined by the center C _beta and radius R _beta out of the points of the re-search space position is corrected to extract the cylindrical surface point group described above (S216) .

The utility pole extraction process S82 ends when the cylindrical surface point group is extracted in the process S216. Further, when there is no point group to be processed in the block space (in the case of “No” in S202), and when the center C _α or C _β cannot be calculated (“No” in any of S206 and S212). In the case of), the process is terminated without extracting the cylindrical surface point group.

When the above-described utility pole extraction processing S82 is performed on a plurality of block spaces set as the target partial space and the cylindrical surface point group is extracted, the cylindrical shape determining means 34 is based on the cylindrical surface point group as described above. Te, center C, the radius R, the maximum seek altitude _{a H} and the minimum altitude _{a L} as a parameter for specifying a cylindrical surface (S86~S90).

The cylindrical shape determining means 34 generates a wire frame model representing the shape of the cylindrical surface specified by the parameter (S92), and the cylindrical feature registering means 36 registers the lines constituting the model in the storage device 6 (S94). ). FIG. 10 is a schematic flowchart of the utility pole shape registration process. A polygonal line approximating a circle defined by the center C and the radius R is generated (S250). It is checked whether another figure has already been registered at the position of the generated line (S252). If another figure has already been registered (in the case of “Yes” in S254), it means that the areas of the utility pole and other features overlap, and it is estimated that the detection result is not correct. Therefore, in this case, the utility pole shape is not registered. On the other hand, if no other graphic has been registered (“No” in S254), the lines constituting the wire model representing the utility pole shape are registered (S256). Specifically, utility pole line approximating the largest circle of the cross section of a highly A _H of the cylindrical surface, the minimum altitude A line approximating a circle cross section at _L, and the maximum polygon that approximates the altitude of the circle thereof A line on the side of the utility pole that connects the vertex and the vertex of the polygon that approximates the circle of the minimum height is obtained and registered (S256).

  The utility pole search process described above is performed for the primary subspace and the secondary subspace as described above. The secondary subspace is basically set at a position shifted with respect to the primary subspace, and the utility pole search process is performed on the subspaces that are shifted from each other, so that any one of the utility poles that are not extracted in one of the utility pole search processes. It becomes possible to extract by the other telephone pole search process. That is, the utility pole search process for the primary subspace and the utility pole search process for the secondary subspace complement each other, and the utility pole can be suitably detected.

  Specifically, the range scanned by the laser scanner is generally a part of the entire circumference of the cylindrical surface of the utility pole. For example, a plurality of point groups obtained from an arc-shaped surface having a relatively narrow central angle are provided. The point cloud divided into each partial space is easily misrecognized as a wall surface. When the erroneous recognition occurs, the point group is removed in the wall surface determination S78 and is not extracted as a utility pole. On the other hand, by setting the partial space to be shifted, it is possible to prevent the point group obtained from the arc-shaped surface from being divided or to reduce the number of divisions, and to make it difficult to determine the wall surface. .

  In addition, for example, there is a case where a utility pole is installed at an eaves end for drawing an electric wire into a house. The point cloud obtained from the eaves of the sloped roof has a wide distribution range in the horizontal plane, and since the distribution range in the height direction is narrow, it is difficult to be determined as a wall surface. Therefore, it may happen that the point group of the roof is mixed into the point group in the block space to be processed in the utility pole extraction process S82, and the cylindrical surface cannot be suitably extracted. Also in this case, if the partial space is shifted, the influence of the point group other than the utility pole is reduced in any of the partial spaces, and the possibility that the utility pole can be extracted increases.

As described above, the utility pole is usually tapered, and the diameter of the concrete utility pole is reduced by 1 cm as it goes up, for example, 75 cm. As already described, it is preferable to consider this when setting ΔR. Specifically, it is possible to include a half of the amount of change in diameter at a height H _B of the block space and re-search space for the first time the discovery and re-discovery of the cylindrical surface to [Delta] R. Further, in the above-described embodiment, the circular radius of the cross section of the cylindrical surface is specified on the assumption that the radius R _α or R _β of the cylindrical surface is around 10 cm. The radius according to this assumption may be changed according to the altitude at which the block space or re-search space to be processed is set. Further, in the above-described process S86, the center C and the radius R are obtained by projecting the cylindrical surface point group extracted from each of the plurality of re-search spaces in the height direction onto the horizontal plane. On the other hand, on the premise that the cylindrical surface where the cylindrical surface point group extracted from the plurality of stages has a taper, processing S86 is performed with a conical surface adapted to the three-dimensional distribution of the cylindrical surface point group. You may replace with the process to calculate.

  As described above, the present invention for extracting cylindrical features has been described by taking a system for extracting utility poles as an example. However, the present invention can also be used for extraction of cylindrical features other than utility poles, for example, road signs and street lights. It can also be applied to the detection of the column.

  2 electric pole extraction system, 4 arithmetic processing device, 6 storage device, 8 input device, 10 output device, 20 partial space setting means, 22 partial space selection means, 24 block space setting means, 26 wall surface searching means, 28 cylindrical surface searching means 30 re-search space setting means, 32 cylindrical surface re-search means, 34 cylindrical shape determining means, 36 cylindrical feature registration means.

Claims (7)

A data analysis device for detecting a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in a target space,
A partial space setting means for dividing the target space into a two-dimensional mesh shape in a horizontal plane by a vertical plane, and setting a plurality of columnar partial spaces;
A partial space selecting means for selecting, as a partial space of interest, a point cloud distributed to a height equal to or higher than a predetermined threshold according to the height of the target feature in the partial space;
Among the line segments in the horizontal plane of the target partial space, the point cloud projected to the horizontal plane is searched for a point cloud concentrated line segment that is gathered in the vicinity of a predetermined reference or more, and the point cloud concentrated line segment is searched for the feature. A wall surface search means for estimating the position of the wall surface in the horizontal plane;
Cylindrical surface search means for obtaining a residual point group excluding the point group corresponding to the wall surface and searching for the position of the cylindrical surface of the target feature based on the distribution of the residual point group in a horizontal plane. When,
A data analysis apparatus characterized by comprising: The data analysis apparatus according to claim 1,
A block space setting unit configured to divide the partial space of interest by a horizontal plane and set a plurality of block spaces stacked vertically;
The wall surface searching means searches for a wall surface position for each block space,
The cylindrical surface search means searches for the position of the cylindrical surface using the point group in the block space where the wall surface position is not detected as the remaining point group,
A data analysis device characterized by In the data analysis device according to claim 2,
The wall surface search means searches for a connected line segment connected to the point group concentrated line segment in the space adjacent to the block space and connected to the point group concentrated line segment detected in the block space, and the connected line. A data analysis apparatus characterized by estimating the point cloud concentration line segment of the block space as the wall surface position when there is a minute. In the data analysis device according to claim 2 or 3,
When the position of the cylindrical surface is detected in the block space, a columnar re-position having a horizontal cross section located at a height corresponding to the block space and having a size including the horizontal cross section of the target feature. Re-search space setting means for setting the search space around the central axis of the cylindrical surface;
A cylindrical surface that obtains the position of the cylindrical surface based on the distribution of the point group in the re-search space in the horizontal plane, and extracts a cylindrical surface point group existing on the surface of the cylindrical surface from the point group Re-search means;
A data analysis apparatus characterized by comprising: The data analysis apparatus according to claim 4, wherein
A data analysis apparatus comprising: a cylindrical shape determining means for obtaining a three-dimensional shape of the cylindrical surface based on the cylindrical surface point group extracted corresponding to the block space. A data analysis method for detecting a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in a target space,
A partial space setting step of dividing the target space into a two-dimensional mesh shape in a horizontal plane by a vertical plane, and setting a plurality of columnar partial spaces;
A partial space selection step of selecting, as the partial space of interest, a point cloud distributed to a height equal to or higher than a threshold set in advance according to the height of the target feature in the partial space;
Among the line segments in the horizontal plane of the target partial space, the point cloud projected to the horizontal plane is searched for a point cloud concentrated line segment that is gathered in the vicinity of a predetermined reference or more, and the point cloud concentrated line segment is searched for the feature. A wall search step for estimating the position of the wall surface in the horizontal plane;
A cylindrical surface search step for obtaining a residual point group excluding the point group corresponding to the wall surface and searching for a position of the cylindrical surface of the target feature based on a distribution of the residual point group in a horizontal plane. When,
A data analysis method characterized by comprising: A program for causing a computer to perform processing for detecting a target feature having a cylindrical shape based on point cloud data representing a three-dimensional shape of a feature surface in a target space, the computer comprising:
A partial space setting means for dividing the target space into a two-dimensional mesh in a horizontal plane by a vertical plane, and setting a plurality of columnar partial spaces;
A partial space selection means for selecting, as the partial space of interest, a point cloud distributed to a height equal to or higher than a predetermined threshold according to the height of the target feature among the partial spaces;
Among the line segments in the horizontal plane of the target partial space, the point cloud projected to the horizontal plane is searched for a point cloud concentrated line segment that is gathered in the vicinity of a predetermined reference or more, and the point cloud concentrated line segment is searched for the feature. A wall surface searching means for estimating the position of the wall surface in the horizontal plane, and obtaining a residual point group excluding the point group corresponding to the wall surface, based on the distribution of the residual point group in the horizontal plane, A program which functions as a cylindrical surface search means for searching a position of a cylindrical surface of the target feature.

Images