A method, device and system for detecting the position of an inclined beam without purlins installed

Abstract

The present application discloses a method, device and system for detecting the position of an inclined beam without purlins installed, which belongs to a point cloud processing system. After acquiring the point cloud data of a photovoltaic power station, the point cloud data is projected on the YOZ plane and the XOZ plane respectively, and then a straight line fitting is performed on the projected point cloud. The fitting line represents the ground, so after removing the points whose distance to the fitting line is less than a preset distance from the point cloud data, the target point cloud data obtained does not contain ground points, which is convenient for subsequent processing without interference from ground points; then the target point cloud data is projected on the XOY plane, and a connected domain analysis is performed on the projected points to obtain a candidate rectangular frame, and then it is determined whether the candidate rectangular frame is the target inclined beam, and finally the position of the target inclined beam is obtained and output.

Description

Technical Field

[0001] The present application relates to the field of point cloud processing technology, and in particular, to a method, device and system for detecting the position of an inclined beam without purlins installed. Background Art

[0002] With the development of power modernization, the demand for photovoltaic power stations is increasing, which can not only reduce the excessive consumption of resources, but also alleviate environmental pollution, achieve energy conservation and emission reduction, and lay the foundation for intensive development. However, the progress monitoring and management of photovoltaic power station construction is still imperfect, and the position detection of the inclined beams without purlins installed in photovoltaic infrastructure is still mainly based on traditional manual detection. This traditional method is not only inefficient and labor-intensive, but also leads to missed detection and false detection. Summary of the invention

[0003] In order to overcome the shortcomings of the prior art, the present application provides a method, device and system for detecting the position of an inclined beam without purlins installed, so as to solve the problem that the position detection of the inclined beam without purlins installed in photovoltaic infrastructure is still mainly based on traditional manual detection. This traditional method is not only inefficient and labor-intensive, but also leads to missed detection and false detection.

[0004] The technical solution adopted by this application to solve its technical problem is:

[0005] In a first aspect, a method for detecting the position of an inclined beam without installing purlins is provided, comprising:

[0006] Obtain point cloud data of photovoltaic power stations;

[0007] Projecting the point cloud data on the YOZ plane and the XOZ plane respectively, performing straight line fitting on the projected point cloud, removing points whose distance to the fitting straight line is less than a preset distance from the point cloud data, and obtaining target point cloud data;

[0008] Project the target point cloud data on the XOY plane, and perform connected domain analysis on the projected points to obtain a candidate rectangular frame;

[0009] Determine whether the candidate rectangular frame is a target inclined beam without purlins installed;

[0010] If yes, the target inclined beam position is obtained and output.

[0011] Further, the point cloud data is projected on the YOZ plane and the XOZ plane respectively, a straight line fitting is performed on the projected point cloud, and points whose distance to the fitting straight line is less than a preset distance are removed from the point cloud data to obtain the target point cloud data, including:

[0012] trimming the point cloud data into parts of a preset size;

[0013] For each portion of the point cloud data, increase its horizontal scale to a first preset horizontal scale;

[0014] Project the XOZ plane and YOZ plane respectively to obtain the first projection point cloud;

[0015] Obtain the target area with the largest number of point clouds in the first projection point cloud;

[0016] Performing straight line fitting on the point cloud of the target area;

[0017] Removing points in the first projected point cloud whose distance to the fitting straight line is less than a preset distance;

[0018] The removed point cloud is statistically filtered and the horizontal scale is restored to obtain the target point cloud data.

[0019] Furthermore, the target point cloud data is projected on the XOY plane, and a connected domain analysis is performed on the projected points to obtain a candidate rectangular frame, including:

[0020] Transforming the horizontal scale of the target point cloud data into a second preset horizontal scale;

[0021] Projecting the transformed target point cloud data on the XOY plane to obtain a second projected point cloud;

[0022] dilating the image of the second projected point cloud;

[0023] A connected domain analysis is performed based on the dilated image, and the rectangular box fitted by the connected domain is used as a candidate rectangular box.

[0024] Further, the determining whether the candidate rectangular frame is a target inclined beam without purlins installed includes:

[0025] Calculate the corner point coordinates of the vertices of the four corners of the candidate rectangular frame;

[0026] Calculate the length and width of the candidate rectangular frame according to the corner point coordinates;

[0027] If the length and width of the candidate rectangular frame are respectively within the preset ranges, the candidate rectangular frame is determined to be a target inclined beam without purlins installed.

[0028] Furthermore, the calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame includes:

[0029] Obtain the coordinates of one corner of the circumscribed rectangle of the candidate rectangular frame and the length and width of the circumscribed rectangle, where the length of the circumscribed rectangle is parallel to the y-axis and the width of the circumscribed rectangle is parallel to the x-axis;

[0030] Determine the rectangle type of the candidate rectangular frame, where the rectangle type includes two types: a regular rectangle and a non-regular rectangle;

[0031] The corner point coordinates of the vertices of the four corners of the candidate rectangular frame are calculated according to the rectangle type, the coordinates of one of the corners of the circumscribed rectangle, and the length and width of the circumscribed rectangle.

[0032] Further, the calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame according to the rectangle type and the coordinates of one corner of the circumscribed rectangle and the length and width of the circumscribed rectangle includes:

[0033] When the rectangle type is a regular rectangle, the vertex coordinates of the four corners of the circumscribed rectangular frame are obtained according to the coordinates of one corner of the circumscribed rectangle of the circumscribed rectangular frame, and the length and width of the circumscribed rectangle;

[0034] The vertex coordinates of the four corners of the circumscribed rectangular frame are used as the corner point coordinates of the vertices where the four corners of the candidate rectangular frame are located.

[0035] Further, the calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame according to the rectangle type and the coordinates of one corner of the circumscribed rectangle and the length and width of the circumscribed rectangle includes:

[0036] When the rectangle type is a non-normal rectangle, obtaining the centroid coordinates of the candidate rectangular frame;

[0037] Find the ordinate y1 of the first corner point, where the coordinate of one of the corners of the circumscribed rectangle is (x1, y4), find all the y values ​​of the candidate rectangular box at x=x1 by indexing, and take the maximum value of all y values ​​as the ordinate y1 of the first corner point; the abscissa of the first corner point is x1;

[0038] Find the abscissa x4 of the fourth corner point, find all x values ​​of the candidate rectangular box at y=y4 by index, and take the maximum value of all x values ​​as the abscissa x4 of the fourth corner point; the ordinate of the fourth corner point is y4;

[0039] The longitudinal coordinate of the second corner point and the transverse coordinate of the third corner point are obtained based on the length and width of the circumscribed rectangle respectively. The transverse coordinate of the second corner point is calculated based on the transverse coordinate of the centroid of the candidate rectangular box and the transverse coordinate of the fourth corner point. The longitudinal coordinate of the third corner point is calculated based on the longitudinal coordinate of the centroid of the candidate rectangular box and the longitudinal coordinate of the first corner point.

[0040] Further, the obtaining and outputting of the target inclined beam position includes:

[0041] Calculate the intersection and union ratio of the rectangular box where each first target inclined beam is located and the rectangular box of any other second target inclined beam;

[0042] If the intersection-and-union ratio is less than or equal to a threshold, the position of the first target inclined beam is output;

[0043] If the intersection-and-union ratio is greater than the threshold, the areas of the rectangular box of the first target oblique beam and the rectangular box of the repeated second target oblique beam are compared, and only the target oblique beam with the largest area is retained and its position is output. The repeated second target oblique beam is the second target oblique beam whose intersection-and-union ratio with the rectangular box of the first target oblique beam is greater than the threshold.

[0044] In a second aspect, a device for detecting the position of an inclined beam without installing a purlin is provided, comprising:

[0045] Point cloud data acquisition module, used to acquire point cloud data of photovoltaic power stations;

[0046] A ground point removal module is used to project the point cloud data on the YOZ plane and the XOZ plane respectively, perform straight line fitting on the projected point cloud, and remove points whose distance to the fitting straight line is less than a preset distance from the point cloud data to obtain target point cloud data;

[0047] A rectangular region acquisition module is used to project the target point cloud data on the XOY plane and perform a connected domain analysis on the projected points to obtain a candidate rectangular frame;

[0048] A target inclined beam judging module, used to judge whether the candidate rectangular frame is a target inclined beam without purlins installed;

[0049] The inclined beam position output module is used to obtain and output the target inclined beam position if so.

[0050] In a third aspect, a system for detecting the position of an inclined beam without installing purlins is provided, comprising:

[0051] at least one processor and at least one memory;

[0052] The memory stores executable instructions of the processor;

[0053] The processor is configured to execute the above-mentioned method for detecting the position of an inclined beam without installing purlins.

[0054] Beneficial effects:

[0055] The technical solution of the present application provides a method, device and system for detecting the position of an inclined beam without purlins installed. After acquiring the point cloud data of the photovoltaic power station, the point cloud data is projected on the YOZ plane and the XOZ plane respectively, and then the projected point cloud is fitted with a straight line. The fitting straight line represents the ground, so the points whose distance to the fitting straight line is less than the preset distance are removed from the point cloud data. The target point cloud data obtained does not contain ground points, which is convenient for subsequent processing without interference from ground points; then the target point cloud data is projected on the XOY plane, and the connected domain analysis is performed on the projected points to obtain a candidate rectangular frame, and then it is determined whether the candidate rectangular frame is the target inclined beam, and finally the position of the target inclined beam is obtained and output. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 It is a flow chart of a method for detecting the position of an inclined beam without purlins installed provided in an embodiment of the present application;

[0058] Figure 2 It is a flow chart of a specific method for detecting the position of an inclined beam without purlins installed provided in an embodiment of the present application;

[0059] Figure 3 is a schematic diagram of a cropped point cloud block provided in an embodiment of the present application;

[0060] Figure 4 is a ground point removal flow chart provided in an embodiment of the present application;

[0061] Figure 5 is a schematic diagram of a first projection point cloud on a YOZ plane provided in an embodiment of the present application;

[0062] Figure 6 is a schematic diagram of a first projection point cloud on the XOZ plane provided in an embodiment of the present application;

[0063] Figure 7 This is a method provided by the embodiment of the present application. Figure 5 The schematic diagram of the projected fitting straight line is shown;

[0064] Figure 8 This is a method provided by the embodiment of the present application. Figure 6 The schematic diagram of the projected fitting straight line is shown;

[0065] Fig. 9is a schematic diagram of target point cloud data provided by an embodiment of the present application;

[0066] Fig.10 is a schematic diagram of a second projection point cloud provided in an embodiment of the present application;

[0067] Fig.11 is a schematic diagram of a projected image after expansion provided by an embodiment of the present application;

[0068] Fig.12 This is a flowchart of calculating the coordinates of a rectangular corner point provided in an embodiment of the present application;

[0069] Fig.13 It is a non-positive rectangular frame connected domain image provided by an embodiment of the present application;

[0070] Fig.14 is a dilated point image provided by an embodiment of the present application;

[0071] Fig.15 It is a non-regular rectangular corner point diagram provided in an embodiment of the present application;

[0072] Fig.16 This is a schematic diagram of a non-deduplication detection result provided in an embodiment of the present application;

[0073] Fig.17 This is a schematic diagram of the detection results after deduplication provided in an embodiment of the present application. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application is described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other implementation methods obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0075] Reference Figure 1 , an embodiment of the present application provides a method for detecting the position of an inclined beam without installing purlins, comprising:

[0076] S11: Obtaining point cloud data of a photovoltaic power station; This application uses a drone to collect point cloud data of a photovoltaic power station site. Since the collected point cloud data includes not only the inclined beams but also other objects, especially the point cloud data of the inclined beams with installed purlins or photovoltaic panels and the inclined beams without installed purlins are different, they cannot be processed directly.

[0077] S12: Project the point cloud data on the YOZ plane and the XOZ plane respectively, perform straight line fitting on the projected point cloud, remove points whose distance to the fitting straight line is less than a preset distance from the point cloud data, and obtain target point cloud data.

[0078] Preferably, the point cloud data is cut into parts of preset size; because the overall point cloud data is large, it is inconvenient to process directly and the efficiency is low. The point cloud data of each part is larger than the point cloud data of a single inclined beam. In addition, in order to ensure that there is no omission, it is necessary to ensure that adjacent point cloud data overlap. Generally, the length of the overlapping part is the point cloud length of the inclined beam.

[0079] For each part of the point cloud data, its horizontal scale is increased to the first preset horizontal scale; the point clouds actually collected are relatively dense, and if directly projected, the point clouds cannot be distinguished, so it is necessary to increase the distance between the point clouds year on year.

[0080] The first projection point cloud is obtained by projecting the XOZ plane and the YOZ plane respectively; thus, the first projection point cloud includes both the ground points and the point cloud of the inclined beam part.

[0081] Obtain the target area with the largest number of point clouds in the first projection point cloud; because the ground points are densest in the first projection point cloud, select the area with a point cloud number greater than the preset point cloud number for straight line fitting. Exemplarily, a rectangular box of a preset size is moved in the first projection point cloud, and the area with the largest number of point clouds is taken as the target area. The point cloud of the target area is fitted with a straight line. The obtained fitting line represents the ground.

[0082] The points in the first projected point cloud whose distance to the fitting straight line is less than a preset distance are removed; that is, the ground points are removed.

[0083] The removed point cloud is statistically filtered and the horizontal scale is restored to obtain the target point cloud data.

[0084] S13: Projecting the target point cloud data on the XOY plane, and performing connected domain analysis on the projected points to obtain a candidate rectangular frame.

[0085] As a preferred implementation method of the embodiment of the present application, the horizontal scale of the target point cloud data is transformed into a second preset horizontal scale; the first preset horizontal scale and the second preset horizontal scale can be equal to or different from each other, and can be set according to actual needs.

[0086] The changed target point cloud data is projected on the XOY plane to obtain the second projection point cloud; after the ground points are removed, the point cloud in the second projection point cloud is the point cloud of the inclined beam, or the point cloud of the inclined beam with the purlin or photovoltaic panel installed. Since the two point clouds are different, they can be distinguished.

[0087] The image of the second projected point cloud is expanded; a connected domain analysis is performed based on the expanded image, and a rectangular frame after the connected domain fitting is used as a candidate rectangular frame. The candidate rectangular frames at this time include three types: an inclined beam without purlins installed, an inclined beam with purlins or photovoltaic panels installed, and other objects projected as rectangles.

[0088] S14: Determine whether the candidate rectangular frame is a target inclined beam without purlins installed;

[0089] Specifically, the corner point coordinates of the vertices of the four corners of the candidate rectangular frame are calculated; the length and width of the candidate rectangular frame are calculated according to the corner point coordinates; if the length and width of the candidate rectangular frame are respectively within the preset range, the candidate rectangular frame is judged to be a target oblique beam without purlins installed. The length and width of the candidate rectangular frame corresponding to the oblique beam with purlins or photovoltaic panels installed are larger than the length and width of the candidate rectangular frame corresponding to the oblique beam without purlins installed, and other objects projected as rectangles can be removed according to the position later even if the length and width of the candidate rectangular frame corresponding to them are also within the preset range.

[0090] The step of calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame includes:

[0091] Obtain the coordinates of one corner of the circumscribed rectangle of the candidate rectangular frame and the length and width of the circumscribed rectangle, where the length of the circumscribed rectangle is parallel to the y-axis and the width of the circumscribed rectangle is parallel to the x-axis;

[0092] Determine the rectangle type of the candidate rectangular box, the rectangle type includes two types: a normal rectangle and a non-normal rectangle; wherein a normal rectangle refers to a side of the candidate rectangular box that is perpendicular or parallel to the x-axis. All rectangles except the normal rectangle are non-normal rectangles. In one embodiment, the rectangle type of the candidate rectangular box can be determined based on the ratio of the length and width of the circumscribed rectangle, and the ratio greater than the threshold is generally a normal rectangle. In another embodiment, the expression of the straight line on which one side of the candidate rectangular box lies can be calculated, and if it is y=a, or x=b, it is judged to be a normal rectangle, and a and b are constants.

[0093] The corner point coordinates of the vertices of the four corners of the candidate rectangular frame are calculated according to the rectangle type, the coordinates of one of the corners of the circumscribed rectangle, and the length and width of the circumscribed rectangle.

[0094] When the rectangle type is a regular rectangle, the vertex coordinates of the four corners of the circumscribed rectangular box are obtained according to the coordinates of one of the corners of the circumscribed rectangle of the circumscribed rectangular box, and the length and width of the circumscribed rectangle; the vertex coordinates of the four corners of the circumscribed rectangular box are used as the corner point coordinates of the vertices of the four corners of the candidate rectangular box.

[0095] When the rectangle type is a non-normal rectangle, obtain the coordinates of the centroid of the candidate rectangular box; calculate the ordinate y1 of the first corner point, wherein the coordinates of one of the corners of the circumscribed rectangle are (x1, y4), and find all the y values ​​of the candidate rectangular box at x=x1 through indexing, and take the maximum value of all the y values ​​as the ordinate y1 of the first corner point; the abscissa of the first corner point is x1; calculate the abscissa x4 of the fourth corner point, and find all the x values ​​of the candidate rectangular box at y=y4 through indexing, and take the maximum value of all the x values ​​as the abscissa x4 of the fourth corner point; the ordinate of the fourth corner point is y4; obtain the ordinate of the second corner point and the abscissa of the third corner point based on the length and width of the circumscribed rectangle, and calculate the abscissa of the second corner point based on the abscissa of the centroid of the candidate rectangular box and the abscissa of the fourth corner point, and calculate the ordinate of the third corner point based on the ordinate of the centroid of the candidate rectangular box and the ordinate of the first corner point.

[0096] S15: If yes, the target inclined beam position is obtained and output.

[0097] Since there are overlapping areas during clipping, deduplication is required at this time. Specifically, the intersection and union ratio of the rectangular box where each first target inclined beam is located and the rectangular box of any other second target inclined beam is calculated;

[0098] If the intersection-and-union ratio is less than or equal to a threshold, the position of the first target inclined beam is output;

[0099] If the intersection-and-union ratio is greater than the threshold, the areas of the rectangular box of the first target oblique beam and the rectangular box of the repeated second target oblique beam are compared, and only the target oblique beam with the largest area is retained and its position is output. The repeated second target oblique beam is the second target oblique beam whose intersection-and-union ratio with the rectangular box of the first target oblique beam is greater than the threshold.

[0100] In order to explain the present application scheme in more detail, a specific method for detecting the position of an inclined beam without installing purlins is provided below.

[0101] During the infrastructure construction phase of a photovoltaic power station, the method of an embodiment of the present application is based on the on-site point cloud data collected by a drone. By performing structured processing on the point cloud data, the inclined beam portion where the purlins are not installed is segmented, and finally the coordinate information of the four corner points of the inclined beam is obtained to realize the position detection of the inclined beam where the purlins are not installed in the photovoltaic infrastructure.

[0102] This method mainly includes six steps: point cloud clipping, ground point removal, rectangle fitting, rectangle corner point coordinate calculation, inclined beam screening and inclined beam deduplication. Figure 2As shown, first, the point cloud is scaled to a fixed size through point cloud preprocessing, and then the entire point cloud is overlapped and cropped into multiple small blocks according to the fixed size and step length of the ground plane. After that, by traversing each point cloud, the YOZ plane and XOZ plane are projected respectively, and the ground points are fitted with a straight line, and the ground is removed according to the distance from the point to the straight line. Then, the point cloud after removing the ground is projected on the XOY plane, and the connected domain analysis is performed on the projected plane points to extract the rectangular area. Then the coordinates of the four corner points of the extracted rectangular frame are calculated. In addition, the position coordinates of the inclined beam are screened out according to the width and length of the inclined beam. Then, the iou (the ratio of the intersection and union of two target detection frames) of the position of the rectangular frame of the inclined beam and the position of the rectangular frame of other inclined beams is judged to remove duplicates. Finally, the position information of the four corner points of each inclined beam is output.

[0103] The details are as follows:

[0104] 1. Point cloud clipping: according to the size of the photovoltaic beam, the point cloud of the photovoltaic power station is clipped into multiple small point clouds on the XOY plane to facilitate the removal of the ground point cloud. The specific process is as follows:

[0105] Step 1: Preprocess the point cloud of the entire photovoltaic infrastructure site, scale the point cloud of the photovoltaic infrastructure site to a fixed scale size according to a certain ratio, and perform voxel downsampling according to a certain threshold.

[0106] Step 2: Calculate the maximum (ymax, xmax) and minimum (ymin, xmin) values ​​of the pre-processed photovoltaic power station cloud along the Y direction (Y axis direction) and X direction (X axis direction). The length of the oblique beam point cloud under the scale of this system is 3m, and the length and width of the cropped point cloud cube are 5m. In order to ensure that the oblique beam can be fully detected, the overlapping area of ​​each cropped point cloud area along the X and Y directions is 3m, so the cropping step should be set to 2m.

[0107] Step 3: Cut along the Y direction. Formula (1-1) is the starting point of each cutting segment, and step is the step length. When it is greater than step, the clipping is terminated.

[0108] (1-1)

[0109] Step 4: Crop each segment of the point cloud cropped along the Y direction along the X direction. Formula (1-2) is the starting point of each segment cropping, step is the step length, and xmax is the maximum value of the point cloud along the X direction. When it is greater than step, the cropping is terminated, and the cropped point cloud block is as follows Figure 3 shown.

[0110] (1-2)

[0111] 2. Ground removal: traverse all the cropped point clouds and remove the ground point clouds from each cropped point cloud. The removed point clouds are the oblique beam point clouds and the remaining obstacle point clouds except the ground. The specific steps and flow chart are as follows: Figure 4 As shown:

[0112] Step 1: In order to ensure that the point cloud can be projected onto the plane more completely, it is necessary to increase the scale of the cropped point cloud block. The original horizontal scale of the cropped point cloud block is 5×5, and the increased horizontal scale becomes 400×400.

[0113] Step 2: The point cloud block is projected on the XOZ plane and the YOZ plane respectively. The coordinate point after projection is still the coordinate point in the geographic coordinate system. The geographic coordinate system point is converted to the image coordinate system point, as shown in formula (1-3), where (u, v) is the point coordinate in the image coordinate system, point[x] and point[y] are the point coordinates in the geographic coordinate system, w and h are the length values ​​along the X and Y directions after the point cloud is projected, max_x and max_y are the maximum values ​​along the X and Y directions after projection, and the first projected point cloud images after projection are as follows: Figure 5 and Figure 6 shown.

[0114] (1-3)

[0115] Step 3: Ground point straight line fitting, fitting the ground straight line based on the projected point image First, fit the projection on the YOZ plane with a straight line. k and b are the slope and intercept of the fitted straight line. If k and b are both 0, it means that the projected ground points on the YOZ plane are relatively scattered. Then continue to fit the ground straight line on the projection on the XOZ plane. Figure 7 and Figure 8 They are the ground lines fitted by the projection of the YOZ plane and the XOZ plane respectively. The fitted line is the parameter of the polar coordinate system. According to the polar coordinate equation, the coordinates (u1, v1) and (u2, v2) of the two points in the linear coordinate system can be obtained. The two coordinate points in the rectangular coordinate system are restored to the geographic coordinate system, as shown in formula (1-4), where (u, v) and (point_x, point_y) are the coordinates in the image coordinate system and the geographic coordinate system respectively, w and h are the length values ​​along the X and Y directions after the point cloud is projected, and max_x and max_y are the maximum values ​​along the X and Y directions after projection. Finally, k and b of the ground line are calculated by the coordinates of the two points in the geographic coordinate system.

[0116] (1-4)

[0117] Step 4: Calculate the distance from the point to the straight line after projection of the point cloud based on the fitted straight line. If the distance is less than the threshold, it means that the point is a ground point, otherwise it is a beam point or other obstacle point. Finally, perform statistical filtering on the retained beam points and restore the scale. The target point cloud data after removing the ground and filtering is as follows: Fig. 9 shown.

[0118] 3. Rectangle fitting: traverse all point clouds after removing the ground, project the point clouds after removing the ground along the Z direction, perform connected domain analysis on the projected area points, and extract rectangle information. The specific steps are as follows:

[0119] Step 1: Scale the input point cloud. The scale after the scale transformation is 500×500. Project the point cloud onto the XOY plane and perform image point coordinate transformation according to equation (1-3). The transformed second projected point cloud image is as follows: Fig.10 shown.

[0120] Step 2: dilate the projection point image, the rectangular dilation kernel size is 5×5, and the image after 6 iterations is as follows Fig.11 shown.

[0121] Step 3: Perform connected domain analysis based on the expanded image. After the connected domain is fitted, the coordinates of the upper left corner of the circumscribed rectangular box, the length and width of the rectangular box, the center of mass of the rectangular box, and labels can be output. Labels is a matrix of the same size as the image. Each connected area has a unique identification value. The background identification value is 0, and the connected area identification value starts from 1.

[0122] 4. Calculate the coordinates of the rectangle corner points. Calculate the coordinates of the four corner points of the rectangle based on the results of rectangle fitting. The specific flow chart and steps are as follows: Fig.12 As shown:

[0123] Step 1: Based on the length and width of the rectangular box fitted by the connected domain, determine whether the inclined beam rectangle is positive. If the length-to-width ratio of the rectangular box is greater than the threshold, the inclined beam is facing due north-south or due east-west. Otherwise, it means that the inclined beam is non-positive or other obstacle rectangle.

[0124] Step 2: 1. If the aspect ratio is greater than the threshold, and the inclined beam is assumed to be positive, then perform the following steps:

[0125] First, based on the information output by the connected domain, the coordinates of the upper left corner of the rectangular box (x1, y1) and the length and width of the box h and w are obtained. The coordinates of the four corner points are calculated based on the coordinates and length and width values, as shown in formula (1-5). (x2, y2) is the coordinate of the lower left corner of the rectangular box, (x3, y3) is the coordinate of the lower right corner of the rectangular box, and (x4, y4) is the coordinate of the upper right corner of the rectangular box. Finally, the coordinates of the corner points of the rectangular box in the image coordinate system are restored to the geographic coordinate system according to formula (1-4), and the coordinates of the four corner points bbox are output.

[0126] (1-5)

[0127] 2. If the aspect ratio is less than the threshold, the inclined beam is non-positive and the following steps are performed:

[0128] The coordinate information of the connected domain output is as follows Fig.13 As shown, (x, y) is the coordinate of the upper left corner of the circumscribed rectangle of the rectangular frame, p is the centroid of the rectangular frame, and this system sets the leftmost point as the coordinate of the first corner point of the inclined beam. The coordinate order of the remaining corner points is calibrated counterclockwise as shown in the figure, where x1=x, x3=x+w, y2=y+h, and y4=y.

[0129] This embodiment obtains the remaining values ​​of the coordinates of the four corner points of the inclined beam based on the labels matrix output by the connected domain. Fig.14 is the expanded point image, the labels matrix is ​​a rectangle of the same size as the image, each connected area has a unique identifier i, which starts from 0, and its corner point value is found by indexing the labels matrix and finding the maximum value. The specific steps are as follows:

[0130] (1) Find the ordinate y1 of the first corner point. The identification value of the i-th rectangle is i. Use the index to find all the y values ​​with the identification value i at x = x1. Take the maximum value of all the y values ​​as the ordinate y1 of the first corner point.

[0131] (2) Find the horizontal coordinate x4 of the fourth corner point. Use the index to find all x values ​​whose identification value is i at y = y4, and take the maximum value of all x values ​​as the horizontal coordinate x4 of the fourth corner point.

[0132] (3) Find the horizontal coordinate x2 of the second corner point, through the horizontal coordinate x4 of the fourth corner point and the horizontal coordinate x of the centroid p p Calculated, similarly, the ordinate y3 of the third corner point is calculated from the ordinate y1 of the first corner point and the ordinate y of the centroid p p The specific formula is shown in formula (1-6). Fig.15 A visualization of the corner points of a non-regular rectangle.

[0133] (1-6)

[0134] (4) Finally, according to formula (1-4), the corner point coordinates are restored to the geographic coordinate system, and the four corner point coordinate bboxes are output.

[0135] 5. Slanted beam screening: judge whether the rectangular area is a slanted beam based on the size of the slanted beam on the horizontal plane. Calculate the length and width of each rectangular box using the corner point coordinates. When the length and width of the rectangular box are within the preset range, the rectangular box is considered to be a slanted beam, otherwise the rectangular box is other obstacles.

[0136] 6. Deduplication of inclined beams. Since the point cloud clipping contains overlapping areas, the same inclined beam will be detected by multiple clipped point clouds, such as Fig.16 As shown, it is necessary to remove the weight of the inclined beam. The specific process is as follows:

[0137] Step 1: Calculate the IOU (intersection-over-union) of each rectangular box in the bbox with other rectangular boxes. If the IOU is greater than the threshold, it means that the two rectangular boxes detect the same inclined beam, and retain the index of the rectangular box surrounding the inclined beam with a larger area.

[0138] Step 2: Filter the coordinates of the corner points of the inclined beam in the bbox by index. The detection results after deduplication are as follows: Fig.17 As shown in the figure, the red framed area is the inclined beam area.

[0139] Based on the same inventive concept, the present application also provides a device for detecting the position of an inclined beam without installing purlins, comprising:

[0140] The point cloud data acquisition module is used to acquire the point cloud data of the photovoltaic power station.

[0141] A ground point removal module is used to project the point cloud data on the YOZ plane and the XOZ plane respectively, perform straight line fitting on the projected point cloud, and remove points whose distance to the fitting straight line is less than a preset distance from the point cloud data to obtain target point cloud data;

[0142] Specifically, the point cloud data is trimmed into parts of preset sizes; for each part of the point cloud data, its horizontal scale is increased to a first preset horizontal scale; the XOZ plane and the YOZ plane are projected respectively to obtain a first projected point cloud; a target area with the largest number of point clouds in the first projected point cloud is obtained; a straight line is fitted to the point cloud in the target area; points in the first projected point cloud whose distance to the fitting straight line is less than a preset distance are removed; the removed point cloud is statistically filtered and the horizontal scale is restored to obtain target point cloud data.

[0143] The rectangular area acquisition module is used to project the target point cloud data on the XOY plane and perform a connected domain analysis on the projected points to obtain a candidate rectangular frame; specifically, the horizontal scale of the target point cloud data is transformed into a second preset horizontal scale; the transformed target point cloud data is projected on the XOY plane to obtain a second projected point cloud; the image of the second projected point cloud is expanded; the connected domain analysis is performed based on the expanded image, and the rectangular frame after the connected domain is fitted is used as the candidate rectangular frame.

[0144] The target inclined beam judgment module is used to judge whether the candidate rectangular box is a target inclined beam without purlins installed; specifically, calculate the corner point coordinates of the vertices of the four corners of the candidate rectangular box; calculate the length and width of the candidate rectangular box according to the corner point coordinates; if the length and width of the candidate rectangular box are respectively within the preset range, then judge that the candidate rectangular box is a target inclined beam without purlins installed.

[0145] Wherein, calculating the corner point coordinates of the vertices where the four corners of the candidate rectangular box are located includes: obtaining the coordinates of one of the corners of the circumscribed rectangle of the candidate rectangular box and the length and width of the circumscribed rectangle, wherein the length of the circumscribed rectangle is parallel to the y-axis, and the width of the circumscribed rectangle is parallel to the x-axis; determining the rectangle type of the candidate rectangular box, wherein the rectangle type includes two types: a regular rectangle and a non-regular rectangle; and calculating the corner point coordinates of the vertices where the four corners of the candidate rectangular box are located according to the rectangle type and the coordinates of one of the corners of the circumscribed rectangle and the length and width of the circumscribed rectangle.

[0146] When the rectangle type is a regular rectangle, the vertex coordinates of the four corners of the circumscribed rectangular box are obtained according to the coordinates of one of the corners of the circumscribed rectangle of the circumscribed rectangular box, and the length and width of the circumscribed rectangle; the vertex coordinates of the four corners of the circumscribed rectangular box are used as the corner point coordinates of the vertices of the four corners of the candidate rectangular box.

[0147] When the rectangle type is a non-normal rectangle, obtain the coordinates of the centroid of the candidate rectangular box; calculate the ordinate y1 of the first corner point, wherein the coordinates of one of the corners of the circumscribed rectangle are (x1, y4), and find all the y values ​​of the candidate rectangular box at x=x1 through indexing, and take the maximum value of all the y values ​​as the ordinate y1 of the first corner point; the abscissa of the first corner point is x1; calculate the abscissa x4 of the fourth corner point, and find all the x values ​​of the candidate rectangular box at y=y4 through indexing, and take the maximum value of all the x values ​​as the abscissa x4 of the fourth corner point; the ordinate of the fourth corner point is y4; obtain the ordinate of the second corner point and the abscissa of the third corner point based on the length and width of the circumscribed rectangle, and calculate the abscissa of the second corner point based on the abscissa of the centroid of the candidate rectangular box and the abscissa of the fourth corner point, and calculate the ordinate of the third corner point based on the ordinate of the centroid of the candidate rectangular box and the ordinate of the first corner point.

[0148] The inclined beam position output module is used to obtain and output the target inclined beam position if so.

[0149] Specifically, the intersection and union ratio of the rectangular box where each first target oblique beam is located and the rectangular box of any other second target oblique beam is calculated; if the intersection and union ratio is less than or equal to the threshold, the position of the first target oblique beam is output; if the intersection and union ratio is greater than the threshold, the area of ​​the rectangular box of the first target oblique beam and the rectangular box of the repeated second target oblique beam are compared, and only the target oblique beam with the largest area is retained and its position is output. The repeated second target oblique beam is the second target oblique beam whose intersection and union ratio with the rectangular box of the first target oblique beam is greater than the threshold.

[0150] Based on the same inventive concept, the present application also provides a system for detecting the position of an inclined beam without installing purlins, comprising:

[0151] at least one processor and at least one memory;

[0152] The memory stores executable instructions of the processor;

[0153] The processor is configured to execute the method for detecting the position of an inclined beam without installing purlins provided in the above embodiment.

[0154] It should be noted that, in the description of this application, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of this application, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0155] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

Claims

1. A method for detecting the position of an inclined beam without installing purlins, characterized in that: include: Obtain point cloud data of photovoltaic power stations; Projecting the point cloud data on the YOZ plane and the XOZ plane respectively, performing straight line fitting on the projected point cloud, removing points whose distance to the fitting straight line is less than a preset distance from the point cloud data, and obtaining target point cloud data; Project the target point cloud data on the XOY plane, and perform connected domain analysis on the projected points to obtain a candidate rectangular frame; Determine whether the candidate rectangular frame is a target inclined beam without purlins installed; If yes, obtain the target inclined beam position and output it; The step of obtaining and outputting the target inclined beam position includes: Calculate the intersection and union ratio of the rectangular box where each first target inclined beam is located and the rectangular box of any other second target inclined beam; If the intersection-and-union ratio is less than or equal to a threshold, the position of the first target inclined beam is output; If the intersection-and-union ratio is greater than the threshold, the areas of the rectangular box of the first target oblique beam and the rectangular box of the repeated second target oblique beam are compared, and only the target oblique beam with the largest area is retained and its position is output. The repeated second target oblique beam is the second target oblique beam whose intersection-and-union ratio with the rectangular box of the first target oblique beam is greater than the threshold.

2. The method according to claim 1, characterized in that: The point cloud data is projected on the YOZ plane and the XOZ plane respectively, a straight line fitting is performed on the projected point cloud, and points whose distance to the fitting straight line is less than a preset distance are removed from the point cloud data to obtain target point cloud data, including: trimming the point cloud data into parts of a preset size; For each portion of the point cloud data, increase its horizontal scale to a first preset horizontal scale; Project the XOZ plane and YOZ plane respectively to obtain the first projection point cloud; Obtain the target area with the largest number of point clouds in the first projection point cloud; Performing straight line fitting on the point cloud of the target area; Removing points in the first projected point cloud whose distance to the fitting straight line is less than a preset distance; The removed point cloud is statistically filtered and the horizontal scale is restored to obtain the target point cloud data.

3. The method according to claim 1, characterized in that: The target point cloud data is projected on the XOY plane, and a connected domain analysis is performed on the projected points to obtain a candidate rectangular frame, including: Transforming the horizontal scale of the target point cloud data into a second preset horizontal scale; Projecting the transformed target point cloud data on the XOY plane to obtain a second projected point cloud; dilating the image of the second projected point cloud; A connected domain analysis is performed based on the dilated image, and the rectangular box fitted by the connected domain is used as a candidate rectangular box.

4. The method according to claim 1, characterized in that: The determining whether the candidate rectangular frame is a target inclined beam without purlins installed includes: Calculate the corner point coordinates of the vertices of the four corners of the candidate rectangular frame; Calculate the length and width of the candidate rectangular frame according to the corner point coordinates; If the length and width of the candidate rectangular frame are respectively within the preset ranges, the candidate rectangular frame is determined to be a target inclined beam without purlins installed.

5. The method according to claim 4, characterized in that: The calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame includes: Obtain the coordinates of one corner of the circumscribed rectangle of the candidate rectangular frame and the length and width of the circumscribed rectangle, where the length of the circumscribed rectangle is parallel to the y-axis and the width of the circumscribed rectangle is parallel to the x-axis; Determine the rectangle type of the candidate rectangular frame, where the rectangle type includes two types: a regular rectangle and a non-regular rectangle; The corner point coordinates of the vertices of the four corners of the candidate rectangular frame are calculated according to the rectangle type, the coordinates of one of the corners of the circumscribed rectangle, and the length and width of the circumscribed rectangle.

6. The method according to claim 5, characterized in that: The step of calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame according to the rectangle type, the coordinates of one corner of the circumscribed rectangle, and the length and width of the circumscribed rectangle includes: When the rectangle type is a regular rectangle, the vertex coordinates of the four corners of the circumscribed rectangle are obtained according to the coordinates of one of the corners of the circumscribed rectangle and the length and width of the circumscribed rectangle; The vertex coordinates of the four corners of the circumscribed rectangle are used as the corner point coordinates of the vertices of the four corners of the candidate rectangular frame.

7. The method according to claim 5, characterized in that: The step of calculating the corner point coordinates of the vertices of the four corners of the candidate rectangular frame according to the rectangle type, the coordinates of one corner of the circumscribed rectangle, and the length and width of the circumscribed rectangle includes: When the rectangle type is a non-normal rectangle, obtaining the centroid coordinates of the candidate rectangular frame; Find the ordinate y1 of the first corner point, where the coordinate of one of the corners of the circumscribed rectangle is (x1, y4), find all the y values ​​of the candidate rectangular box at x=x1 by indexing, and take the maximum value of all y values ​​as the ordinate y1 of the first corner point; the abscissa of the first corner point is x1; Find the abscissa x4 of the fourth corner point, find all x values ​​of the candidate rectangular box at y=y4 by index, and take the maximum value of all x values ​​as the abscissa x4 of the fourth corner point; the ordinate of the fourth corner point is y4; The longitudinal coordinate of the second corner point and the transverse coordinate of the third corner point are obtained based on the length and width of the circumscribed rectangle respectively. The transverse coordinate of the second corner point is calculated based on the transverse coordinate of the centroid of the candidate rectangular box and the transverse coordinate of the fourth corner point. The longitudinal coordinate of the third corner point is calculated based on the longitudinal coordinate of the centroid of the candidate rectangular box and the longitudinal coordinate of the first corner point.

8. A device for detecting the position of an inclined beam without installing purlins, characterized in that: include: Point cloud data acquisition module, used to acquire point cloud data of photovoltaic power stations; A ground point removal module is used to project the point cloud data on the YOZ plane and the XOZ plane respectively, perform straight line fitting on the projected point cloud, and remove points whose distance to the fitting straight line is less than a preset distance from the point cloud data to obtain target point cloud data; A rectangular region acquisition module is used to project the target point cloud data on the XOY plane and perform a connected domain analysis on the projected points to obtain a candidate rectangular frame; A target inclined beam judging module, used for judging whether the candidate rectangular frame is a target inclined beam without purlins installed; The inclined beam position output module is used to obtain and output the target inclined beam position if the target inclined beam position is the same as the output position of the inclined beam, wherein the obtaining and outputting of the target inclined beam position includes: Calculate the intersection and union ratio of the rectangular box where each first target inclined beam is located and the rectangular box of any other second target inclined beam; If the intersection-and-union ratio is less than or equal to the threshold, the position of the first target oblique beam will be output; if the intersection-and-union ratio is greater than the threshold, the area of ​​the rectangular box of the first target oblique beam is compared with the area of ​​the rectangular box of the repeated second target oblique beam, and only the target oblique beam with the largest area is retained and its position is output. The repeated second target oblique beam is the second target oblique beam whose intersection-and-union ratio with the rectangular box of the first target oblique beam is greater than the threshold.

9. A system for detecting the position of an inclined beam without installing purlins, characterized in that: include: at least one processor and at least one memory; The memory stores executable instructions of the processor; The processor is configured to execute the method according to any one of claims 1 to 7.

Images