JP7773423B2 - Structural analysis device, structural analysis method, and program - Google Patents

Description

This invention relates to a structural analysis device, a structural analysis method, and a program.

Historic structures are sometimes restored or relocated for long-term maintenance and preservation. In recent years, when such structures are relocated or dismantled before being restored or restored, restoration is facilitated by taking photographs of the structure in advance and performing photogrammetry or 3D laser scanning to identify the original three-dimensional structure.

Such structures include brick walls. Brick walls are made by stacking roughly rectangular bricks, and the mesh-like joints (joints) form depressions. The reflection intensity of the laser light at these joints differs from that of the brick itself, allowing the joints to be identified and structural information about the brick wall to be obtained. Patent Document 1 describes a technology that detects the position of the gradient peak where the reflection intensity changes in a laser scan as the position of the brick boundary, and then visualizes the brick wall.

AUTOMATIC CREATION OF STRUCTURAL MODELS FROMPOINT CLOUD DATA: THE CASE OF MASONRY STRUCTURES, B. Riveiro et al., ISPRS Annals of Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume II-3/W5, 2015. ISPRS Geospatial Week 2015, September 28-October 3, 2015 (France)

However, as time passes after a structure made of block-shaped structural materials, such as a brick wall, is constructed, numerous irregularities develop due to deterioration and damage. With conventional technology, it has been difficult to accurately distinguish such irregularities from the joints of the structural materials and identify the structure.

The purpose of this disclosure is to provide a structural analysis device, structural analysis method, and program that can extract candidate points for structural joints with higher accuracy.

In order to achieve the above object, the present disclosure provides:
1. A structure analysis device for detecting joints between a plurality of members having flat surfaces on a surface including the flat surfaces of a structure in which the flat surfaces are combined with each other, the joints being oriented in approximately the same direction, the device comprising:
an acquisition means for acquiring point cloud data, which is a set of points on the surface, the positions of which are specified in a three-dimensional orthogonal coordinate system having three axes: a first direction substantially parallel to the surface, a second direction substantially parallel to the surface and perpendicular to the first direction, and a third direction perpendicular to the first and second directions;
a calculation means for calculating, for each local region of the surface, at least one of diagonal components relating to the first direction and the second direction in an inverse matrix of a covariance matrix of three-dimensional coordinates of a plurality of points included in the local region among the plurality of points;
a determination means for determining whether the seam is present in the local region using the calculated diagonal components;
The present invention is characterized by comprising:

This disclosure has the advantage of being able to extract candidate points for structural joints with greater accuracy.

FIG. 1 is a block diagram showing the configuration of a structure analysis system. 1A and 1B are diagrams illustrating an example of a brick wall before and after coordinate transformation. FIG. 10 is a diagram illustrating an example of specifying the position of a joint using LoG. 10 is a flowchart showing a control procedure for a structural analysis process. 10 is a flowchart showing a control procedure for horizontal joint identification processing. 10 is a flowchart showing a control procedure for vertical joint identification processing. FIG. 1 is a diagram showing an example of structural analysis of a joint in a brick wall. FIG. 1 is a diagram showing an example of structural analysis of a joint in a brick wall.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing the configuration of a structure analysis system 100 including a structure analysis apparatus according to this embodiment.
The structure analysis system 100 includes a processing device 1 as a structure analysis device, and a storage device 2 that stores position data of structures.

The processing device 1 may be a typical personal computer (PC, computer) and includes a CPU 11 (Central Processing Unit), RAM 12 (Random Access Memory), a storage unit 13, a communication unit 14, a display unit 15, an operation reception unit 16, etc.

The CPU 11 performs arithmetic processing and provides overall control over the operation of the processing device 1. The CPU 11 may be a single processor, or multiple processors operating in parallel or independently depending on the application.

RAM 12 provides working memory space for CPU 11 and stores temporary data. RAM 12 is not limited to, but may be, for example, DRAM. If CPU 11 has multiple processors, each processor may have its own corresponding RAM, or it may be RAM shared by the multiple processors.

The memory unit 13 is an auxiliary storage device that stores a program 131 related to the brick wall (structure) analysis process, setting data, etc. The memory unit 13 may be, for example, a non-volatile memory such as a flash memory, or an HDD (Hard Disk Drive). The memory unit 13 may be configured externally to the processing device 1 (a configuration external to the processing device 1), or may be an accessible network drive or cloud server that exists on a network. The setting data includes, for example, screen settings and display order settings related to the display of the matching results on the display screen.

The communication unit 14 controls data transmission and reception (communication) with external devices in accordance with a predetermined communication standard. The predetermined communication standard may include, for example, TCP/IP related to a LAN (Local Area Network). The communication unit 14 may also have a connection terminal such as a USB (Universal Serial Bus) and be able to control one-to-one communication with a connected device via USB.

The display unit 15 has a digital display screen, and displays information on the digital display screen under the control of the CPU 11. The digital display screen is not particularly limited, but may be, for example, a liquid crystal display screen (
LCD).
The operation reception unit 16 includes a pointing device such as a mouse, a keyboard, etc., receives input operations, and outputs operation signals according to the received contents to the CPU 11.
The display unit 15 and the operation reception unit 16 may be peripheral devices (configurations external to the processing device 1 (computer)) connected to the connection terminals of the communication unit 14 .

The storage device 2 stores measurement data 21 of the surface of the brick wall. The storage device 2 may be a dedicated database device such as a network drive or cloud server, or a PC with an internal or external auxiliary storage device (HDD) of appropriate capacity. The measurement data 21 is a collection of the three-dimensional positions of numerous points on the surface. The three-dimensional positions may be determined using SfM or similar techniques from multiple images taken by an optical camera, or may be determined by measuring the distance and direction from the laser light emission point using a laser scanner or similar. The spacing between measurement points may be uniform or random, but must be sufficiently small compared to the width of the joints in the brick wall being analyzed.

Next, we will explain the analysis of brick wall joints, which is a structural analysis in this embodiment.

While not particularly limited, a brick wall (wall) subject to structural analysis is generally a structure assembled by arranging multiple structural members (components) with roughly rectangular parallelepiped shapes and exposed flat surfaces horizontally (in a certain direction) so that the flat surfaces (rectangular) are aligned on the same roughly vertical plane (surface) (with normals pointing in roughly the same horizontal direction), with the long sides of the structural members horizontal, and then stacking them vertically (perpendicular to the certain direction) in multiple layers (the horizontal positions of individual structural members may vary), with each structural member connected by mortar joints (seams). The length of the short sides of the structural members, i.e., their vertical width (length), is roughly the same. Therefore, a brick wall has multiple horizontal joints (first joint group) that extend horizontally across multiple structural members and are spaced roughly equally apart in the vertical direction, forming the boundaries of each layer. However, due to changes over time, slight distortion may occur, especially in the mortar parts, so it is often not possible to achieve a perfectly straight line extending horizontally.

The joints where the short sides of structural materials meet (vertical joints roughly perpendicular to horizontal joints, the second group of joints) are often staggered horizontally, meaning that their positions vary for each row. Furthermore, the length of the long sides of structural materials, i.e., their horizontal width (width), is often roughly the same, but may vary depending on the joints and patterns. Therefore, vertical joints must be individually specified for each row, and they do not necessarily appear at equal intervals.

These joints usually have a width perpendicular to their extension direction, typically due to mortar or similar. Therefore, the center position of the joint must be appropriately identified. Since the joint portion often has a different surface position from the brick structural material (it is generally recessed), the position where the change in the position component perpendicular to the surface is at its maximum or minimum can be determined as the center position.

The processing device 1 uses the inverse matrix of the covariance matrix of the three-dimensional coordinates of multiple points within a reference range (local region) around each point as an identification parameter for joint detection. The local region is preferably set to a size approximately equal to the width of the joint so that, when set within the joint, it includes many points within the joint while minimizing points outside the joint. For example, it can be a circular region centered on each point and with a radius approximately half the width of the joint. For this process, the positions of each measured (calculated) point are converted into a coordinate system with an x-y plane that is approximately parallel to the surface of the brick wall and aligned with the joint direction. The term "approximately parallel" here means that a representative plane representing the uneven surface of the brick wall does not need to be strictly determined based on the measurement data of the entire brick wall, and therefore some tilt is acceptable. Identification parameters can be determined separately for vertical and horizontal joints. As will be described later, by defining such xy directions, it becomes possible to judge horizontal joints and vertical joints separately using the identification parameters Qxx and Qyy. However, if there is a slight tilt (for example, 5 to 10 degrees or less), there is little mixing into the other identification parameter, so there are no major problems with the judgment process or judgment accuracy.

FIG. 2 is a diagram showing an example of a brick wall before and after coordinate transformation.
As shown in FIG. 2A, the position data of the surface of the brick wall W before coordinate conversion may be in any coordinate system, such as a geographic coordinate system with the vertical direction being the Z coordinate and an XY plane parallel to the horizontal plane. The orientation of the surface of the brick wall W on the XY plane is also arbitrary and is not limited to overlapping with the X axis (XZ plane) or the Y axis (YZ plane). Three of the four vertices of a rectangle R along the surface of the brick wall W represented in such an XYZ coordinate system are determined by a user inputting operations into the operation reception unit 16. Two vertices P1 and P2 are set approximately parallel to one of the horizontal joints Jh of the brick wall W to the extent possible through input operations. The setting of the third vertex P3 may be limited to a range perpendicular to one of the two vertices P1 and P2 with respect to the line segment connecting the two previously determined vertices P1 and P2.

If the three determined vertices P1 to P3 coincide with the measurement points, the coordinates of the measurement points may be used as is. Vertices that do not coincide with the measurement points may be determined by linear interpolation of the positions of nearby measurement points, or the operation reception unit 16 may receive coordinate input from the user. As described above, the rectangle represented by the three vertices does not have to be strictly parallel to the average surface of the brick wall W, but the three vertices P1 to P3 may be set in parts of the brick B (structural material) that are outside the vertical joints Jv and horizontal joints Jh and have little damage or deterioration, so that the surface is as close to parallel as possible.

A coordinate transformation (affine transformation) is performed to translate and rotate the rectangle R (i.e., three points on the surface of the brick wall) so that it is located on the xy plane of a three-dimensional Cartesian coordinate system (a Cartesian coordinate system) with coordinate axes different from those of the XYZ coordinate system. That is, the coordinates of the three vertices P1 to P3 in the xyz coordinate system are determined so that the z coordinates (third direction) of the three vertices are zero, the y coordinates (second direction) of vertices P1 and P2 are equal, and the x coordinates (first direction) of vertices P2 and P3 are equal. In this case, any one of the vertices may be set as the origin (0,0,0), or any other xy coordinate may be set. Determining the coordinates so that the distance between vertices P1 and P2 and the distance between vertices P2 and P3 is maintained results in a one-to-one transformation, but this is not limited to this. However, when performing a non-one-to-one transformation on laser measurement data expressed in actual size, the scale factor σ (described below) must also be adjusted accordingly. Coordinates may be set by a user's input operation to the operation reception unit 16, and the coordinate components determined according to the conditions as described above may be fixed in advance or according to the input operation of a certain value.

By determining the coordinates of the three vertices P1 to P3 before and after the transformation, the coordinate transformation parameters (each component of a three-dimensional matrix) from the XYZ coordinate system to the xyz coordinate system can be determined. Once the processing device 1 calculates these coordinate transformation parameters, it automatically converts the coordinates of all measurement points into coordinates in the xyz coordinate system, determining a set of positions of multiple points on the brick wall (point cloud data) as shown in Figure 2(b). That is, the x direction is oriented (parallel) along the horizontal joints Jh (the typical extension direction of each joint in the first joint group; it is also the direction in which the structural materials are arranged), the y direction, which is perpendicular to the x direction in a plane parallel to the surface, is oriented roughly along the vertical joints Jv (parallel to the direction in which the structural materials are stacked), and the z component (component in the third direction) represents the unevenness in the direction perpendicular to the surface of the brick wall.

As described above, the surface of the brick wall W is generally flat on the bricks B, but has unspecified irregularities (mainly recesses) due to deterioration, damage, etc. On the other hand, the surface of the brick wall W has regions where the z component extends regularly in the horizontal direction (x direction) along the horizontal joints Jh and regions where the z component extends regularly in the vertical direction (y direction) along the vertical joints Jv and has non-zero z component.

The uv components _Ciuv (where u and v are x, y, or z, respectively) of the covariance matrix Ci used to calculate the identification parameters related to joint detection are the covariance of the u and v components of the coordinates of multiple points located within a reference range (local region) of a certain size from a certain point. Since the measurement points are distributed roughly randomly on the surface of the brick wall W, the variance (variation) of the x and y components of these coordinates is roughly uniform. On the other hand, the variance (variation) of the z component varies depending on whether or not there are vertical joints Jv or horizontal joints Jh within the local region. When a vertical joint Jv is included in a local region near a vertical joint Jv, the yz and zz components become larger due to the inclusion of many non-zero z component points scattered and distributed in the y direction within a certain x range (since the covariance matrix is generally symmetric, the yz and zy components are approximately equal), while in the case of a horizontal joint, the xz and zz components become larger due to the inclusion of many non-zero z component points scattered and distributed in the x direction within a certain y component range (similarly, the xz and zx components are approximately equal).When the z component becomes non-zero due to an irregular shape caused by deterioration or damage, this will appear in Czz if the size is large, but in comparison, it is difficult for clear non-zero components to appear in the off-diagonal components.

The discrimination parameters for the point r _i =(x _i , y _i , z _i ) are obtained from the diagonal components of the inverse matrix Q i of the covariance matrix Ci shown in Equation 1 below.

Since the inverse matrix can be expressed using a cofactor matrix, the (m, n) components of the inverse matrix Qi (m and n are x, y, or z, respectively) correspond to the (n, m) cofactors of the covariance matrix Ci. Therefore, Qi _xx (the diagonal components in the first direction) tend to be larger at the vertical joint Jv (second joint group) containing the yz, zy, and zz components of the covariance matrix Ci, while Qi _yy (the diagonal components in the second direction) tend to be larger at the horizontal joint Jh (first joint group) containing the xz, zx, and zz components of the covariance matrix Ci. The other components tend to be smaller than Qi _xx and Qi _yy . In particular, by using the diagonal components of the inverse matrix Qi rather than directly using each component of the covariance matrix Ci, the effects of noise, damage, and deterioration can be further reduced. Therefore, by using the magnitudes of the diagonal components Qi _xx and Qi _yy of the inverse matrix Qi as the identification parameters Qxx and Qyy, it is possible to distinguish and accurately narrow down (determine) the candidate positions (points where joints may exist) of horizontal joints (first joint group) and vertical joints (second joint group) while reducing the influence of unevenness due to surface deterioration or damage. The threshold value Tc for determining whether or not there is a joint can be determined experimentally in advance. Note that since components other than the xx and yy components of Qi are unnecessary, there is no need to calculate the entire inverse matrix Qi (this may be determined by checking the error, or unnecessary components may be obtained collectively using an existing inverse matrix calculation program).

Since there may be finer irregularities on the surface of the brick wall or variations in measurement, the z-component value of each point may be smoothed to remove noise before calculating the covariance matrix Ci. A well-known technique, for example, a Gaussian filter, may be used to remove noise. For example, for a certain point _ri = ( _xi , _yi , _zi ), the z-component (irregularity) of each point may be smoothed as zi = _Σj ( _g · _zj ), which is the sum of the x- and y-components ( _xj , _yj ) of the coordinates of each neighboring point _rj (including _{point ri} itself) weighted by a weight g that corresponds to the relative position (distance) from ( _xi , _yi ) and a scale factor σ.

Here, the scale factor σ is preferably set to be smaller than the joint width, and in particular, 3σ, which is the range where the value of the Gaussian filter g cannot be ignored (near each point r _i ), is smaller than the joint width, for example, approximately 0.3 times the joint width. By setting the scale factor σ in this way, it is possible to appropriately reduce the influence of small irregularities inside the joint and bricks (structural materials) while not eliminating the difference in z-components inside and outside the joint. The measurement points are dense enough that multiple points fall within a radius of 3σ from a given point.

On the other hand, the center position of the vertical joint and horizontal joint candidates extracted as described above is determined by the extreme value related to the change in the gradient of the z coordinate (coordinate component in the third direction). For this determination, for example, a Laplacian filter (second-order differential) is used. Alternatively, Laplacian of Gaussians (LoG; the second-order differential of a Gaussian filter, i.e., gradient change) combined with smoothing and noise removal using the Gaussian filter described above may be used. When using this LoG, the gradient change Ez _i of a certain point r _i is expressed as the sum of the contour components of each point within the neighborhood, with the LoG value for each point r _j within the neighborhood as a weight gL, i.e., Ez _i = -Σ _j (gL _j · z _j ). The LoG also includes a scale factor σ, whose value can be set to the same as the scale factor σ of the Gaussian filter described above. By appropriately defining the scale factor σ in LoG, this gradient change Ez _i approaches zero at points other than the contour, i.e., the joint, and becomes a large value at the joint, with an extreme value near the center.

When LoG is used for the coordinates of each point to which the Gaussian filter has been applied, σ ² is multiplied to adjust the scale, resulting in Ez _i = -Σ _j (σ ² · gL _j · z _j ). ^{σ 2} · gL _j may be calculated collectively and then multiplied by z _i .

FIG. 3 is a diagram illustrating an example of specifying the position of a joint using LoG.
For example, when detecting a horizontal joint extending in the x direction, among the measurement points indicated by circles, those with an identification parameter Qyy greater than a reference value (indicated by hatching, white dots, and crosses) are arranged in the x direction with a width corresponding to the width of the joint. One of these points, _ri (indicated by a cross), is selected, and points within a reference range (±Tw/2 in the x direction, ±Th/2 in the y direction) from the selected point _ri are detected (indicated by white dots). The width Th is set to be narrower than the width of the joint but wider than the average spacing between the measurement points. The width Tw may be approximately the same as the width Th, but may also be wider. In other words, it is assumed that multiple measurement points are included within the reference range. The detected points are arranged in order in the y direction intersecting the horizontal joint, and when the change in gradient change Ez _j is observed, if the selected point _ri has an extreme value (if the selected point _ri is an extreme value point), the point _ri is identified as _a candidate point for the joint. By changing the selected points in order and performing the same process, multiple candidate joint points are identified along the joint.

When detecting a vertical joint, the points within the reference range are arranged in order in the x direction that intersects with the vertical joint, and when the change in gradient Ez _j is observed, if the selected point _ri is at the extreme value of the gradient change, the point _ri is identified as a candidate point for the joint.

In this way, the position of the joint is approximately determined by approximating a line representing the joint that extends according to the arrangement of joint candidate points extracted using the identification parameters and identified using the gradient distribution. This line may be represented by an approximate curve or straight line of the arrangement of the joint measurement points. As mentioned above, joints may become distorted over time after construction, so for example, long horizontal joints that are approximately perpendicular to the vertical direction are approximated by a curve, and short vertical joints that are approximately parallel to the vertical direction are approximated by a straight line. Note that long horizontal joints are easier to identify, so it is best to identify the horizontal joints first, followed by the vertical joints.

The above processing allows the joint positions to be obtained from the measurement point data, but any joints whose spacing or length are inappropriate for joints can be excluded as erroneous judgments. After these processes are performed, the user can visually check and correct the detection results, significantly reducing the user's effort and enabling the joint positions to be determined with high accuracy.

By converting the joints obtained in this approximate manner back into the original XYZ coordinate system and drawing them, an elevation drawing showing the position of the joints in the brick wall can be obtained. Note that if the Z-axis direction is not parallel to the vertical direction, the elevation drawing may be further coordinate-converted so that the Za-axis is oriented vertically.

Figure 4 is a flowchart showing the control procedure by the CPU 11 of the structural analysis process executed by the processing device 1 of this embodiment. This structural analysis process, which includes the structure analysis method of this embodiment, is started and executed by the CPU 11 as part of the program 131 in response to an input operation related to a start request by the user, for example, while position measurement data of a brick wall is stored and held in the storage device 2.

The CPU 11 acquires point cloud data indicating the position of the brick wall from the storage device 2 (external) (step S101). The acquisition means (CPU 11) converts the coordinates of each point in the acquired point cloud data (step S102; acquisition step, acquisition function). The acquisition means displays an image of the brick wall on the display unit 15, waits for an operation to set three vertices via the operation reception unit 16, and acquires the operation content. The acquisition means calculates coordinate transformation parameters based on the coordinates of the acquired three vertices before and after coordinate transformation. The acquisition means converts the coordinates of each point in the acquired point cloud data using the calculated coordinate transformation parameters.

The CPU 11 removes (reduces) noise in the position data of the converted point cloud data (step S103). For each point in the point cloud data, the CPU 11 smoothes the z-component value based on the scale factor σ. The scale factor σ may be determined in advance, or may be a numerical value received by the operation reception unit 16. Alternatively, an appropriate numerical value may be determined based on the positions of both ends of the width direction of the joint specified by an operation received by the operation reception unit 16 for the brick wall displayed on the display unit 15.

The calculation means (CPU 11) selects one point from the point cloud data and sets a circular local region centered on the selected point (step S104). The calculation means calculates a covariance matrix Ci of the three-dimensional coordinates of multiple points (including the selected point) within the local region (step S105). The calculation means calculates the inverse matrix of this covariance matrix Ci (step S106). The processing of steps S104 to S106 constitutes the calculation step (calculation function) in the structure analysis method (program) of this embodiment. The detection means (CPU 11) calculates the gradient change Ez _i of the selected point (step S107). As described above, the gradient change is obtained by taking the weighted average (sum) of LoG for the z-component of the coordinates of each point within the local region.

The CPU 11 determines whether all points in the point cloud data have been selected. If it is determined that all points have not been selected (there are unselected points) ("NO" in step S108), the processing of the CPU 11 returns to step S104.

If it is determined that all points have been selected ("YES" in step S108), the CPU 11 executes the horizontal joint identification process described below (step S109). The CPU 11 then executes the vertical joint identification process described below (step S110).

The CPU 11 displays the estimated horizontal and vertical joint positions superimposed on the image of the brick wall on the display unit 15. The CPU 11 waits for an input operation to the operation reception unit 16 and acquires the editing operation details for the horizontal and vertical joints accepted by the input operation (step S111). The CPU 11 adjusts (including deleting and adding) the positions of the horizontal and vertical joints based on the acquired editing operation details (step S112). Note that if no editing operation has been performed (if an approval confirmation operation, etc. has been performed), the processing of step S112 is omitted.

The generation means (CPU 11) converts the obtained joint positions from coordinates in the xyz coordinate system to coordinates in the XYZ coordinate system, and generates an elevation drawing (graphic data) of the brick wall (step S113). Then, the CPU 11 ends the structural analysis process.

FIG. 5 is a flowchart showing the control procedure of the horizontal joint identification process in step S109 of FIG.
In the horizontal joint identification process, the determination unit (CPU 11) extracts points from the point cloud data whose identification parameter Qyy is equal to or greater than a threshold value Tc (first threshold value) (step S201). The CPU 11 selects one of the extracted points that has not yet been selected (step S202). The CPU 11 detects points within a reference range (a range of ±Tw/2 in the x direction and a range of ±Th/2 in the y direction) from the selected point, and sorts the detected points in order of their y coordinates (step S203).

Assuming that the extraction point selected in step S202 is the ith extraction point in the y-coordinate order, the CPU 11 determines whether the gradient change Ez( _i) of this ith extraction point is greater than the gradient change Ez _(i+1) of the (i+1)th point in the sort order and greater than the gradient change Ez _(i-1) of the (i-1)th point in the sort order (step S204). If it is determined that the gradient change Ez _{( i} ) is greater than both the gradient change Ez(i-1) and the gradient change Ez _{(i+1) (} "YES" in step S204), the CPU 11 sets the selected extraction point as a joint candidate point (step S205). The CPU 11 then determines whether all extraction points have been selected (step S206). If it is determined that not all extraction points have been selected (i.e., there are unselected extraction points) ("NO" in step S206), the CPU 11 returns to step S202. If it is determined in the determination process of step S204 that the gradient change Ez _i is smaller than at least one of the gradient change Ez _(i+1) and the gradient change Ez _(i-1) ("NO" in step S204), the process of the CPU 11 proceeds to step S206.

If it is determined in the determination process of step S206 that all extraction points have been selected ("YES" in step S206), the CPU 11 selects one unselected joint candidate point from the set joint candidate points (step S207). The CPU 11 extracts joint candidate points having y coordinates within the reference range ±Th2/2 relative to the y coordinate of the selected joint candidate point (step S208). Th2 is, for example, approximately the width of the joint or slightly wider.

The generation means (CPU 11) performs curve approximation on the extracted joint candidate points to set (determine) a line (first shape) representing the horizontal joint (step S209). The approximated curve is, but is not limited to, a cubic curve. The number of approximated curves may be determined, for example, using the least squares method or the RANSAC (Random Sample Consensus) method. Note that if there are very few extracted joint candidate points (e.g., three or less), there is a high possibility of erroneous extraction, so the corresponding joint candidate points may be excluded without curve approximation. The CPU 11 determines that all joint candidate points used in the curve approximation have been selected (step S210). The CPU 11 determines whether all joint candidate points have been selected (step S211). If it is determined that all joint candidate points have not been selected (there are unselected joint candidate points) ("NO" in step S211), the processing of the CPU 11 returns to step S207.

If it is determined that all joint candidate points have been selected ("YES" in step S211), the CPU 11 calculates the average spacing between the multiple approximate curves obtained (step S212). Here, the CPU 11 calculates the average spacing by dividing the distance between the approximate curve located furthest on the +y side and the approximate curve located furthest on the -y side by the number of spacings corresponding to the number of approximate curves (number of approximate curves minus 1). Because the distance between curves is not constant, approximate straight lines may be obtained in addition to the curve approximation to determine the distance between the approximate straight lines, or the minimum or average of the differences (absolute values) in the y component when comparing positions of the same x component on the approximate curves may be used.

The CPU 11 multiplies the calculated average interval by a predetermined percentage, for example, 30%, to set the lower limit interval Dth (step S213). The CPU 11 then determines whether the interval between the approximate curve at the top (with the largest y component) and the approximate curve directly below (on the -y side) is smaller than the lower limit interval Dth (step S214). If it determines that the interval is smaller than the lower limit interval Dth ("YES" in step S214), the CPU 11 deletes the approximate curve at the top (step S215). The CPU 11 then proceeds to step S216. If it determines that the interval is not smaller than the lower limit interval Dth ("NO" in step S214), the CPU 11 proceeds to step S216.

The CPU 11 determines whether the distance between the approximate curve at the bottom (with the smallest y component) and the approximate curve immediately above it (on the +y side) is smaller than the lower limit distance Dth (step S216). If it is determined that the distance is smaller than the lower limit distance Dth ("YES" in step S216), the CPU 11 deletes the approximate curve at the bottom (step S217). Then, the CPU 11 proceeds to step S218. If it is determined that the distance is not smaller than the lower limit distance Dth ("NO" in step S216), the CPU 11 proceeds to step S218.

The CPU 11 determines whether an approximate curve was deleted in step S215 or step S217 (step S218). If an approximate curve was obtained by detecting the top or bottom end of a brick wall rather than the joints, these may not depend on the width of the bricks (structural materials), and so are excluded in the processing of steps S215 and S217, respectively. If it is determined that an approximate curve was deleted ("YES" in step S218), the CPU 11 recalculates the average spacing between the remaining approximate curves (step S219). Then, the processing of the CPU 11 proceeds to step S220. If it is determined that an approximate curve was not deleted ("NO" in step S218), the processing of the CPU 11 proceeds to step S220.

When the process proceeds to step S220, CPU 11 selects the approximate curve with the smallest y component from among the unselected approximate curves (step S220). CPU 11 determines whether the distance between the selected approximate curve and the approximate curve immediately above it (on the +y side) is smaller than the lower limit distance Dth (step S221). If it is determined that the distance is smaller than the lower limit distance Dth ("YES" in step S221), CPU 11 deletes the upper of the compared approximate curves (step S222). In this case, it is assumed that the upper approximate curve has been erroneously detected due to damage, deterioration, or the like. CPU 11 then returns to step S221.

If it is determined that the distance is not smaller than the lower limit distance Dth ("NO" in step S221), the CXPU 11 determines that the two compared approximate curves have been selected (step S223). The CPU 11 then determines whether all of the approximate curves that have not been deleted have been selected (step S224). If it is determined that all of the approximate curves have not been selected (there are approximate curves that have not been selected) ("NO" in step S224), the processing of the CPU 11 returns to step S220. If it is determined that all of the approximate curves have been selected ("YES" in step S224), the CPU 11 ends the horizontal joint identification processing and returns the processing to the structural analysis processing.

FIG. 6 is a flowchart showing the control procedure of the vertical joint identification process, which is step S110 in FIG.
In the vertical joint identification process, the CPU 11 divides the acquired point cloud data into intervals of the approximate curves of the previously identified horizontal joints, i.e., into rows of bricks (structural materials) (step S301).The CPU 11 then selects one of the divided point clouds (step S302).

The CPU 11 selects one unselected point belonging to the selected point cloud (step S303). The determination means (CPU 11) determines whether the identification parameter Qxx of the selected point is equal to or greater than a threshold value Tc (second threshold value) (step S304). The threshold value Tc used in the horizontal joint identification process may be the same as or different from the threshold value Tc used in the vertical joint identification process. If it is determined that the identification parameter Qxx is equal to or greater than the threshold value Tc ("YES" in step S304), the CPU 11 detects points within a reference range from the selected point (within a range of ±Th/2 for the x component of the selected point and within a range of ±Tw/2 for the y component of the selected point) and sorts the detected points in order of their x coordinates (step S305).

If the extracted point selected in step S303 is the i-th point in the order of the x-coordinate, the CPU 11 determines whether the gradient change Ez( _i ) of this i-th selected point is greater than the gradient change Ez _{(i+1) of the (i+1)} th point in the order of the x-component and greater than the gradient change Ez _{(i-1) of the (i-1} )th point in the order of the x-component (step S306). If it is determined that the gradient change Ez( _i ) is greater than both the gradient change Ez _(i+1) and the gradient change Ez _(i-1) ("YES" in step S306), the CPU 11 sets the selected point as a joint candidate point (step S307). Then, the CPU 11 proceeds to step S308. If it is determined that the gradient change Ez _i is not greater than at least one of the gradient change Ez _(i+1) and the gradient change Ez _(i-1) ("NO" in step S306), the processing of the CPU 11 proceeds to step S308.

Furthermore, if the determination process of step S304 determines that the identification parameter Qxx is not greater than or equal to the threshold value Tc ("NO" in step S304), the CPU 11 proceeds to step S308.

When the process proceeds to step S308, the CPU 11 determines whether all points in the selected point cloud have been selected (step S308). If it is determined that not all points have been selected (there are unselected points) ("NO" in step S308), the process of the CPU 11 returns to step S303.

If it is determined that all points have been selected ("YES" in step S308), the CPU 11 selects one of the set joint candidate points that has not yet been selected in this process (step S309). The CPU 11 extracts joint candidate points within a reference range (±Tw2/2) in the x direction from the selected joint candidate point (step S310). The reference range (±Tw2/2) is set to be approximately the width of the vertical joint or slightly wider so that all joint candidate points located within the same vertical joint as the selected joint candidate point are extracted, and joint candidate points located in different vertical or horizontal joints are minimized. The total width of the reference range, |Tw2|, may be the same as or different from |Th2|, the total width of the reference range in the horizontal joint identification process.

The generation means (CPU 11) performs linear approximation of the extracted joint candidate points to set (determine) a line segment (second figure) representing the vertical joint (step S311). The linear approximation may be performed using the least squares method or the RANSAC method, etc. In this case, since the vertical joint is approximately parallel to the y direction, the coefficients α and β in the approximation equation x = αy + β, with the y coordinate of the joint candidate point as the independent variable, may be found. The y coordinates of both ends of the line segment may be determined, for example, by the maximum and minimum y coordinates, respectively, of the extracted joint candidate points. Note that if there are very few extracted joint candidate points (e.g., three or less), there is a high possibility of erroneous extraction, so the corresponding joint candidate points may be excluded without linear approximation. The CPU 11 considers all joint candidate points used in the linear approximation to have been selected (step S312).

The CPU 11 determines whether all of the set joint candidate points have been selected (step S313). If it is determined that not all of the joint candidate points have been selected (there are unselected joint candidate points) ("NO" in step S313), the CPU 11 returns to step S309. If it is determined that all of the joint candidate points have been selected ("YES" in step S313), the CPU 11 determines whether all of the divided point groups have been selected (step S314). If it is determined that not all point groups have been selected (there are unselected point groups) ("NO" in step S314), the CPU 11 returns to step S302.

If it is determined that all divided point clouds have been selected ("YES" in step S314), the CPU 11 calculates the distance Lu between the top end of each line segment of the acquired vertical joint and the horizontal joint closest to that top end (step S315). The CPU 11 excludes line segments whose distance Lu is greater than the upper limit value Dc from the vertical joint (step S316). The upper limit value Dc may be determined based on the average density of measurement points (average spacing) or the average spacing of horizontal joints. For example, the upper limit value Dc can be determined by multiplying the average spacing calculated in step S212 of the horizontal joint identification process or recalculated in step S219 by a predetermined percentage, such as 25%.

The CPU 11 calculates the distance Ll between the bottom end of the acquired vertical joint line segment and the horizontal joint closest to that bottom end (step S317). The CPU 11 excludes line segments for which the distance Ll is greater than the upper limit value Dc from the vertical joints (step S318). The CPU 11 then ends the vertical joint identification process and returns the process to the structural analysis process.

The processes branching to "YES" in step S201 of the horizontal joint identification process and step S304 of the vertical joint identification process constitute the judgment steps (judgment functions) in the structural analysis method (program) of this embodiment. Furthermore, the processes in step S209 of the horizontal joint identification process, step S311 of the vertical joint identification process, and step S113 of the structural analysis process constitute the generation steps (generation functions) in the structural analysis method (program) of this embodiment.

7 and 8 are diagrams showing an example of structural analysis of a joint in a brick wall.
Figure 7(a) shows the discrimination parameters in contour. It can be seen that the discrimination parameter Qxx increases at vertical joints and is displayed as black (darker), while the discrimination parameter Qyy increases at horizontal joints and is displayed as white (brighter). Meanwhile, Figure 7(b) shows the distribution of the gradient change of the z component based on LoG in shading. In this gradient change distribution, the joints are clearly displayed as white, but many white streaks also appear between the joints.

As shown in Figure 8(a), when a one-dimensional plot of the gradient change in the z-component obtained on the horizontal line between each horizontal joint is superimposed on the contour display of the identification parameter, it can be seen that Qxx is kept small even in areas where the gradient change Ez undulates up and down. Therefore, the identification parameter narrows down joint candidates with greater precision, and when used in conjunction with the gradient distribution, it becomes easier to identify the joint locations. As a result of the above processing, as shown in Figure 8(b), in the elevation view showing the joint distribution, the joint locations are accurately and appropriately identified from the measurement point data of the brick wall.

[Modification]
The present invention is not limited to the above-described embodiment, and various modifications are possible.
For example, in the above embodiment, the determination of whether each measurement point, distributed roughly randomly, is located inside a joint can be performed. However, the determination of whether a point is located inside a joint can also be performed using a grid point of a mesh defined in a two-dimensional matrix having a unit area. For example, a square mesh of a certain size can be defined, and the three-dimensional position of each grid point can be determined by interpolating (e.g., weighted averaging) the position data of the surface of the brick wall W located within the four meshes surrounding each grid point. The grid point data thus obtained can be used as the target point cloud data, and the acquired point cloud data can be used to determine whether each grid point is located inside a joint (whether a joint exists in the local region). In this case, the local region can be defined, for example, to be approximately √2 to 1.5 times the length of the mesh edge, so as to include the target grid point and its surrounding eight grid points.

Alternatively, instead of first defining the extent of the local region, a predetermined number N of points (e.g., 3 to 8 points) may be defined as points within the local region in order of proximity to the point of interest. In other words, in this case, the local region may be a circle with the point of interest at its center and the distance from this point to the Nth closest point as its radius.

In addition, in the above embodiment, the position of the joint was identified based on gradient changes, but this process may be omitted and the joint range may be narrowed down solely by judgment using identification parameters. The number of points within a joint with width will be greater than when narrowing down to only the extreme values of gradient changes, but it is also possible to use all of these points to find a line that approximates the joint.

Furthermore, false detection judgments that take into account the uniformity of joint spacing do not need to be made. In particular, such judgments are excluded if the joint spacing is not uniform.

In addition, in the above embodiment, coordinate transformation was performed by the processing device 1, but it is also possible to obtain data that has already undergone coordinate transformation and then perform the process of identifying seams directly.

In addition, in the above embodiment, horizontal joints are approximated by cubic curves and vertical joints by straight lines (line segments), but this is not limited to this. If the distortion is not too great, horizontal joints may also be approximated by straight lines, and conversely, if the size of the bricks (structural materials) is large, vertical joints may also be approximated by curves.

In addition, in the above embodiment, we have described two types of seams, one along the x-axis and one along the y-axis. However, if there is only one type, there is no need to detect seams in the other direction. Furthermore, if there is a seam that does not necessarily extend parallel to the x-axis and y-axis, it is possible to determine whether a point is within the seam area using the square root of the sum of the squares of the two components or the sum of the absolute values of both components, rather than dividing the identification parameter into two components. In this case, the directions of the x-axis and y-axis may be set arbitrarily; for example, the z-axis may be set first, and then the x-axis and y-axis may be set arbitrarily in a plane perpendicular to the z-axis. Furthermore, there is no need to detect vertical and horizontal joints in order; all seams may be detected at once.

Furthermore, in the above embodiment, LoG was used to calculate the gradient change, but other methods, such as DoG (Difference of Gaussian Filter), may also be used.

In addition, while the above embodiment has been described as detecting joints in brick walls, this is not limited to this. It can be used in general to detect joints in structures made up of a combination of generally defined structural materials, such as concrete blocks (block walls, etc.) or tile structures. Furthermore, analysis can also be performed on structures that are not upright like walls, but are tilted or curved, such as road retaining walls or structures laid or placed on the ground, based on data obtained by photographing or surveying them from the air.

Furthermore, the above technology can be applied even if the structural members on both sides of the joint are not the same height. In other words, the structure may have a step at the joint.

Furthermore, if the joint is not straight, for example, if it is circular, the position may be determined using a three-dimensional Cartesian coordinate system such as a cylindrical coordinate system.

Furthermore, in the above embodiment, the horizontal joints were described as being continuous across the entire width of the brick wall, but this is not limited to this. If the structure clearly changes midway in the x direction, the image may be divided using that point as a boundary, but if this is not the case or if this is not possible, the horizontal joints may also be detected as line segments.

In particular, when the processing device 1 is used in conjunction with the analysis of structures with different joint thicknesses, the scale factor σ, the widths Th and Tw defining the local region, and the threshold value Tc may be changeable at any time. The changeable settings may be performed by automatically setting an appropriate value by specifying the joint width on the original measurement data, or the value itself may be input and set via the operation reception unit 16.

In addition, in the above embodiment, noise removal was performed using a Gaussian filter, but this is not limited to this. Alternatively, noise removal may not be necessary depending on the quality of the measurement points, etc.

In addition, in the above embodiment, the structural analysis processing was performed by a single processing device 1, but the processing may be divided among multiple processing devices. Furthermore, the measurement data may be acquired from a portable storage medium such as a DVD-ROM instead of the storage device 2.

In the above description, the storage unit 13 is described as an example of a computer-readable medium for storing the program 131 related to structural analysis control of the present invention, which is composed of a nonvolatile memory such as an HDD or flash memory. However, the present invention is not limited to this. Other computer-readable media may include other nonvolatile memories such as MRAM, and portable storage media such as CD-ROMs and DVD disks. Furthermore, a carrier wave may also be used as a medium for providing program data related to the present invention via a communication line.
In addition, the specific configurations, contents and procedures of the processing operations, etc. shown in the above embodiments can be modified as appropriate without departing from the spirit of the present invention. The scope of the present invention includes the scope of the invention described in the claims and its equivalents.

As described above, the processing device 1, which is a structure analysis device according to this embodiment, includes a CPU 11. The CPU 11 serves as an acquisition unit for detecting joints (vertical joints Jv, horizontal joints Jh) between bricks B on a surface including a flat surface of a structure (brick wall W) in which a plurality of structural materials (bricks B) having flat surfaces are assembled with the flat surfaces facing substantially in the same direction, the acquisition unit acquires point cloud data, which is a set of multiple points on the surface whose positions are specified and represented in a three-dimensional Cartesian coordinate system (Cartesian coordinate system) having three axes defined as an x direction substantially parallel to the surface, a y direction substantially parallel to the surface and perpendicular to the x direction, and a z direction perpendicular to the x and y directions, the calculation unit calculates, for each local region (each point r _i in this case) on the surface, at least one of diagonal components relating to the x direction and the y direction of an inverse matrix Q i of a covariance matrix C i of the three-dimensional coordinates of multiple points included in the local region among the multiple points, and the determination unit uses the calculated diagonal components to determine whether a seam is likely to exist in the local region (whether each point r _i is likely to exist within the seam).
According to the processing device 1, it is possible to determine with lower noise than before whether each point r _i is a point within a seam, and it is possible to exclude many points that are prone to misjudgment due to damage, deterioration, etc., so it is possible to detect seams with higher accuracy than before. Therefore, it is easier to accurately identify the structure even in structures that are heavily damaged or deteriorated, while significantly reducing the effort required by the user.

The structure (brick wall W) also has a first seam group consisting of a plurality of seams (joints) extending substantially parallel to one another. The CPU 11, as an acquisition unit, acquires point cloud data in which positions are represented in a three-dimensional orthogonal coordinate system (Cartesian coordinate system) in which the extension direction of the plurality of seams is substantially parallel to the x direction, as a calculation unit, calculates at least the diagonal component (identification parameter Qyy) relating to the y direction of the inverse matrix Qi, and as a determination unit, determines that a seam (joint) of the first seam group may be present in a local region (point) where the identification parameter Qyy is equal to or greater than a predetermined threshold Tc.
In this way, by previously aligning the coordinates of the acquired data with the direction of extension of the joint, it becomes possible to detect the joint using only one identification parameter, and the joint can be easily identified.

The structure (brick wall W) also has a second seam group consisting of seams extending substantially perpendicular to the extension direction of the seams of the first seam group. The CPU 11, as a determination means, determines that a seam of the second seam group may exist in a local region where a diagonal component in the x direction is equal to or greater than a predetermined threshold value Tc (that is, that point r _i may exist in a seam of the second seam group).
In this way, when joints extend in perpendicular directions, the coordinate axis direction roughly coincides with the joints, allowing for the separation of each identification parameter. Therefore, it is possible to more accurately and easily determine whether or not a particular joint corresponds to each other while suppressing the influence of other joints.

The CPU 11 also functions as a detection means, detecting extreme points related to gradient changes in the z-direction component for the z-direction components at multiple positions, and as a generation means, generating graphic data that approximates the seam using point ri, which is a representative point of a local area where it has been determined that a seam may exist, for the extreme points included in the local area. By combining seam determination using the identification parameters Qxx and Qyy with conventional technology, it is possible to more appropriately identify the center position of the seam while excluding points outside the seam, thereby enabling more accurate determination of an approximate line representing the seam.

The structural material is a brick B having a rectangular plane, the structure is a brick wall W in which bricks B are arranged in a certain direction (horizontal direction) and stacked in a direction perpendicular to that direction (vertical direction), and the joints are joints (horizontal joints Jh, vertical joints Jv) connecting the bricks B. The CPU 11, as an acquisition means, acquires data of a cloud of points in which the x direction is parallel to the arrangement direction of the bricks B and the y direction is parallel to the direction in which the bricks B are stacked; as a calculation means, calculates the diagonal components related to the x and y directions in the inverse matrix Qi; as a determination means, determines that a horizontal joint Jh may exist in a local region where the diagonal component related to the y direction is equal to or greater than a predetermined threshold Tc (point _ri may exist in the horizontal joint Jh); and determines that a vertical joint Jv may exist in a local region where the diagonal component related to the x direction is equal to or greater than a predetermined threshold Tc (point _ri may exist in the vertical joint Jv); and as a generation means, generates graphic data including a first graphic (line segment) that approximates the horizontal joint Jh using extreme points included in the local region where it is determined that a horizontal joint Jh may exist, and a second graphic (curve) that approximates the vertical joint Jv using extreme points included in the local region where it is determined that a vertical joint Jv may exist.
In this way, by using the above-mentioned identification parameters Qxx and Qyy to detect the vertical joints Jv and horizontal joints Jh of a brick wall W made of bricks B arranged vertically and horizontally, it is possible to reduce the influence of noise, particularly deterioration and damage, more than before and obtain lines that more accurately approximate these.
In addition, when the CPU 11 generates graphic data including a second figure as a generation means, it is preferable that the CPU 11 generates graphic data including the above-mentioned first figure and a second figure that approximates the vertical joint Jv using extreme points included in the local area where it is determined that the vertical joint Jv may exist, and that has connectivity with the first figure.
In this way, by generating a first figure relating to the horizontal joint Jh, which is long and easy to identify, and then generating as figure data only those second figures relating to the vertical joint Jv that have connectivity with the first figure, it is possible to effectively eliminate erroneous detection of the vertical joint Jv, which is shorter than the horizontal joint Jh and more likely to be erroneously detected.

Furthermore, the structural analysis method of this embodiment is a structural analysis method for detecting seams between components on a surface including a plane of a structure (brick wall W) in which a plurality of components (bricks B) having flat surfaces are combined with the planes facing in approximately the same direction, and includes the following steps: acquiring point cloud data, which is a collection of multiple points on the surface whose positions are specified and expressed in a three-dimensional orthogonal coordinate system (Cartesian coordinate system) with the three axes being an x-direction that is approximately parallel to the surface, a y-direction that is approximately parallel to the surface and perpendicular to the x-direction, and a z-direction that is perpendicular to the x- and y-directions; calculating, for each local region (point r _i ) on the surface, at least one of the diagonal components relating to the x- and y-directions of an inverse matrix Q i of a covariance matrix Ci of the three-dimensional coordinates of multiple points included in the local region among the multiple points; and determining whether a _seam may exist in the local region using the calculated diagonal components.
According to this structure analysis method, it is possible to determine with lower noise than before whether each point r _i is a point within a seam, and it is possible to exclude many points that are prone to misjudgment due to damage, deterioration, etc., making it possible to detect seams with higher accuracy than before. Therefore, it is easier to accurately identify the structure even in structures that are heavily damaged or deteriorated, while significantly reducing the effort required by the user.

By installing and running the program 131 relating to the structural analysis method on a computer, it becomes possible to execute the series of processes relating to the detection of the above-mentioned joints (joints) in software, making it possible to identify joints more accurately and easily using a conventional computer.

1 Processing device 2 Storage device 11 CPU
12 RAM
13 Memory unit 131 Program 14 Communication unit 15 Display unit 16 Operation reception unit 21 Measurement data 100 Structure analysis system B Brick Ci Covariance matrix Dc Upper limit value Dth Lower limit interval Ez Gradient change Jh Horizontal joint Jv Vertical joint Qi Inverse matrix Qxx, Qyy Identification parameter Tc Threshold value Th, Tw Width W Brick wall

Claims (7)

1. A structure analysis device for detecting joints between a plurality of members having flat surfaces on a surface including the flat surfaces of a structure in which the flat surfaces are combined with each other, the joints being oriented in approximately the same direction, the device comprising:
an acquisition means for acquiring point cloud data, which is a set of points on the surface, the positions of which are specified in a three-dimensional orthogonal coordinate system having three axes: a first direction substantially parallel to the surface, a second direction substantially parallel to the surface and perpendicular to the first direction, and a third direction perpendicular to the first and second directions;
a calculation means for calculating, for each local region of the surface, at least one of diagonal components relating to the first direction and the second direction in an inverse matrix of a covariance matrix of three-dimensional coordinates of a plurality of points included in the local region among the plurality of points;
a determination means for determining whether the seam is present in the local region using the calculated diagonal components;
A structural analysis device comprising: the structure has a first group of seams consisting of a plurality of the seams extending generally parallel to one another;
the acquisition means acquires data of the point cloud in which positions are represented in the three-dimensional orthogonal coordinate system in which the extension directions of the plurality of seams are substantially parallel to the first direction;
the calculation means calculates at least diagonal components of the inverse matrix in the second direction;
The structural analysis device according to claim 1, wherein the determination means determines that a seam of the first seam group may exist in the local region where a diagonal component in the second direction is equal to or greater than a predetermined first threshold value. the structure has a second group of seams extending generally perpendicular to the direction of extension of the seams of the first group of seams;
the calculation means calculates diagonal components of the inverse matrix relating to the first direction and the second direction;
The structural analysis device according to claim 2, wherein the determination means determines that a seam of the second seam group may exist in the local region where a diagonal component in the first direction is equal to or greater than a predetermined second threshold value. a detection means for detecting an extreme value point related to a change in the gradient of the third direction component at the positions of the plurality of points;
a generating means for generating graphic data approximating the seam by using the extreme value points included in the local region in which it has been determined by the determining means that the seam may exist;
4. The structure analysis device according to claim 1, further comprising: the member is a brick having a rectangular plane,
the structure is a wall in which the bricks are arranged in a certain direction and stacked in a direction perpendicular to the certain direction,
The joints are joints connecting the bricks,
the acquiring means acquires the data of the point cloud in which the first direction is parallel to the direction in which the bricks are arranged and the second direction is parallel to the direction in which the bricks are stacked,
the calculation means calculates diagonal components of the inverse matrix relating to the first direction and the second direction;
The determination means determines that a horizontal joint extending in the first direction may exist in the local region where a diagonal component related to the second direction is equal to or greater than a predetermined first threshold, and determines that a vertical joint extending in the second direction may exist in the local region where a diagonal component related to the first direction is equal to or greater than a predetermined second threshold,
The structural analysis device according to claim 4, characterized in that the generating means generates graphic data including a first graphic approximating the horizontal joint using the extreme value points included in the local region where it is determined that the horizontal joint may exist, and a second graphic approximating the vertical joint using the extreme value points included in the local region where it is determined that the vertical joint may exist. 1. A structure analysis method for detecting joints between a plurality of members having flat surfaces on a surface including the flat surfaces of a structure in which the flat surfaces are combined with each other, the joints being oriented in approximately the same direction, the method comprising:
an acquisition step of acquiring point cloud data, which is a set of points on the surface, the positions of which are specified in a three-dimensional orthogonal coordinate system having three axes: a first direction substantially parallel to the surface, a second direction substantially parallel to the surface and perpendicular to the first direction, and a third direction perpendicular to the first and second directions;
a calculation step of calculating, for each local region of the surface, at least one of diagonal components relating to the first direction and the second direction in an inverse matrix of a covariance matrix of three-dimensional coordinates of a plurality of points included in the local region among the plurality of points;
a determination step of determining whether the seam may exist in the local region using the calculated diagonal components;
A structural analysis method comprising: A program that causes a computer to execute a function of detecting joints between a plurality of members having flat surfaces on a surface including the flat surfaces of a structure in which the flat surfaces are combined with each other with the flat surfaces facing in approximately the same direction, the program comprising:
The functions include:
an acquisition function for acquiring point cloud data, which is a set of points on the surface, the positions of which are specified in a three-dimensional orthogonal coordinate system having three axes: a first direction substantially parallel to the surface, a second direction substantially parallel to the surface and perpendicular to the first direction, and a third direction perpendicular to the first and second directions;
a calculation function for calculating, for each local region of the surface, at least one of diagonal components relating to the first direction and the second direction in an inverse matrix of a covariance matrix of three-dimensional coordinates of a plurality of points included in the local region among the plurality of points;
a determination function for determining whether the seam may exist in the local region using the calculated diagonal components;
A program comprising: