A highway deformation disease detection method and system based on single-line laser point cloud

Abstract

The application discloses a kind of highway deformation class disease detection method and system based on single line laser point cloud, first record the pose of unmanned aerial vehicle at different time, and laser radar collects the road point cloud data of cross section form, then fitting road surface reference cross section;Fitting road datum plane and road curved surface model, determine disease point cloud position, divide 10m road surface area;Finally, calculate disease maximum depth and deformation volume parameters from three-dimensional angle, formulate index and evaluate highway deformation class disease.The application realizes the calculation of deformation class disease depth from three-dimensional angle, compared with traditional road deformation class disease detection has higher precision, proposes the evaluation index of three-dimensional road deformation class disease, enriches the evaluation method of deformation class disease, and preliminarily locates the position of disease road section needing maintenance section.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of road engineering, and particularly relates to a highway deformation disease detection method and system based on single-line laser point cloud. BACKGROUND

[0002] With the rapid economic development, asphalt road has gradually become the main pavement type of expressway in China. However, with the continuous increase of traffic volume and the increasing problems of overload, various damages occur on asphalt pavement, especially on highway asphalt pavement.

[0003] Deformation disease is a common disease of asphalt highway pavement, which is caused by external force and has obvious deformation. On the one hand, deformation disease leads to a large amount of deformation of the pavement, causing sudden changes in driving elevation, thereby affecting the comfort and stability of driving. On the other hand, the generation of deformation disease can cause the pavement structure to protrude or sag, triggering a series of other diseases, causing drainage problems, reducing the strength and service life of the pavement, and reducing the safety of the pavement. Therefore, it is necessary to systematically and comprehensively evaluate the deformation disease of the pavement and take targeted maintenance measures.

[0004] At present, the evaluation index of deformation disease at home and abroad is often based on the depth and width index of cross-section disease, and there is no unified evaluation index for deformation disease. At the same time, due to the lack of pavement detection technology, the precision and density of collected data are difficult to guarantee, and the harm of deformation disease is rarely considered from the three-dimensional perspective.

[0005] In the past, due to the backwardness of electronic technology, manual methods were the main method for detecting pavement diseases, mainly through the use of a ruler to measure the depth and width of rut. However, manual measurement relies on experience and has a large error, and has been eliminated today. In recent years, with the continuous development of electronic technology, deformation disease detection methods based on deep learning or image processing have been gradually applied, mainly using a detection vehicle as a mounting device, and realizing the detection of road deformation disease by arranging multiple multi-line laser radars. However, this method still has some drawbacks, such as being easily affected by other vehicles on the road, high cost and maintenance cost of the detection vehicle and multi-line radar, strict speed requirement, etc. Although the laser radar measurement method of pavement point cloud can guarantee the accuracy of the data, the mounting device limits the performance of the radar. Compared with the detection vehicle, the unmanned aerial vehicle can achieve lower cost, lighter weight and higher efficiency, and is theoretically a better mounting device for laser radar. At the same time, the unmanned aerial vehicle detects the deformation of the road from a high altitude, and a single-line radar is sufficient to meet the detection needs. Compared with the multi-line radar, the data processing of the single-line laser radar is more simple and simple, and it also has an advantage in the amount of calculation. After a certain point cloud data processing, the single-line radar can also generate high-density and accurate point cloud. SUMMARY

[0006] To solve the technical problems mentioned in the background art, the present application proposes a highway deformation disease detection method and system based on single-line laser point cloud.

[0007] To achieve the above technical purpose, the technical scheme of the present application is:

[0008] A highway deformation disease detection method based on single-line laser point cloud, comprising the following steps:

[0009] S1, mount a single-line laser radar based on triangulation on a UAV, record the pose of the UAV at different times during the flight of the UAV along the specified route, and collect road point cloud data in the form of a cross section.

[0010] S2, during the data collection process, there is noise interference, resulting in a series of isolated points and interference points. The collected road point cloud data is denoised to effectively exclude outliers, obtaining a single-frame point cloud of the road, screening the outside lane point cloud, fitting the standard cross section in a straight line, and inserting the inside lane single-frame point cloud according to the slope of the fitted straight line, splicing the inserted point cloud and the outside lane point cloud, and obtaining a three-dimensional point cloud of the road.

[0011] S3, according to the three-dimensional point cloud of step S2, divide the road into specific interval pavement areas, fit the reference driving surface of the road by thin plate spline interpolation method, locate the disease area on the divided pavement area, so as to study the severity of the disease, and determine the actual curved surface equation of the road by thin plate spline interpolation method.

[0012] S4, calculate the maximum settlement of the divided disease area, the maximum protrusion of the disease, and the corresponding deformation volume, determine the severity of the disease, and judge whether maintenance is needed.

[0013] Further, the specific steps of step S2 are as follows:

[0014] S201, adopt a radius filtering algorithm, make a circle with the center of each point cloud, if the number of points contained in the circle is greater than a specified value, retain it, if it is less than a specified value, delete it, adopt a statistical filtering algorithm, for each point cloud, solve the average value of the distance of all points in the domain K, solve the average value μ and the variance σ of all average values, set μ+nv as the threshold value, n is a specified multiple, take the threshold value as the screening value, thereby preliminarily remove the noise data of the road point cloud data, obtain a single-frame point cloud of the road; determine the road point cloud range based on the road cross slope and the lowest point elevation of the road surface, so as to extract accurate cross section point cloud.

[0015] S202. Using a method to extract the outer point cloud of a single frame based on point cloud coordinates, select the point cloud 0.5m-3m outside the cross section of the single frame point cloud as the emergency lane point cloud. Fit the emergency lane point cloud with a straight line to form a standard cross section. Based on the slope of the straight line, insert the single frame point cloud inside the emergency lane at point intervals within the road area to simulate the reference driving surface.

[0016] S203. Because the UAV carrying the radar may deviate and twist due to wind during operation, it is necessary to use sensors to know the specific trajectory and pose of the mounted equipment in order to locate the three-dimensional position of the cross-section it collects. Based on the single-line radar pose recorded in step S1, the cross-section point cloud extracted in S201, and the single-frame point cloud in S202, the point cloud data of the continuous frames are stitched together through coordinate transformation from NED (North East Down) to LLA (longitude, latitude, altitude) to form three-dimensional road point cloud data and reference driving surface three-dimensional point cloud data.

[0017] Furthermore, the specific steps of step S3 are as follows:

[0018] S301. Based on the emergency lane point cloud collected in step S202, extract the planar coordinates of the point cloud, fit the overall road alignment using a cubic curve to reflect the road's trend, and divide the curve into 10m segments. Extract the x and y coordinates of each segment and the normal plane of the curve at each segment. Take the point on the xOy projection line of the normal plane that is 20m inside the trend curve to determine the road area enclosed by the normal plane, thus achieving the division of the road surface into 10m segments on the reference driving surface; the coordinates of the points on the normal plane are as follows:

[0019] QD = [[x 01 y 01 x 02 y 02 ],......]

[0020] Where, x 01 The x-coordinate of the intersection point of the trend curve and the first normal plane is represented by y. 01 The x-coordinate represents the ordinate of the intersection point of the trend curve and the first normal plane. 02 The x-coordinate of the point taken inside the first normal plane is y. 02 This represents the ordinate of a point taken inside the first normal plane.

[0021] S302, Since the point cloud of the emergency lane well preserves the road elevation information at the designed place, facilitating subsequent operation, based on the reference driving surface three-dimensional point cloud data of step S203, a plane fitting algorithm is used to fit the reference driving surface of the road, and the parameters A0, A1 and A2 of the plane equation about the reference driving surface are obtained, and the specific formula is:

[0022] z=A0x+A1y+A2

[0023] Wherein, A0, A1 and A2 are respectively the parameters of the plane about x and y coordinates and the intercept of the plane when x and y coordinates are equal to 0.

[0024] S303, Based on the three-dimensional road point cloud data of step S203 and the road reference driving surface equation of S302, the elevation difference between the three-dimensional road point cloud data and the reference surface corresponding to the xOy coordinate is calculated.

[0025] S304, Based on the grid method, the xOy coordinate of the road reference surface is divided into 1dm*1dm grid, if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located below the reference surface, then the point is a subsidence disease point, and the grid area where the point is located is a subsidence disease area; if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located above the reference surface, then the point is a protrusion disease point, and the grid area where the point is located is a protrusion disease area; thereby preliminarily dividing the subsidence disease area Sa and the protrusion disease area Sb.

[0026] S305, Based on the subsidence disease area and the protrusion disease area point cloud, the three-dimensional surface equation of the road disease area is fitted by using the thin plate spline interpolation method, and the specific formula is:

[0027]

[0028] U(x)=r 2 lnr

[0029] Wherein, p(x, y) is any one point on the surface, U(x) is the radial basis function, and ||p-p i || represents the distance from point p to a certain control point, and the control points 1, 2, 3…, N are known, ω i Indicates the weighting of different radial bases, and m0, m1 and m2 are the parameters of the plane.

[0030] S306, The control point matrix and the height matrix of the point cloud data are established, and the specific formula is:

[0031] (1) Control point matrix

[0032]

[0033] Wherein, n is the number of control points, the second and third columns represent the (x, y) coordinates of the control points.

[0034] (2) Height matrix

[0035]

[0036] Wherein, v1 to v n represent the coordinates of each control point in the z direction.

[0037] S307, calculate the radial basis function value of any two control points, the specific formula is:

[0038]

[0039] Wherein, r ij is the distance between control points i and j, U(r ij ) is the value of the radial basis function corresponding to the distance r ij .

[0040] S308, define the matrix L as:

[0041]

[0042] The above matrix has the following relationship:

[0043] Y=L*(ω1, …ω N , m0, m1, m2) T .

[0044] Based on the principle of thin plate spline interpolation, substitute the condition function about the control points, calculate all parameters of the road three-dimensional surface equation, and complete the interpolation, the specific matrix is:

[0045]

[0046] Wherein, ω ij is the weight of the jth radial basis on the ith segment, m i0 , m i1 , m i2 is the m0, m1, m2 coefficient on the ith segment.

[0047] Further, the specific steps of step S4 are:

[0048] S401, extract the point positions M and N corresponding to the maximum elevation difference in step S3, take the grid where the M and N points are located and its eight surrounding grids as the selected area, and solve the gradient of the point positions M and N based on the steepest descent / ascent method. The specific formula is:

[0049] f=z 实际 -z 基准

[0050]

[0051] Where f is the difference between the actual road surface elevation and the road reference driving surface, z 实际 z represents the actual elevation of the road surface. 基准 Let M be the reference driving surface of the road; let M be the known minimum point in the selected defect area, and let M be the point obtained after k iterations. k P k M represents k The direction of the maximum rate of change of the point surface, d k This represents the gradient.

[0052] S402. According to the formula in step S305, we obtain z. 基准 Value, for z 基准 Solving the differential using the value formula yields the following formula:

[0053]

[0054]

[0055] d k The final expression is:

[0056]

[0057] Where N is the number of control points for fitting the reference surface, Q is the number of control points for fitting the actual road surface, and M... k (x, y) is the starting point of the current iteration; x i y i r represents the coordinates of the corresponding control point; k (x,y) is the distance from each control point.

[0058] S403. Solve for the elevation of the next point with a step size of 0.01m and iterate until an extreme point appears. The elevation difference between the extreme point and the reference plane of the corresponding xOy coordinate is the depth / height of the damage on the road surface.

[0059] S404. Take 9 points regularly in each disease grid and calculate the average elevation. Pick The product of the area and the mesh area is the deformable volume V of the mesh. i The specific formula is as follows:

[0060]

[0061]

[0062] Where S is the grid area of ​​4 cm² 2 Vi For each grid of the pavement area from left to right, and then from bottom to top, the deformation volume of the deformed disease.

[0063] Add all the grid deformation volumes, that is, the volume V of the pavement area subsidence / protrusion a , V b .

[0064] S405, based on the pavement disease depth / disease height, and the pavement area subsidence / protrusion volume, complete the three-dimensional evaluation of the road deformation disease.

[0065] Further, the present application also proposes a highway deformation disease detection system based on single line laser point cloud, comprising

[0066] The information acquisition module is used for mounting the single line laser radar on the unmanned aerial vehicle, recording the pose of the unmanned aerial vehicle at different times during the flight of the unmanned aerial vehicle along the specified route, and collecting the road point cloud data in the form of cross section.

[0067] The road three-dimensional point cloud acquisition module is used for denoising the collected road point cloud data to obtain a single frame point cloud of the road, screening the outside lane point cloud, fitting the standard cross section in a straight line, inserting the inside lane single frame point cloud according to the slope of the fitted straight line, splicing the inserted point cloud and the outside lane point cloud, and obtaining the three-dimensional point cloud of the road.

[0068] The disease area positioning module is used for dividing the road into specific pavement areas according to the road three-dimensional point cloud, fitting the reference driving surface of the road by the thin plate spline interpolation method, positioning the disease area in the divided pavement area and determining the actual curved surface equation of the road.

[0069] The disease area deformation volume calculation module is used for calculating the maximum subsidence, the maximum protrusion of the disease and the corresponding deformation volume of the divided disease area.

[0070] Further, in the road three-dimensional point cloud acquisition module, the specific steps are as follows:

[0071] Step 1, using the radius filtering algorithm, making a circle with the center of each point cloud, retaining the points in the circle whose number is greater than a specified value, and deleting the points whose number is less than a specified value, using the statistical filtering algorithm, solving the average value of the distance of each point cloud to all points in the domain K, solving the average value μ and the variance σ of all average values, setting μ + nσ as the threshold value, n is a specified multiple, using the threshold value as the screening value, thereby preliminarily removing the noise data of the road point cloud data, obtaining the single frame point cloud of the road; based on the road cross slope and the lowest point elevation of the pavement, the pavement point cloud range is determined, so as to extract the accurate cross section point cloud.

[0072] Step 2, using the method of extracting single-frame data outside point cloud based on point cloud coordinates, screening the point cloud of 0.5m-3m outside the cross section direction in single-frame point cloud as the emergency lane point cloud, fitting the emergency lane point cloud in a straight line to form a standard cross section, and inserting the single-frame point cloud inside the emergency lane according to the point spacing based on the slope of the straight line.

[0073] Step 3, based on the single-line radar pose, cross section point cloud and single-frame point cloud, through the coordinate transformation from NED to LLA coordinate system, the point cloud data of continuous frames is spliced to form three-dimensional road point cloud data and three-dimensional point cloud data of the reference driving surface.

[0074] Further, in the disease area positioning module, the specific steps are as follows:

[0075] Step 1, based on the emergency lane point cloud, the point cloud plane coordinates are extracted, the overall line type of the road is fitted in a cubic curve, and the curve is divided into 10m small sections, the x, y coordinates at the section are extracted, and the normal plane of the curve at the section is extracted, the point inside the 20m of the normal plane on the xOy projection straight line from the trend curve is taken, the road area between the normal planes is determined, and the 10m road surface area division on the reference driving surface is realized; the coordinates of the point taken by the normal plane are marked as:

[0076] QD = [[x 01 , y 01 , x 02 , y 02 ],...]

[0077] Wherein, x 01 represents the horizontal coordinate of the intersection point of the trend curve and the first normal plane, y 01 represents the vertical coordinate of the intersection point of the trend curve and the first normal plane, x 02 represents the horizontal coordinate of the point taken inside the first normal plane, and y 02 represents the vertical coordinate of the point taken inside the first normal plane.

[0078] Step 2, based on the three-dimensional point cloud data of the reference driving surface, the reference driving surface of the road is fitted by using the plane fitting algorithm to obtain the parameters A0, A1 and A2 of the plane equation of the reference driving surface, and the specific formula is:

[0079] z = A0x + A1y + A2

[0080] Wherein, A0, A1 and A2 are respectively the parameters of the plane about x, y coordinates and the intercept of the z coordinate when the plane in x, y coordinates is equal to 0.

[0081] Step 3, based on the three-dimensional road point cloud data and the equation of the reference driving surface of the road, the elevation difference between the three-dimensional road point cloud data and the reference surface corresponding to the xOy coordinate is calculated.

[0082] Step 4, based on the grid method, the xOy coordinates of the road datum surface are divided into 1dm x 1dm grids, if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located below the datum surface, then the point is a subsidence disease point, and the grid area where the point is located is a subsidence disease area; if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located above the datum surface, then the point is a protrusion disease point, and the grid area where the point is located is a protrusion disease area; thereby preliminarily dividing the subsidence disease area Sa and the protrusion disease area Sb.

[0083] Step 5, based on the subsidence disease area and the protrusion disease area point cloud, the three-dimensional surface equation of the road disease area is fitted by the thin plate spline interpolation method, and the specific formula is:

[0084]

[0085] U(x)=r 2 lnr

[0086] Wherein, p(x,y) is any one point on the surface, U(x) is the radial basis function, ||p-p i || represents the distance from point p to a certain control point, the control points 1, 2, 3…, N are known, ω i represents the weighting of different radial basis, m0, m1, m2 are the parameters of the plane.

[0087] Step 6, the control point matrix and the height matrix of the point cloud data are established, and the specific formula is:

[0088] (1) Control point matrix

[0089]

[0090] Wherein, n is the number of control points, the second and third columns represent the (x, y) coordinates of the control points.

[0091] (2) Height matrix

[0092]

[0093] Wherein, v1 to v n represent the coordinates of each control point in the z direction.

[0094] Step 7, calculate the radial basis function value of any two control points, and the specific formula is:

[0095]

[0096] Wherein, r ij represents the distance between control points i and j, U(r ij ) is the radial basis function corresponding to the distance r ijThe value of the matrix L is defined.

[0097] Step 8, define the matrix L as:

[0098]

[0099] The above matrix has the following relationship:

[0100] Y=L*(ω1, …ω N , m0, m1, m2) T .

[0101] Based on the thin plate spline interpolation principle, the condition function about the control point is substituted, all parameters of the road three-dimensional surface equation are calculated, and interpolation is completed, and the specific matrix is:

[0102]

[0103] Where, ω ij is the weight of the jth radial basis on the ith segment, and m i0 , m i1 , m i2 is the m0, m1, m2 coefficient on the ith segment.

[0104] Further, in the disease area deformation volume calculation module, the specific steps are as follows:

[0105] Step 1, extract the point positions M and N corresponding to the maximum elevation difference, take the grid where the M and N points are located and the eight surrounding grids as the selected area, and solve the gradient of the point M and N based on the steepest descent / ascent method. The specific formula is:

[0106] f=z 实际 -z 基准

[0107]

[0108] Where, f is the difference between the actual road surface elevation and the road reference driving surface, z 实际 is the actual road surface elevation, and z 基准 is the road reference driving surface; let the known minimum point in the selected disease area be M, and the point obtained after k iterations be M k , P k represents the direction of the largest surface change rate of the M k point, and d k represents the gradient.

[0109] S402, according to the formula of step S305, the value of z 基准 is obtained, and the differential of the z 基准 value formula is solved to obtain the following formula:

[0110]

[0111]

[0112] d k Finally expressed as:

[0113]

[0114] Wherein, N is the number of control points of the reference surface fitting, Q is the number of control points when the actual road surface is fitted, M k (x,y) is the starting point of the current iteration;x i , y i is the coordinate of the corresponding control point;r is the distance from M k (x,y) to each control point.

[0115] Step 3, the height of the next point is solved with 0.01m as a step, and iteration is carried out until the extreme point appears, and the height difference of the extreme point and the corresponding xOy coordinate of the reference surface is the disease depth / disease height on the road surface area.

[0116] Step 4, 9 points are regularly taken in each disease grid, and the average value of the height thereof is calculated Take The product of the grid area is the deformation volume V of the grid. i , the specific formula is:

[0117]

[0118]

[0119] Wherein, S is the grid area 4cm 2 ; V i is the deformation volume of the grid existing deformation disease from left to right and then from bottom to top in each road surface area.

[0120] Add all the grid deformation volumes, that is, the volume V a , V b of the road surface area subsidence / convexity.

[0121] Step 5, based on the road disease depth / disease height and the road surface area subsidence / convexity volume, the three-dimensional evaluation of the road deformation disease is completed.

[0122] The above technical scheme is adopted in the present application, compared with the prior art, and the significant technical effects are as follows:

[0123] The application fits the equation of the road curve surface and the road reference driving surface, simply and effectively positions the road disease area, completes the calculation of the depth and volume of the deformation type disease in three-dimensional hierarchy, and proposes a comprehensive and effective evaluation index according to the specification. Based on the unmanned aerial vehicle laser radar collection technology, a lighter and more cost-effective loading device is used, a single-line laser radar with smaller data volume is used, and a low-computational three-dimensional deformation disease extraction algorithm is designed, which reduces the operation time, increases the inspection efficiency, guarantees the accuracy of the deformation disease depth calculation, and realizes the evaluation of the deformation disease from the three-dimensional angle.

[0124] Meanwhile, the application uses the thin plate spline interpolation method to fit the equation of the road curve surface, and writes the gradient formula of the equation based on the equation, and further improves the accuracy of the deformation disease depth detection in combination with the principle of the steepest descent / ascent method, and proposes a comprehensive deformation disease evaluation index according to the previous specification, which provides certain guidance and help for the selection of subsequent maintenance road sections and the calculation of the use amount of maintenance materials. BRIEF DESCRIPTION OF DRAWINGS

[0125] Figure 1 is the overall flowchart of the embodiment of the application.

[0126] Figure 2 is a schematic diagram of a frame of cross-section point cloud collected by the single-line laser radar of the embodiment of the application.

[0127] Figure 3 is a three-dimensional point cloud image after point cloud splicing of the embodiment of the application.

[0128] Figure 4 is a schematic diagram of point cloud disease area detection depth calculation of the embodiment of the application.

[0129] Figure 5 is a schematic diagram of micro-element square point selection for calculating disease volume of the embodiment of the application. DETAILED DESCRIPTION

[0130] In order to make the technical means, creative features, purposes and effects realized by the application easy to understand, the following will combine embodiments and drawings to specifically describe the highway deformation disease detection method based on single-line laser point cloud.

[0131] The highway deformation disease detection method based on single-line laser point cloud, the flowchart is as shown in Figure 1 The specific steps include the following steps:

[0132] S1. In this embodiment, the data acquisition device is a DJI M600 Pro drone, and the LiDAR is a Sick LMS511 single-line LiDAR. The LiDAR is mounted under the drone, and the drone is positioned above the road to take pictures. The flight speed is set to 1 m / s for inspection.

[0133] During the flight of the UAV along the prescribed route, a single-line radar detects at a frequency of 25 Hz, with an angular resolution of 0.1667° for each detection. Each frame collects approximately 1,100 data points, recording the UAV's pose at different times, as well as 10-meter segment of road point cloud data in cross-sectional form collected by lidar.

[0134] S2. Denoise the collected road point cloud data to effectively eliminate outliers and obtain single-frame point clouds of the road. Filter the outer lane point clouds and fit them to a standard cross-section using a straight line. Insert the inner lane single-frame point cloud based on the slope of the fitted line. Stitch the inserted point cloud and the outer lane point cloud together to obtain the 3D point cloud of the road. The specific steps are as follows:

[0135] S201, road point cloud data consists of over a thousand frames of road cross-sectional data collected by a single-line lidar, such as... Figure 2 As shown. Using Open3D to read a single-frame point cloud, the lowest point in the point cloud data is first selected. This point is the lowest point of the road surface point cloud. The road surface data is initially filtered using the elevation of this point plus 30cm as the boundary. A radius filtering algorithm is used: for each point cloud, a circle is drawn with its center. Points within the circle that are greater than a certain value are retained, while those less than the certain value are deleted. In this embodiment, the certain value is set to 2cm. A statistical filtering algorithm is also used: for each point cloud, the average distance to all points in its neighborhood is calculated. The mean μ and variance σ of all averages are calculated. μ + nσ is set as the threshold, where n is a specified multiple. This threshold is used as the value for the radius filtering algorithm to filter the point cloud. Combined with the road surface elevation and the normal vector of the point cloud, noise data in the single-frame point cloud is initially removed, resulting in the single-frame point cloud of the road.

[0136] The normal vector of a point cloud is the normal vector of the plane fitted by the point cloud of a given point and its neighborhood. For a road surface, the direction of the normal vector is mainly distributed along the z-coordinate. The normal vector is calculated using the estimate normals function of Open3D, and the magnitude of the normal vector is corrected to 1, with the direction being the positive z-coordinate.

[0137] If the z-coordinate of the normal vector is greater than 0.9, then the point is a road surface point without defects. The maximum and minimum values ​​of the y-coordinates of a single frame point cloud are extracted, and all point clouds within this range are considered road surface points. The range of the road surface point cloud is determined based on the road cross slope and the elevation of the lowest point of the road surface, thereby extracting accurate cross-sectional point clouds for stitching together 3D point cloud data.

[0138] S202. Since emergency lanes generally do not experience deformation or damage during road operation, extending the cross-section of the outer emergency lane as a straight line can yield a result similar to the standard cross-section. In the absence of existing design data, it can be regarded as the standard cross-section.

[0139] The processed road surface point cloud was read using Open3D. A method based on point cloud coordinates was employed to extract the outer point cloud from a single frame. Points within 0.5m-3m of the cross-sectional direction in each frame were selected as the emergency lane point cloud. A straight line was fitted to the emergency lane point cloud to form a standard cross-section. Based on the slope of the line, single-frame point clouds were inserted into the inner side of the emergency lane within the road area at 2cm intervals to simulate the baseline driving surface. The inserted point clouds and their corresponding frame numbers were recorded in the array DY.

[0140] DY = [[x, y, z, (frame number)], ...].

[0141] S203. Based on the single-line radar pose recorded in step S1, the cross-sectional point cloud extracted in S201, and the single-frame point cloud in S202, the point cloud data of consecutive frames are stitched together through coordinate transformation from NED to LLA coordinate system to form three-dimensional road point cloud data and reference driving surface three-dimensional point cloud data.

[0142] To facilitate point cloud modeling, stationing, and region division, the single-line radar pose recorded in step S1, the cross-sectional point cloud extracted in S201, and the single-frame point cloud in S202 are converted into an array DY1 using NumPy. A new column is added after all point cloud data to mark the original frame number of the point cloud. DY1 is:

[0143] DY1 = [[x, y, z, (frame number)], ...].

[0144] S3. Based on the 3D point cloud from step S2, the road is divided into road surface areas spaced 10m apart. The reference driving surface of the road is fitted using thin-plate spline interpolation. The areas where defects are located are then located on the divided road surface areas to facilitate the study of the severity of the defects. Finally, the actual surface equation of the road is determined using thin-plate spline interpolation. The specific steps are as follows:

[0145] Figure 3 This is a 3D point cloud image of this embodiment. The black area circled in the circle represents the point cloud region of the disease in this embodiment.

[0146] S301, based on the emergency lane point cloud plane coordinates collected in step S202, the python library function is used to complete the cubic curve fitting of the overall line type of the road, label the stake number of the curve, and divide it into 10m small sections, extract the x, y coordinates at the segmented place and the normal plane of the curve at the segmented place, take the point on the xOy projection straight line of the normal plane which is 20m inside the trend curve, determine the road area between the normal planes, and record the coordinates in the numpy array, so as to realize the division of 10m road surface area on the reference driving surface. The coordinates of the point taken by the normal plane are marked as:

[0147] QD = [[x 01 , y 01 , x 02 , y 02 ],...... ]

[0148] Wherein, x 01 represents the horizontal coordinate of the intersection point of the trend curve and the first normal plane, y 01 represents the vertical coordinate of the intersection point of the trend curve and the first normal plane, x 02 represents the horizontal coordinate of the point taken inside the first normal plane, and y 02 represents the vertical coordinate of the point taken inside the first normal plane.

[0149] S302, based on the three-dimensional point cloud data of the reference driving surface in step S203, the plane fitting algorithm is used to fit the reference driving surface of the road, and the parameters A0, A1 and A2 about the reference driving surface are obtained. The specific formula is:

[0150] z = A0x + A1y + A2

[0151] Wherein, A0, A1 and A2 are respectively the parameters of the plane about x, y coordinates and the intercept of the plane when x, y coordinates are equal to 0.

[0152] S303, based on the three-dimensional road point cloud data in step S203 and the road reference driving surface equation in S302, the elevation difference between the three-dimensional road point cloud data and the reference surface equation under the corresponding xOy coordinates is calculated. If the value is negative, it means that the point is below the reference surface, otherwise it is above the reference surface.

[0153] S304, based on the grid method, the xOy coordinates of the road reference surface are divided into 1dm x 1dm grid.

[0154] In the array DY1 described in step S203, a column of data is added for judging non-deformation disease point clouds. If the absolute value of the point cloud elevation difference is greater than a preset threshold value, and the point is located below the reference surface, the point is a subsidence disease point, and the grid area where the point is located is a subsidence disease area; if the absolute value of the point cloud elevation difference is greater than a preset threshold value, and the point is located above the reference surface, the point is a protrusion disease point, and the grid area where the point is located is a protrusion disease area. The threshold value set in this embodiment is ±5 mm.

[0155] After adding a column of data to the array DY1, DY2 is obtained:

[0156] DY2 = [[x, y, z, (frame number), (none / subsidence / protrusion)],...]

[0157] In order to divide the position of the point cloud and facilitate subsequent operation processing, the disease point cloud is divided by grid. The information of the grid is represented by an array WG, which is specifically:

[0158] WG = [[x, y, (subsidence / protrusion), (pavement area serial number)],...]

[0159] Among them, the first two columns of data represent the coordinate position of the grid, such as (2, 3) representing the area surrounded by x=2 cm, x=4 cm, y=4 cm, and y=6 cm in the coordinate system; the third column of data is used to mark the subsidence and protrusion of the grid. According to the fourth column of data of the array DY2, if all the point clouds in the grid are not subsidence or protrusion point clouds, there is no disease area in the grid; if there are subsidence or protrusion in the grid, the grid is a subsidence disease area Sa or a protrusion disease area Sb, which is used for calculating the subsidence volume V a and the protrusion volume V b ; the fourth column of data is the pavement area serial number of the grid with disease, which can be determined according to the position relationship between the grid and the normal plane (straight line after projection onto the xOy plane) described in step S301, so as to assign a value to the fourth column of data of the grid. For the grid without disease, in order to simplify the calculation, it is defaulted to 0.

[0160] The method for judging the position relationship between the grid and the normal plane is: taking the midpoint G of the disease grid, calculating the vector of point G to the points recorded in the QD array; calculating the vector inner product of the adjacent normal planes from the starting surface, if the vector inner product is negative, G is between the adjacent normal planes corresponding to the vector, thereby determining the pavement area where the grid is located; if the vector inner product is positive, G is outside the adjacent normal planes corresponding to the vector, and the vector inner product of the next group of normal planes and G is calculated.

[0161] S305, based on the subsidence disease area and the raised disease area point cloud, a three-dimensional surface equation of the road disease area is fitted by using a thin plate spline interpolation method, and a specific formula is as follows:

[0162]

[0163] U(x)=r 2 lnr

[0164] wherein p(x, y) is any one point on the surface, U(x) is a radial basis function, and ||p-p i || represents the distance from the point p to a certain control point, the control points 1, 2, 3, …, N are known, and ω i represents the weighting of different radial bases, and m0, m1, m2 are parameters of the plane.

[0165] S306, the three-dimensional surface equation has point cloud number + 3 parameters, a control point matrix and a height matrix of the point cloud data are established, and a specific formula is as follows:

[0166] (1) Control point matrix

[0167]

[0168] wherein n is the number of control points, and the second and third columns represent the (x, y) coordinates of the control points.

[0169] (2) Height matrix

[0170]

[0171] wherein v1 to v n represent the coordinates in the z direction of each control point.

[0172] S307, the radial basis function values of any two control points are calculated, the condition function of the thin plate spline interpolation is substituted, all parameters of the equation are calculated, and the interpolation is completed, and a specific formula is as follows:

[0173]

[0174] wherein r ij represents the distance between the control points i and j, and U(r ij ) is the value of the radial basis function corresponding to the distance r ij .

[0175] S308, the matrix L is defined as:

[0176]

[0177] The above matrix has the following relationship:

[0178] Y=L*(ω1,…ωN , m0, m1, m2) T .

[0179] Based on the thin plate spline interpolation principle, the condition function about the control point is substituted, all parameters of the road three-dimensional surface equation are calculated, and interpolation is completed, and the specific matrix is:

[0180]

[0181] Where, ω ij is the weight of the jth radial basis on the ith segment, m i0 , m i1 , m i2 is the m0, m1, m2 coefficient on the ith segment.

[0182] S4, calculate the maximum settlement of the divided disease area, the maximum protrusion of the disease and the corresponding deformation volume, determine the disease severity, and judge whether maintenance is needed, the specific steps are as follows:

[0183] S401, the max and min functions of python are used to solve the point positions M and N corresponding to the maximum and minimum values of the elevation difference, the grid and the elevation difference, and the extreme value in the neighborhood of the point positions M and N is the maximum value of the deformation on the road surface area.

[0184] The grid where the point positions M and N are located and the surrounding 8 grids are selected as the selected area, and the steepest descent / ascent method is used to solve the extreme value of the actual model of the area road, which is the disease depth / disease height of the road surface area. The gradient of the point positions M and N is solved, as shown in the left figure of Figure 4 , the left figure selects the grid containing the maximum elevation difference point, and the right figure uses the steepest descent / ascent method in the area to solve the maximum elevation difference point of the model. The specific formula is:

[0185] f = z 实际 -z 基准

[0186]

[0187] Where, f is the difference between the actual road surface elevation and the road reference driving surface, z 实际 is the actual road surface elevation, and z 基准 is the road reference driving surface; let the known minimum point in the selected disease area be M, and the point obtained after k iterations be M k , P k represents the direction of the maximum surface change rate of the point M k , d k represents the gradient;

[0188] S402, according to the formula of step S305, z 基准Value, z 基准 The differential is solved by the value formula, and the following formula is obtained:

[0189]

[0190]

[0191] d k The final expression is:

[0192]

[0193] Where N is the number of control points for fitting the reference surface, Q is the number of control points for fitting the actual road surface, M k (x, y) is the starting point of the current iteration; x i , y i is the coordinate of the corresponding control point; r is the distance from (x, y) to each control point. k

[0194] S403, in order to ensure the accuracy of the solution, the search step is λ = 1 cm, and M k is searched along the direction d k , M k+1 = M k + λd k . Check whether the modulus of the gradient of the original point is less than 0.01, if it is satisfied, the point is the extreme value point, if it is not satisfied, iterate until the modulus of the gradient is less than 0.01, end the iteration. Output f and M k , which is the maximum value H a or H b of the deformation depth, and the coordinates corresponding to the maximum value. The maximum depth of the disease calculated by the method in this embodiment is 82.9 mm.

[0195] S404, regularly take 9 points in each disease grid, and the taking method is as shown in Figure 5 , calculate the average elevation Take the product of the grid area as the deformation volume V i of the grid, and the specific formula is:

[0196]

[0197]

[0198] Where, is the average of the elevations of the nine points taken, and S is the grid area 4 cm 2 ; V i is the deformation volume of the grid with deformation disease from left to right, and then from bottom to top in each pavement area.​

[0199] If V i is positive, the grid calculates the volume of the protrusion; if V i is negative, the volume of the depression is obtained. According to the positive and negative of V i , the deformation volume of all grids is added respectively according to the road surface area to which the grid belongs, that is, the volume V a of the depression / protrusion of the road surface area, V b .

[0200] The volume index V a , V b of the road section is larger, and the future development of serious diseases is possible, and it can help and guide the subsequent selective maintenance of the road.

[0201] The volume of the depression disease calculated by the embodiment is 0.1532m 3 , and no protrusion disease is detected.

[0202] S405, based on the road disease depth / disease height and the road surface area depression / protrusion volume, the three-dimensional evaluation of the road deformation disease is completed.

[0203] The evaluation index of deformation diseases such as bump, wave, rut, and depression is mostly 5mm, 15mm, and 25mm as a boundary. The maximum deformation depth H a , H b of the application is also 5mm, 15mm, and 25mm as a boundary, and the deformation severity is divided, as shown in Table 1. The difference is that the commonly used deformation disease index is the distance from the line to the lowest point by the virtual pull line method or the virtual ruler method. The maximum deformation depth H a , H b of the application is the distance from the fitted road datum to the lowest point of the disease. Therefore, for the same road section, the calculated value of the maximum deformation depth H a , H b is relatively smaller.

[0204] Table 1 Disease severity classification table

[0205]

[0206] In combination with Table 1, in the detection of actual road deformation diseases, H a , H b , V a , V b , and the severity of each road section are listed in Table 2 for evaluating the deformation of the whole road. Among them, H a , H bThe severity of the road deformation disease is determined, and Va and Vb are the volumes of the subsidence and the protrusion. For the volume index V a , V b Both can provide certain information support for the maintenance and monitoring scheme. For the road sections with medium and low severity, if there is a larger deformation volume, the possibility of subsequent development of the disease into a high severity disease is larger, which needs to be focused on in subsequent monitoring and maintenance, and is labeled in the remarks.

[0207] Table 2: Road surface area deformation depth and volume summary table

[0208]

[0209]

[0210] The embodiment of the present application also provides a highway deformation disease detection system based on single-line laser point cloud, which comprises an information acquisition module, a road three-dimensional point cloud acquisition module, a disease area positioning module, a disease area deformation volume calculation module and a computer program capable of running on a processor. It should be noted that each module in the above system corresponds to the specific steps of the method provided by the embodiment of the present application, has the corresponding function modules and beneficial effects of the execution method. Technical details not described in detail in the present embodiment can be referred to the method provided by the embodiment of the present application.

[0211] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can adopt a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes. The scheme in the embodiments of the present application can be implemented in various computer languages, such as object-oriented programming language Java and interpreted scripting language JavaScript.

[0212] The present application is described with reference to flowcharts and / or block diagrams according to the method, device (system) and computer program product of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a machine that implements the flowcharts and / or block diagrams. Figure 1 one flow or multiple flows and / or blocksFigure 1 means for performing the function specified by the block or blocks.

[0213] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions Figure 1 one or more flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0214] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions Figure 1 one or more flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0215] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the preferred embodiments by those of skill in the art once they have the benefit of the present disclosure. Therefore, the appended claims are intended to encompass within their scope all possible variations and modifications of the preferred embodiments. 1

[0216] It is apparent that a person skilled in the art can make a variety of changes and modifications to the application without departing from the spirit and scope thereof. Thus, if these modifications and changes fall within the scope of the claims and their equivalents, it is intended to include them in the application.

Claims

1. A method for detecting highway deformation disease based on single-line laser point cloud, characterized in that, The method comprises the following steps: S1, mounting a single-line laser radar on a UAV, recording the pose of the UAV at different times during the flight of the UAV along a specified route, and collecting road point cloud data in the form of a cross section; S2, denoising the collected road point cloud data to obtain a single-frame point cloud of the road, screening the point cloud of the outer lane, fitting a standard cross section in a straight line manner, and inserting the point cloud of the inner lane according to the slope of the fitted straight line, splicing the inserted point cloud and the point cloud of the outer lane to obtain a three-dimensional point cloud of the road; S3, according to the three-dimensional point cloud of step S2, dividing the road into road surface areas with a certain interval, fitting the reference driving surface of the road by a thin plate spline interpolation method, positioning the disease area on the divided road surface area, and determining the actual curved surface equation of the road; S4, calculating the maximum settlement of the divided disease area, the maximum protrusion of the disease, and the corresponding deformation volume; specifically: S401, extracting the point positions M and N corresponding to the maximum height difference, taking the grid where the M and N points are located and the eight surrounding grids as the selected area, and solving the gradient of the point positions M and N based on the steepest descent / ascent method, and the specific formula is: f = z 实际 - z 基准 wherein f is the difference between the actual road surface elevation and the road reference driving surface, z 实际 is the actual road surface elevation, z 基准 is the road reference driving surface; the known minimum point provided in the selected disease area is M, and the point obtained after k iterations is denoted as M k , P k represents the direction of the maximum change rate of the M k point surface, d k represents the gradient; S402、According to the formula of step S305, z is obtained 基准 value, z 基准 The differential is solved by the formula as follows: d k The final expression is: Wherein, N is the number of control points of the reference surface fitting, Q is the number of control points when the actual road surface is fitted, M k (x, y) is the starting point of the current iteration; x i , y i is the coordinate of the corresponding control point; r is the distance from M k (x, y) to each control point; S403, solving the elevation of the next point with a step size of 0.01 m, and iterating until the extreme value point appears, and the elevation difference between the extreme value point and the corresponding xOy coordinate of the reference surface is the disease depth / disease height on the road surface area; S404、In each disease grid, take 9 points regularly, calculate the average value of its elevation Take The product of the grid area and the grid deformation volume V i The specific formula is: wherein S is the grid area 4 cm 2 ; V i is the volume of deformation of the grid in each pavement area from left to right, from bottom to top, which presents deformation disease. Summing all the grid deformation volumes, i.e. the volumes V of the road surface area that are sunken / raised a , V b ; S405, based on the road surface disease depth / disease height and the road surface area settlement / protrusion volume, completing the three-dimensional evaluation of the road deformation disease.

2. The single-line laser point cloud based highway deformation disease detection method according to claim 1, characterized in that, The specific steps of step S2 are as follows: S201, using a radius filtering algorithm, making a circle with the center of each point cloud, and retaining the points in the circle if the number of points is greater than a specified value, and deleting the points if the number of points is less than a specified value, using a statistical filtering algorithm, solving the average distance of each point cloud to all points in the domain, and then solving the average value μ and the variance σ of all average values, setting μ+nσ as the threshold value, n being a specified multiple, using the threshold value as the screening value, thereby preliminarily removing the noise data of the road point cloud data, obtaining a single-frame point cloud of the road, and determining the road point cloud range based on the road cross slope and the lowest point elevation of the road surface, thereby extracting accurate cross section point cloud; S202, using a method of extracting single-frame data outer point cloud based on point cloud coordinates, screening the point cloud of 0.5m-3m outside the cross section direction in the single-frame point cloud as the emergency lane point cloud, fitting the emergency lane point cloud in a straight line manner to form a standard cross section, and inserting the single-frame point cloud inside the emergency lane according to the slope of the straight line within the road range; S203, based on the single-line radar pose recorded in step S1, the cross section point cloud extracted in S201, and the single-frame point cloud of S202, the coordinate transformation from NED to LLA coordinate system is used to splice the point cloud data of consecutive frames to form three-dimensional road point cloud data and reference driving surface three-dimensional point cloud data.

3. The single-line laser point cloud based highway deformation disease detection method according to claim 2, characterized in that, The specific steps of step S3 are as follows: S301, based on the emergency lane point cloud collected in step S202, the point cloud plane coordinates are extracted, the overall line type of the road is fitted by using a cubic curve method, and the curve is divided into 10m small sections, the x, y coordinates at the segmented positions and the normal plane of the curve at the segmented positions are extracted, the point on the inside of the normal plane 20m away from the trend curve on the xOy projection straight line is taken, the road area clamped by the normal plane is determined, and thus the division of the 10m road surface area on the reference driving surface is realized; the normal plane point coordinate is marked as: QD = [[x 01 , y 01 , x 02 , y 02 ],...] wherein x 01 represents the abscissa of the intersection point of the trend curve with the first normal plane, y 01 represents the ordinate of the intersection point of the trend curve with the first normal plane, x 02 represents the abscissa of the inside point of the first normal plane, y 02 represents the ordinate of the inside point of the first normal plane; S302, based on the three-dimensional point cloud data of the reference driving surface in step S203, a plane fitting algorithm is used to fit the reference driving surface of the road, and the parameters A1, A0 and A2 of the plane equation about the reference driving surface are obtained, and the specific formula is: z=A0x+A1y+A2 Wherein, A0, A1, A2 are respectively the parameters of the plane about x, y coordinates and the intercept of the plane when x, y coordinates are equal to 0, z coordinate; S303, based on the three-dimensional road point cloud data in step S203 and the road reference driving surface equation in S302, the elevation difference between the three-dimensional road point cloud data and the corresponding xOy coordinate reference surface is calculated; S304, based on the grid method, the xOy coordinates of the road reference surface are divided into 1dm*1dm grids, if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located below the reference surface, then the point is a subsidence disease point, and the grid area where the point is located is a subsidence disease area; if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located above the reference surface, then the point is a protrusion disease point, and the grid area where the point is located is a protrusion disease area; thus, the subsidence disease area Sa and the protrusion disease area Sb are preliminarily divided; S305, based on the subsidence disease area and the protrusion disease area point cloud, the three-dimensional surface equation of the road disease area is fitted by using the thin plate spline interpolation method, and the specific formula is: U(x) = r 2 lnr where p(x, y) is any point on the surface, U(x) is the radial basis function, and ||p-p i || represents the distance from point p to a certain control point, and control points 1, 2, 3, …, N are known, ω i represents the weighting of different radial bases, and m0, m1, m2 are parameters of the plane. S306, the control point matrix and the height matrix of the point cloud data are established, and the specific formula is: (1) Control point matrix Wherein, n is the number of control points, the second and third columns represent the (x, y) coordinates of the control points; (2) Height matrix wherein v1 to v n represent the coordinates in the z-direction of each control point z; S307, the radial basis function value of any two control points is calculated, and the specific formula is: Where, r ij U(r) represents the distance between control points i and j. ij ) represents the distance r corresponding to the radial basis function. ij The value; S308, define the matrix L as: Then the above matrix has the following relationship: Y = L * (ω1,... ω N , m0, m1, m2) T ; Based on the thin plate spline interpolation principle, the condition function about the control point is substituted, all parameters of the road three-dimensional surface equation are calculated, and the interpolation is completed, and the specific matrix is: where ω ij is the weight of the jth radial basis on the ith segment, m i0 ,m i1 ,m i2 are the m0, m1, m2 coefficients on the ith segment.

4. A highway deformation disease detection system based on single-line laser point cloud, characterized in that, The information acquisition module is used to mount the single-line laser radar on the unmanned aerial vehicle, record the poses of the unmanned aerial vehicle at different times during the flight of the unmanned aerial vehicle along the specified route, and collect the road point cloud data in the form of cross section; The road three-dimensional point cloud acquisition module is used to denoise the collected road point cloud data to obtain a single frame of point cloud of the road, screen the outside lane point cloud, fit the standard cross section in a straight line, insert the inside lane single frame point cloud according to the slope of the fitted straight line, splice the inserted point cloud and the outside lane point cloud, and obtain the three-dimensional point cloud of the road; The disease area positioning module is configured to divide the road into specific interval pavement areas according to the road three-dimensional point cloud, fit the reference driving surface of the road by using a thin plate spline interpolation method, position the disease area on the divided pavement area, and determine the actual curved surface equation of the road. The disease area deformation volume calculation module is configured to calculate the maximum settlement, the maximum protrusion, and the corresponding deformation volume of the divided disease area; specifically, the following steps are performed: Step 1: Extract the point positions M and N corresponding to the maximum height difference, take the grid where the M and N points are located and the eight surrounding grids as the selected area, and solve the gradient of the point positions M and N based on the steepest descent / ascent method. The specific formula is: f = z 实际 - z 基准 Where f is the difference between the actual road surface elevation and the road reference driving surface, z 实际 z represents the actual elevation of the road surface. 基准 Let M be the reference driving surface of the road; let M be the known minimum point in the selected defect area, and let M be the point obtained after k iterations. k P k M represents k The direction of the maximum rate of change of the point surface, d k Represents the gradient; S402、According to the formula of step S305, z is obtained 基准 value, z 基准 The differential of the value formula is solved to obtain the following formula: d k The final expression is: Wherein, N is the number of control points of the reference surface fitting, Q is the number of control points in the actual road surface fitting, M k (x, y) is the starting point of the current iteration; x i , y i is the coordinate of the corresponding control point; r is the distance from M k (x, y) to each control point; Step 3: Solve the elevation of the next point with a step size of 0.01 m and iterate until the extreme point is obtained. The elevation difference between the extreme point and the corresponding xOy coordinate of the reference surface is the disease depth / disease height on the pavement area. Step 4, take 9 points regularly in each disease grid, calculate the average value of its height Take The product of the grid area and the deformation volume V of the grid i The specific formula is: wherein, is the average of the elevations of the nine taken points, S is the area of the grid 4 cm 2 ; V i is the volume of deformation of the grid in each pavement area, from left to right and then from bottom to top, in which there is a deformation disease. Summing all the grid deformation volumes, i.e. the volumes V of the road surface area that are sunken / raised a , V b ; Step 5: Based on the pavement disease depth / disease height and the pavement area settlement / protrusion volume, complete the three-dimensional evaluation of the road deformation disease.

5. The single-line laser point cloud based highway deformation disease detection system according to claim 4, characterized in that, In the road three-dimensional point cloud acquisition module, the specific steps are as follows: Step 1: Use the radius filtering algorithm to make a circle with the center of each point cloud. If the number of points contained in the circle is greater than a certain value, the point is retained; otherwise, the point is deleted. Use the statistical filtering algorithm to solve the average distance of all points within the domain K for each point cloud, solve the mean μ and variance σ of all average values, set μ+nσ as the threshold value, and n as a specified multiple. Use the threshold value as the screening value to preliminarily remove the noise data of the road point cloud data, and obtain the single-frame point cloud of the road. Determine the pavement point cloud range based on the road cross slope and the lowest pavement elevation, and thus extract the accurate cross section point cloud. Step 2: Use the method of extracting single-frame data outside point cloud based on point cloud coordinates to screen the point cloud 0.5-3 m outside the cross section direction in the single-frame point cloud as the emergency lane point cloud. Fit the emergency lane point cloud in a straight line to form a standard cross section, and insert the single-frame point cloud inside the emergency lane according to the slope of the straight line with a point spacing. Step 3: Based on the single-line radar pose, cross section point cloud, and single-frame point cloud, the point cloud data of continuous frames is spliced through the coordinate transformation from NED to LLA coordinate system to form three-dimensional road point cloud data and reference driving surface three-dimensional point cloud data.

6. The single-line laser point cloud based highway deformation disease detection system according to claim 4, wherein, In the disease area positioning module, the specific steps are as follows: Step 1: Based on the emergency lane point cloud, extract the point cloud plane coordinates, fit the overall line type of the road in a cubic curve manner, divide the curve into 10 m small sections, extract the x and y coordinates at the section points and the normal plane of the curve at the section points, take the points inside the 20 m straight line on the xOy projection of the normal plane, determine the road area between the normal planes, and thus divide the 10 m pavement area on the reference driving surface; the coordinates of the points on the normal plane are denoted as: QD = [[x 01 , y 01 , x 02 , y 02 ],...] wherein x 01 represents the abscissa of the intersection point of the trend curve with the first normal plane, y 01 represents the ordinate of the intersection point of the trend curve with the first normal plane, x 02 represents the abscissa of the inside point of the first normal plane, y 02 represents the ordinate of the inside point of the first normal plane; Step 2: Based on the reference driving surface three-dimensional point cloud data, fit the reference driving surface of the road by using a plane fitting algorithm to obtain the parameters A0, A1, and A2 of the plane equation of the reference driving surface. The specific formula is: z=A0x+A1y+A2 Wherein, A0, A1, A2 are the parameters of the plane about x, y coordinates and the intercept of the plane in z coordinate when x, y coordinates are equal to 0 respectively; Step 3, based on the three-dimensional road point cloud data and the road reference driving surface equation, the elevation difference between the three-dimensional road point cloud data and the corresponding xOy coordinate reference surface is calculated; Step 4, based on the grid method, the xOy coordinates of the road reference surface are divided into 1dm*1dm grid, if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located below the reference surface, then the point is a subsidence disease point, and the grid area where the point is located is a subsidence disease area; if the absolute value of the point cloud elevation difference is greater than the preset threshold value, and the point is located above the reference surface, then the point is a protrusion disease point, and the grid area where the point is located is a protrusion disease area; thereby preliminarily dividing the subsidence disease area Sa and the protrusion disease area Sb; Step 5, based on the subsidence disease area and the protrusion disease area point cloud, the three-dimensional surface equation of the road disease area is fitted by the thin plate spline interpolation method, and the specific formula is: U(x) = r 2 lnr where p(x, y) is any point on the surface, U(x) is the radial basis function, and ||p-p i || represents the distance from point p to a certain control point, and the control points 1, 2, 3, …, N are known, ω i represents the weighting of different radial bases, and m0, m1, m2 are parameters of the plane. Step 6, the control point matrix and the height matrix of the point cloud data are established, and the specific formula is: (1) Control point matrix Wherein, n is the number of control points, the second and third columns represent the (x, y) coordinates of the control points; (2) Height matrix wherein v1 to v n represents the coordinate in the z direction of each control point; Step 7, the radial basis function value of any two control points is calculated, and the specific formula is: where r ij represents the distance between control points i and j, U(r ij ) is the value of the radial basis function corresponding to the distance r ij . Step 8, define the matrix L as: Then the above matrix has the following relationship: Y = L * (ω1,... ω N , m0, m1, m2) T ; Based on the thin plate spline interpolation principle, substitute the condition function about the control point, calculate all parameters of the road three-dimensional surface equation, and complete the interpolation, and the specific matrix is: where ω ij is the weight of the jth radial basis on the ith segment, m i0 ,m i1 ,m i2 are the m0, m1, m2 coefficients on the ith segment.

Images