A method for rapid identification and detection of pothole volume for road repair

Abstract

The present application relates to the field of road pit volume identification, in particular to a pit groove volume rapid identification and detection method for road repair;The three-dimensional point cloud data containing the road surface groove region obtained is preprocessed to reduce the point cloud noise, then the suspected upper edge points and suspected lower edge points are obtained based on the horizontal angle threshold and the depth threshold, and then the complete groove region edge is obtained by the bidirectional edge search method of the suspected upper edge points and suspected lower edge points, the randomness of the determination region edge of the conventional method is reduced, the integrity and accuracy of the groove segmentation are improved, finally the groove point cloud data is segmented to obtain the slice point cloud, and the slice point cloud boundary is obtained based on the Alpha shapes operator, so that the cross-sectional area and volume of each slice point cloud are calculated to obtain the volume of the whole groove;Solve the problem of low measurement accuracy of road groove volume in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of road pit volume identification, in particular to a pit groove volume rapid identification and detection method for road repair. BACKGROUND

[0002] The road groove is a pit groove formed by the loss of road material in the local area of the road surface due to the factors such as the accumulated vehicle load, the aging of road material and the erosion of rain and snow, which seriously affects the service life, function and use feeling of the road, and even can cause traffic accidents. The existing detection methods for the road groove are mostly limited to the measurement of the groove depth and the area of the road surface, and a small amount of measurement of the groove volume is limited by the irregular shape of the groove and the randomness of the detection environment, and the measurement accuracy of the volume is low. SUMMARY

[0003] In view of the defects of the prior art, the present application provides a pit groove volume rapid identification and detection method for road repair, which solves the problem of low measurement accuracy of the road groove volume in the prior art.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] A pit groove volume rapid identification and detection method for road repair, the method comprising the following steps:

[0006] S1, obtaining three-dimensional point cloud data containing a road groove area;

[0007] S2, preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud, and respectively obtaining suspected upper edge points and suspected lower edge points based on the horizontal angle threshold and the depth threshold of the point cloud in the preprocessed point cloud; for the upper edge points of the pit, the angle between the vector values of the two adjacent points in the vertical direction and the horizontal plane is close to 0, so that the suspected upper edge points can be determined, and the suspected lower edge points can be determined by further processing the coordinate values of each point in the preprocessed point cloud;

[0008] S3, defining the suspected upper edge points and the suspected lower edge points as seed points, searching the preprocessed point cloud based on the region growing algorithm, and marking the same point cloud contained in the two search results as groove point cloud data;

[0009] S4, cutting the groove point cloud data into several slice point clouds in the depth direction and projecting them onto the horizontal plane, and then performing noise reduction on the slice point clouds by the k-means clustering algorithm to obtain denoised point clouds;

[0010] S5, fitting a plane to each denoised point cloud by an Alpha shapes operator to obtain a slice point cloud boundary;

[0011] S6、Calculate the cross-sectional area of each slice point cloud according to the slice point cloud boundary, and calculate the sum of the volumes of each slice point cloud to obtain the groove volume.

[0012] As preferred, in step S1, specifically comprises the following steps:

[0013] S11, the three-dimensional ground penetrating radar is opposite to the groove area on the road surface;

[0014] S12, get the maximum groove spacing of the groove area in the horizontal direction;

[0015] S12, set the shooting distance of three-dimensional ground penetrating radar and ground to 1-1.5 times of the maximum groove spacing to obtain three-dimensional point cloud data of the groove area.

[0016] As preferred, in step S2, specifically comprises the following steps:

[0017] S21, down-sampling and median filtering the three-dimensional point cloud data to obtain pre-processed point cloud;

[0018] S22, connect the adjacent points in the pre-processed point cloud along the cross section in the driving direction of the road in turn, and calculate the horizontal angle between the connecting line of the adjacent points and the horizontal plane X axis; the calculation formula of the horizontal angle is:

[0019]

[0020] In the above formula, represents the horizontal angle of the connecting line of any two adjacent points i and j and the horizontal plane X axis, and respectively represent the x-axis coordinate and z-axis coordinate of any point i, and respectively represent the x-axis coordinate and z-axis coordinate of any point j;

[0021] S23, set the horizontal angle threshold, and judge whether the horizontal angle corresponding to any two adjacent points is less than or equal to the horizontal angle threshold in turn;

[0022] If yes, mark the two adjacent points as suspected upper edge points;

[0023] If not, go to the next step;

[0024] S24, calculate the depth threshold according to the z-axis coordinate of each point in the pre-processed point cloud; the calculation formula of the depth threshold is:

[0025]

[0026] In the above formula, represents the depth threshold, z-axis coordinate of the i-th point in the pre-processed point cloud, the number of points in the pre-processed point cloud with z-axis coordinate

[0027] S25, selecting a plurality of points with z-axis coordinate less than the depth value threshold from the pre-processed point cloud, and screening a plurality of points with the smallest z-axis coordinate to obtain suspected lower edge points.

[0028] As preferred, in step S21, the following steps are specifically included:

[0029] S211, calculating the normal vector of any point in the three-dimensional point cloud data;

[0030] S212, calculating the normal angle of any two points;

[0031] S213, setting a normal angle threshold, and sequentially judging whether the normal angle of any two points is less than the normal angle threshold;

[0032] If yes, deleting one of the two points;

[0033] If no, retaining the two points;

[0034] S214, marking the retained points as a filtered point cloud, clustering the points according to the direction of the normal vector of the points in the filtered point cloud to obtain a plurality of cluster centroids, and performing down-sampling;

[0035] S215, performing median filtering on the down-sampled filtered point cloud to obtain the pre-processed point cloud. As preferred, in step S4, the following steps are specifically included:

[0036] S41, cutting the groove point cloud data into a plurality of parts along the depth direction, and projecting it onto the horizontal plane to obtain a slice point cloud;

[0037] S42, determining the center point coordinates and k value of each slice by using the k-means clustering algorithm to perform secondary noise reduction on each slice point cloud, then calculating the Euclidean distance of each point in the slice to the center point, and then removing the points in the groove point cloud data with Euclidean distance greater than the distance threshold to obtain a noise-reduced point cloud.

[0038] As preferred, in step S6, the following steps are specifically included:

[0039] S61, obtaining the coordinates of each point in each slice according to the boundary of the projected point cloud, and calculating the cross-sectional area of each slice along the depth direction; the calculation formula of the cross-sectional area is:

[0040]

[0041] In the above formula, ​represents the cross-sectional area of the i-th slice along the depth direction, and the number of points in the i-th slice is m, and respectively represent the horizontal coordinate and the vertical coordinate of the i-th point in the i-th slice;

[0042] S62, calculate the volume of each slice according to the cross-sectional area of each slice along the depth direction and the total depth of the groove; the calculation formula of the volume of each slice is:

[0043]

[0044] In the above formula, represents the volume of the i-th slice, represents the cross-sectional area of the i-th slice along the depth direction, and the number of points in the i-th slice is m, and the total depth of the groove is H;

[0045] S63, calculate the volume of the groove area according to the volume of each slice; the calculation formula of the volume of the groove area is:

[0046]

[0047] In the above formula, represents the volume of the groove area.

[0048] Compared with the prior art, the present application provides a pit volume rapid identification and detection method for road repair, which has the following beneficial effects:

[0049] 1、The road surface groove volume measurement method provided by the present application adopts an alpha shapes operator instead of a conventional RANSAC plane fitting algorithm when fitting to obtain the slice point cloud boundary, avoids errors caused by the use of a conventional RANSAC algorithm because the center of gravity of the point cloud plane is outside the plane, and thus improves the measurement accuracy.

[0050] 2、The road surface pit volume measurement method provided by the present application uses a depth threshold to remove the redundant area of the road when segmenting the groove area, as much as possible to retain the effective part of the groove area, reduces the complexity of the segmentation method, and improves the segmentation speed.

[0051] 3、The road surface groove volume measurement method provided by the present application first obtains the seed point on the upper surface of the pit by calculating the horizontal included angle, and simultaneously determines the seed point on the lower surface using a depth threshold method, and then uses a bidirectional edge search method to obtain the complete edge of the groove area, reduces the randomness of the determination of the area edge in the conventional method, and improves the completeness and accuracy of the groove segmentation. BRIEF DESCRIPTION OF DRAWINGS

[0052] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0053] Figure 1 Flow chart of the method for rapid identification and detection of pit volume for road repair of the present application;

[0054] Figure 2 Three-dimensional point cloud data containing a road groove area of the present application;

[0055] Figure 3 Three-dimensional point cloud data of a partial groove of the present application;

[0056] Figure 4 Groove point cloud data of the present application;

[0057] Figure 5 Sliced point cloud of the groove after plane fitting of the present application. DETAILED DESCRIPTION

[0058] In order to make the above objectives, features and advantages of the present application more apparent, the following will further describe the present application in detail with reference to the accompanying drawings and specific embodiments. The implementation process of how the present application applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented.

[0059] Those of ordinary skill in the art can understand that all or part of the steps of the following embodiments can be completed by programs instructing related hardware, therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product in 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 code.

[0060] In daily maintenance of roads, accurate measurement of the volume of road grooves is of great help to maintain roads, including but not limited to accurately determining the number of capsules used during recent asphalt maintenance, future road maintenance will inevitably be automated, and in the case of reducing labor and maintenance material costs, it is necessary to improve the efficiency of maintenance, therefore, in order to solve the problem of low accuracy of measuring the volume of road grooves in the prior art, the present application provides a method for rapid identification and detection of pit volume for road repair, as shown in Figures 1-5 The method for rapid identification and detection of pit volume for road repair of the present application, as shown in

[0061] S1, acquire three-dimensional point cloud data containing the groove area of the road surface; in order to ensure that the radar can acquire all the point clouds in the groove area of the road surface as much as possible, the relative distance and position of the three-dimensional ground penetrating radar and the ground need to be controlled, in step S1, specifically including the following steps:

[0062] S11, the three-dimensional ground penetrating radar is directly opposite the groove area on the road surface;

[0063] S12, acquire the maximum groove interval of the groove area in the horizontal direction;

[0064] S12, set the shooting distance of the three-dimensional ground penetrating radar and the ground to 1-1.5 times of the maximum groove interval to obtain the three-dimensional point cloud data of the groove area, and when shooting, use the three-dimensional ground penetrating radar to move around the road surface area slightly larger than the groove area and above the pit, so as to collect three-dimensional point cloud data containing the groove and the surrounding part of the road surface;

[0065] S2, pre-process the three-dimensional point cloud data to obtain pre-processed point cloud, and respectively acquire suspected upper edge points and suspected lower edge points based on the horizontal angle threshold and depth threshold of the point cloud in the pre-processed point cloud; for the upper edge points of the pit, the angle between the vector values of the two points adjacent in the vertical direction and the horizontal plane is close to 0, so that the suspected upper edge points can be determined, and the suspected lower edge points can be determined by further processing the coordinate values of each point in the pre-processed point cloud, in step S2, specifically including the following steps:

[0066] S21, downsample and median filter the three-dimensional point cloud data to obtain pre-processed point cloud; in order to further illustrate the way of pre-processing the three-dimensional point cloud data, in step S21, specifically including the following steps:

[0067] S211, calculate the normal vector of any point in the three-dimensional point cloud data;

[0068] S212, calculate the normal angle of any two points;

[0069] S213, set the normal angle threshold, and judge whether the normal angle of any two points is less than the normal angle threshold in turn;

[0070] If yes, delete one of the points;

[0071] If not, keep the two points;

[0072] S214, mark the kept points as filtered point cloud, and cluster the points according to the direction of the normal vector of the points in the filtered point cloud to obtain the centroid of several clusters, and downsample;

[0073] S215, the down-sampled filtered point cloud is subjected to median filtering to obtain a pre-processed point cloud, the specific process of the median filtering being: replacing the value of a center pixel (or point) with the median of pixels (or points) in a neighborhood, thereby effectively smoothing the signal and reducing the influence of noise on the image or point cloud, for each pixel or point, a 3x3 or 5x5 neighborhood window is defined, including the pixel and its surrounding pixels, the values of all pixels (or points) in the neighborhood are extracted, and the median thereof is calculated, the formula being as follows: , wherein, is a center point a set of neighborhood pixels around the center point represents calculating the median of the values in the neighborhood, and replacing the value of the center pixel (or point) with the calculated median.

[0074] S22, adjacent points in the pre-processed point cloud are sequentially connected along the cross section in the driving direction of the road, and the horizontal angle between the connecting line of the adjacent points and the horizontal plane X axis is calculated; the formula for calculating the horizontal angle being:

[0075]

[0076] In the above formula, represents the horizontal angle between the connecting line of any two adjacent points i and j and the horizontal plane X axis, and respectively represent the x-axis coordinate and the z-axis coordinate of any point i, and respectively represent the x-axis coordinate and the z-axis coordinate of any point j;

[0077] S23, a horizontal angle threshold is set, and it is sequentially determined whether the horizontal angle corresponding to any two adjacent points is less than or equal to the horizontal angle threshold; the horizontal angle threshold is generally set to 10°;

[0078] If yes, the two adjacent points are marked as suspected upper edge points;

[0079] If no, the next step is entered;

[0080] S24, a depth threshold is calculated according to the z-axis coordinates of the points in the pre-processed point cloud; the formula for calculating the depth threshold being:

[0081]

[0082] In the above formula, represents the depth threshold, represents the z-axis coordinate of the i-th point in the pre-processed point cloud, represents the number of points in the pre-processed point cloud whose z-axis coordinates are ; and

[0083] S25, select a plurality of points with a z-axis coordinate less than a depth value threshold from the pre-processed point cloud, and select a plurality of points with the smallest z-axis coordinate to obtain suspected lower edge points.

[0084] S3, define the suspected upper edge points and the suspected lower edge points as seed points, search the pre-processed point cloud based on a growing algorithm, and mark the same point cloud contained in the two search results as groove point cloud data;

[0085] The region growing algorithm starts from the selected seed point, discriminates according to the pixel value within the similarity threshold, gradually connects the similar pixels, and ends the region growing under certain conditions. The certain condition is that if the 4-neighborhood or 8-neighborhood pixels of the seed point meet the growing criterion, i.e., the pixel value is within the similarity threshold, then they are merged into the same region with the seed point.

[0086] S4, cut the groove point cloud data into several slice point clouds in the depth direction and project it onto the horizontal plane, and then perform noise reduction on it by the k-means clustering algorithm to obtain a denoised point cloud. In order to further illustrate the generation method of the slice point cloud and the generation method of the denoised point cloud, in step S4, the following steps are included:

[0087] S41, cut the groove point cloud data into several parts along the depth direction, i.e., cut the groove point cloud data along the depth direction and set several horizontal planes at certain intervals, and project it onto the horizontal plane to obtain a slice point cloud;

[0088] S42, perform secondary noise reduction on each slice point cloud by using the k-means clustering algorithm, determine the center point coordinates and k value of each slice, then calculate the Euclidean distance of each point in the slice to the center point, and then remove the points in the groove point cloud data with a Euclidean distance greater than a distance threshold to obtain a denoised point cloud.

[0089] S5, perform plane fitting on each denoised point cloud by the Alpha shapes operator to obtain a slice point cloud boundary; the Alpha shapes shape algorithm is a simple, effective and fast boundary point extraction algorithm, which overcomes the disadvantage of point cloud boundary point shape influence, can quickly and accurately extract boundary points, ensures that the center of gravity of the point cloud plane is in the plane, and thus improves the accuracy of the groove boundary measurement.

[0090] S6, calculating the cross-sectional area of each slice point cloud according to the slice point cloud boundary, and calculating the sum of the volume of each slice point cloud to obtain the groove volume, after determining the corresponding slice point cloud boundary of each slice point cloud, the cross-sectional area of each slice point cloud can be calculated according to the coordinates in the slice point cloud, then the volume of each slice point cloud is calculated according to the cross-sectional area and the height of the slice point cloud, and finally the volume of the groove is calculated, in step S6, the following steps are included:

[0091] S61, obtaining the coordinates of each point in each slice according to the boundary of the projection point cloud, and calculating the cross-sectional area of each slice along the depth direction; the calculation formula of the cross-sectional area is:

[0092]

[0093] In the above formula, represents the cross-sectional area of the i-th slice along the depth direction, the number of points in the i-th slice is m, and respectively represent the horizontal coordinate and the vertical coordinate of the i-th point in the i-th slice;

[0094] S62, calculating the volume of each slice according to the cross-sectional area of each slice along the depth direction and the total depth of the groove; the calculation formula of the volume of each slice is:

[0095]

[0096] In the above formula, represents the volume of the i-th slice, represents the cross-sectional area of the i-th slice along the depth direction, the number of points in the i-th slice is m, and the total depth of the groove is H;

[0097] S63, calculating the volume of the groove region according to the volume of each slice; the calculation formula of the volume of the groove region is:

[0098]

[0099] In the above formula, represents the volume of the groove region.

[0100] The present application reduces the randomness of the determination of the region edge by the conventional method, improves the integrity and accuracy of the groove segmentation, and finally segments the groove point cloud data to obtain the slice point cloud and obtains the slice point cloud boundary based on the Alpha shapes operator, so as to finally obtain the volume of the entire groove by calculating the cross-sectional area and volume of each slice point cloud.

[0101] The above embodiments have been described in detail, and the principles and embodiments of the present application have been described by applying specific examples. The above embodiment is only used to help understand the method and core idea of the present application. Meanwhile, for those skilled in the art, according to the idea of the present application, the specific embodiments and application scope will be changed. In summary, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for rapid identification and detection of pothole volume for road repair, characterized in that, The method comprises the following steps: S1, obtaining three-dimensional point cloud data containing a road groove area; S2, preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud, and respectively obtaining suspected upper edge points and suspected lower edge points based on horizontal angle threshold and depth threshold of the point cloud in the preprocessed point cloud; S3, defining the suspected upper edge points and the suspected lower edge points as seed points, searching the preprocessed point cloud based on a region growing algorithm, and marking the same point cloud contained in two search results as groove point cloud data; S4, longitudinally cutting the groove point cloud data into several slice point clouds in the depth direction and projecting them onto the horizontal plane, and then performing noise reduction on the slice point clouds by a k-means clustering algorithm to obtain denoised point clouds; S5, fitting a plane to each of the denoised point clouds by an Alpha shapes operator to obtain a slice point cloud boundary; S6, calculating the cross-sectional area of each slice point cloud according to the slice point cloud boundary, and calculating the sum of the volumes of each slice point cloud to obtain a groove volume: In step S2, the following steps are specifically included: S21, downsampling and median filtering the three-dimensional point cloud data to obtain the preprocessed point cloud; S22, connecting adjacent points in the preprocessed point cloud in sequence along the cross section in the driving direction of the road, and calculating the horizontal angle of the connecting line between the adjacent points and the horizontal plane X axis; the calculation formula of the horizontal angle is: ； In the above formulae, denotes the horizontal angle between the line connecting any two adjacent points i and j and the horizontal plane X-axis, and denote the x-axis coordinate and z-axis coordinate of any point i, respectively, and denote the x-axis coordinate and z-axis coordinate of any point j, respectively; S23, setting a horizontal angle threshold, and sequentially judging whether the horizontal angle corresponding to any two adjacent points is less than or equal to the horizontal angle threshold; if yes, marking the two adjacent points as suspected upper edge points; if no, proceeding to the next step; S24, calculating a depth threshold according to the z-axis coordinates of the points in the preprocessed point cloud; the calculation formula of the depth threshold is: ； In the above formula, denotes a depth threshold value, denotes the z-axis coordinate of the i-th point in the pre-processed point cloud, denotes the number of points in the pre-processed point cloud whose z-axis coordinate is . S25, selecting several points with z-axis coordinates less than the depth threshold from the preprocessed point cloud, and selecting several points with the smallest z-axis coordinates to obtain suspected lower edge points.

2. The identification and detection method according to claim 1, characterized in that, In step S1, the following steps are specifically included: S11, directing the three-dimensional ground penetrating radar at the groove area on the road surface; S12, obtaining the maximum groove spacing of the groove area in the horizontal plane direction; S12, setting the shooting spacing between the three-dimensional ground penetrating radar and the ground to be 1-1.5 times the maximum groove spacing to obtain the three-dimensional point cloud data of the groove area.

3. The identification and detection method of claim 1, wherein, In step S21, the following steps are specifically included: S211, calculating the normal vector of any point in the three-dimensional point cloud data; S212, calculating the normal angle of any two points; S213, setting a normal angle threshold, and sequentially judging whether the normal angle of any two points is less than the normal angle threshold; if yes, deleting one of the two points; if no, retaining both points; S214, marking the retained points as filtered point clouds, and clustering the points according to the direction of the normal vector of the points in the filtered point clouds to obtain the centroids of several clusters, and performing downsampling; S215, median filtering the filtered point clouds after downsampling to obtain the preprocessed point cloud.

4. The identification and detection method of claim 1, wherein, In step S3, the region growing algorithm is started from the selected seed point, and the similar pixels are connected step by step according to the pixel value within the similar threshold. Under certain conditions, the region growing is ended. The certain conditions are that the 4-neighborhood or 8-neighborhood pixels of the seed point are merged into the same region if the pixels satisfy the growth criterion, that is, the pixel value is within the similar threshold.

5. The identification and detection method of claim 1, wherein, In step S4, the following steps are specifically included: S41, the groove point cloud data is cut into several parts along the depth direction, and is projected onto the horizontal plane to obtain the slice point cloud; S42, the k-means clustering algorithm is used to perform secondary noise reduction on each slice point cloud, the center point coordinates and the k value of each slice are determined, then the Euclidean distance of each point in the slice to the center point is calculated, and finally the points in the groove point cloud data with the Euclidean distance greater than the distance threshold are removed to obtain the denoising point cloud.

6. The identification and detection method of claim 1, wherein, In step S6, the following steps are specifically included: S61, the coordinates of each point in each slice are obtained according to the boundary of the projected point cloud, and the cross-sectional area of each slice along the depth direction is calculated; the calculation formula of the cross-sectional area is: ； In the above formulae, denotes the cross-sectional area of the i-th slice along the depth direction, the number of points in the i-th slice is m, and denote the abscissa and ordinate of the i-th point in the i-th slice, respectively; S62, the volume of each slice is calculated according to the cross-sectional area of each slice along the depth direction and the total depth of the groove; the calculation formula of the volume of each slice is: ； In the above formula, represents the volume of the i-th slice, represents the cross-sectional area of the i-th slice along the depth direction, the number of points in the i-th slice is m, and the total depth of the groove is H. S63, the volume of the groove region is calculated according to the volume of each slice; the calculation formula of the volume of the groove region is: ； In the above formula, represents the volume of the recess region.

Images