A method for detecting pavement diseases by line scanning

Abstract

The present application belongs to the field of intelligent detection, and particularly relates to a kind of road surface disease line scanning detection method.The main purpose is to solve how to efficiently and accurately detect road surface disease, so as to timely discover and evaluate road surface disease, improve road life, reduce maintenance cost and ensure driving safety.The line array point cloud data representing road surface shape information is obtained by fixing scanning device at the front end of carrier, and a two-step scanning section fitting information acquisition method is designed accordingly.Furthermore, road surface abnormal points are intelligently identified according to the size of coordinate points deviating from the scanning section fitting straight line.In addition, the identification reading device fixed on the carrier reads the identification arranged at equal intervals on the road shoulder in real time, divides the road surface to be detected into several detection units, and realizes sectional detection.Through clustering and quantitative analysis of abnormal points, the disease area in each detection unit is effectively evaluated.Finally, the threshold discrimination method and reference template comparison method are combined to detect road surface disease anomaly and change before and after.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of intelligent detection, and particularly relates to a line scanning type detection method for road surface diseases. BACKGROUND

[0002] Road surface diseases refer to local damages of road surface such as cracks, potholes, subsidence, rutting, heave, and arching due to construction quality problems or external force, and serious road surface diseases will pose a threat to driving safety, thus leading to serious traffic accidents. Timely detection of road surface diseases can not only ensure driving safety, but also prevent further expansion of road surface diseases, thus improving road life and reducing maintenance costs. However, manual periodic detection or some existing road surface disease detection methods and devices often have the disadvantages of high work intensity, low accuracy, low detection efficiency, and difficulty in comparing changes before and after at the same position. Therefore, it is of great significance to develop an efficient and reliable road surface disease detection method for road surface engineering acceptance, road surface state and aging degree evaluation, and daily maintenance. SUMMARY

[0003] In order to realize efficient detection of road surface diseases, the present application provides a method for detecting road surface diseases by using three-dimensional point cloud processing technology, which can accurately identify, locate and quantify road surface diseases, abnormalities and anomalies, greatly facilitating road surface engineering acceptance, maintenance and repair, and ensuring driving safety. Unlike other detection methods, it has the advantages of convenience and precision.

[0004] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0005] The present application provides a line scanning type detection method for road surface diseases, comprising:

[0006] Vehicle 1: a conventional four-wheel motor vehicle, an identification reading device is fixed at the front end thereof, and is used to read identifications I, II, III, and the like arranged equidistantly on the road shoulder;

[0007] Scanning device 2: installed at the front end of the vehicle, a laser source thereon vertically projects laser light downward to form a laser scanning surface;

[0008] The scanning device obtains y and z coordinates of dense points distributed on the "I" shaped laser line in the width direction of the road surface, and x coordinate is obtained in combination with real-time speed of the vehicle and time stamp of each frame of line array point cloud data, so that line array three-dimensional point cloud data representing road surface information is obtained by line scanning;

[0009] The detection step is:

[0010] (1) performing least square z-y straight line fitting on the line array three-dimensional point cloud data to solve a scanning cross section fitting straight line, and identifying abnormal points representing disease areas according to the deviation distance of each coordinate point on the line array from the fitting straight line;

[0011] (2) The identification is read by the reading device to realize segmented detection, and the abnormal points in each detection unit divided are clustered and quantitatively analyzed to determine the size of the disease area and the number of diseases according to the number of sets obtained by the clustering processing;

[0012] (3) The pavement disease abnormalities and anomalies are detected by threshold method and reference template comparison method.

[0013] In the above scheme, in step (1), two-step solving of pavement fitting straight line:

[0014] First, the z-y straight line fitting based on least square method is performed on all coordinate points of each frame linear array point cloud to obtain fitting straight line

[0015] Second, the distance of each coordinate point to the fitting straight line is calculated, and if the distance is greater than a certain threshold, the point and its neighborhood points are filtered out, and the z-y straight line fitting is performed on the remaining points again to obtain fitting straight line which represents the whole pavement after removing obvious concave-convex accidental factors in the point cloud coordinate system. Wherein, Ay+Bz+C=0 represents the fitting straight line always satisfies the equation Ay+Bz+C=0, wherein A, B and C represent the coefficients of the straight line equation, and z and y represent the vertical and horizontal coordinates in the pavement point cloud data.

[0016] In the above scheme, in step (1), the pavement abnormal points are identified according to the size of the coordinate points deviating from the scanning section fitting straight line

[0017] When a coordinate point p i，j (x i , y i，j , z i，j ) in the linear array point cloud deviates from the scanning section fitting straight line by more than a deviation threshold, the coordinate point is considered to be a pavement disease abnormal point:

[0018] The deviation distance of p i，j (x i , y i，j , z i，j ) from is:

[0019]

[0020] If d i，j ≥T 3σ , p i，j is an abnormal point, and the deviation threshold T​3σ identify the current frame linear array point cloud P i all abnormal points in the current detection unit.

[0021] In the above scheme, the deviation threshold T 3σ Determined by the Rajda criterion: sample the road surface without disease on the road section to be detected, calculate the distance from the coordinate point to the fitting straight line of the scanning section and calculate the standard deviation σ, the deviation threshold T 3σ is 3 times of the standard deviation σ.

[0022] In the above scheme, in step (2), the abnormal point occurrence position is determined by a marked segmented detection mechanism: the identification corresponding serial number is read by the identification reading device fixed on the vehicle, so as to determine that the current detection position is located between which two identifications, and then the abnormal position to the identification can be accurately determined according to the real-time speed information corresponding to the abnormal frame. Specifically, the identifications are arranged at Δx intervals on the road shoulder, and when the reading device is opposite to the identification, the identification corresponding serial number is read, and the specific position of the abnormality is found according to the serial number. For example, it is assumed that the abnormality is identified at time t m , and the time t m between the time t I when the identification I is read and the time t II when the identification II is read, it can be known that the abnormality must occur between the identification I and the identification II; it is assumed that the identification I is read at time t I , the mth frame linear array data appears abnormality, and the abnormality identification time t m corresponds to the mth frame linear array point cloud data after the identification I is read, so the abnormality position can be accurately positioned according to the real-time speed information:

[0023]

[0024] In the formula, v m represents the real-time speed of the vehicle when the mth frame data is obtained, x m_I represents the x-direction distance from the abnormal point to the identification I, v i represents the real-time speed of the vehicle 1 when the i th frame linear array point cloud data is obtained since the identification I is read.

[0025] In the above scheme, in step (2), the abnormal points in each detection unit divided by the identification are clustered, the abnormal points belonging to different disease areas are classified into different sets, and the length, width and maximum fluctuation of each disease area in the detection unit are determined by quantitatively analyzing the sets.

[0026] In the above scheme, the clustering process is specifically:

[0027] a1. Randomly select one abnormal point p from the current detection unit, and construct a set C kAll the unvisited abnormal points in the area with p as the center and r as the radius are classified into the set C k The point p is marked as visited.

[0028] a2. All the abnormal points in the seed point area are traversed, and step 1 is repeated until no new abnormal point is classified into C k The set C is thus constructed. k The construction is completed.

[0029] a3. A new abnormal point is randomly selected from the remaining unvisited points as a new seed point, a new set is constructed, and steps a1 and a2 are repeated until all the points in the detection unit are visited, that is, all the abnormal points are classified into corresponding sets according to the adjacent relationship, and all the abnormal points in the current detection unit are classified into sets {C1, C2, C3, …}, wherein each abnormal point set represents a disease area.

[0030] In the above scheme, the length, width and maximum fluctuation of each disease area are obtained by quantitatively analyzing the sets obtained by clustering. Specifically, principal component analysis is performed on the sets {C1, C2, C3, …}, and the first principal component direction e1 and the second principal component direction e2 are obtained. The distance between the two projection points of e1 at the head and tail is the length L of the disease area, the distance between the two projection points of e2 at the head and tail is the width W of the disease area, and the maximum deviation distance d of each point in the set i，j The maximum value is the maximum fluctuation of the disease area.

[0031] In step (3) of the above scheme, the road disease abnormalities and anomalies are detected in combination with modes A and B.

[0032] Mode A: Threshold method, that is, when the number of diseases or the size of the disease area in the detection unit exceeds the normal allowable value, it is determined that the current detection unit has road disease abnormalities, and the detection unit needs to be checked and positioned to locate the disease position for targeted repair. The size of the disease area includes length, width and maximum fluctuation.

[0033] Mode B: Reference template comparison, specifically, first, pre-scan all the detection units of the entire detection road section, and then, during the actual detection scan, the same processing is performed on each detection unit, the same disease areas in the pre-scan and the actual scan are matched through the position relationship of the centers of gravity, and the changes of the length, width, maximum fluctuation and number of diseases of the same disease area in the pre-scan and the actual scan are compared, so as to judge the changes of the road diseases in the time scale, so as to make maintenance decisions.

[0034] Advantages:

[0035] The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0036] A method for defining road surface position based on line array point cloud is designed by using least square method, two-step solving scanning section fitting straight line, obtaining overall road surface equation expression after removing obvious concave-convex accidental factors; a marking segmented detection mechanism is designed to divide the road section into several detection units for convenient disease positioning; by identifying road surface abnormal points and doing clustering and quantitative analysis, the disease area in each detection unit is effectively evaluated; threshold judgment and benchmark comparison two detection modes are designed to identify road surface disease abnormalities and anomalies.

[0037] Compared with the existing intelligent detection method of road surface disease, the present application has simple arrangement, low calculation complexity, good real-time performance and high detection efficiency, can accurately and comprehensively evaluate road surface disease, can not only distinguish the length, width and fluctuation abnormality of disease area by threshold method, but also can accurately obtain the change amount before and after disease in a period of time by benchmark template comparison method. It can also accurately locate the position of disease abnormality, which is convenient for maintenance. In addition, compared with the conventional image processing detection method based on pixel RGB color value, the present application can avoid the influence of light condition and color difference, and is expected to realize more reliable detection in practical application. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0039] Figure 2 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0040] Figure 3 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0041] Figure 4 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0042] Figure 5 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0043] Figure 6 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0044] Figure 7 The present application is directed to road surface disease detection, based on three-dimensional point cloud processing technology, a unique line scanning detection method is designed, including:

[0045] BRIEF DESCRIPTION OF DRAWINGS

[0046] 1-vehicle, 101-identification reading device, 2-scanning device, 201-laser source, 301-identification I, 302-identification II, 303-identification III. DETAILED DESCRIPTION

[0047] As Figure 1 shown, the present application proposes a road disease line scanning detection method, mainly including a carrier 1 and a scanning device 2.

[0048] The carrier 1 is a conventional four-wheel motor vehicle, and a reading device 101 is fixed at the front end thereof. When the carrier 1 passes, the reading device 101 can read the information of the identification array (identification I301, identification II 302, identification III 303,...) arranged equidistantly on the road shoulder, so as to realize the segmentation of the detected road surface, facilitate the disease detection and positioning.

[0049] The scanning device 2 is also installed at the front end of the carrier, and a laser source 201 installed on the scanning device 2 projects a fan-shaped laser vertically downward to form a laser scanning surface. The scanning device 2 obtains the y, z coordinates of the dense points distributed on the "I" laser line in the road width direction, and the x coordinates of the dense points are obtained in combination with the real-time speed of the carrier and the time stamp of each frame of line array point cloud data (the scanning device coordinate system is defined as: the running direction of the carrier is x, the road width direction is v, and the vertical direction of the carrier is z). Thus, the line array three-dimensional point cloud data {P1{(x1,y 1，1 , z 1，1 ), (x1,y 1，2 , z 1，2 )…, (x1,y 1,n , z l，n )}, P2{(x2,y 2，1 , z 2，1 ), (x2,y 2，2 , z 2，2 )…, (x2,y 2，n , z 2，n )},…P i {(x i , y i，1 , z i，1 ), (x i , y i，2 , z i，2 )…, (x i , y i，n , z i，n )},…} representing the road surface information can be obtained in a line scanning manner. The speed of the carrier can be obtained by detecting the linear speed or rotating speed of the wheel through a special speed sensor, or can be directly obtained in real time by communicating with the instrument of the carrier 1.

[0050] Based on the line scanning data acquisition mode, a two-step scanning section fitting information acquisition method is designed.

[0051] First step: as Figure 2As shown in (a), a least-squares straight line fitting is performed on all coordinate points in the linear point cloud obtained by the current scan to obtain the fitted straight line.

[0052] Step 2: Calculate the distance from each point to the fitted line. The distance, if some point p i，j arrive If the distance is greater than the threshold T, it indicates that the point is a significant pit or protrusion. Therefore, this point and several points in its neighborhood ( Figure 2 (a) All points within the rectangular frame shown are removed from the linear array, as follows: Figure 2 As shown in (b), the remaining points are refitted with the zy line to obtain a fitted line that can characterize the current scanning section. Ay + Bz + C = 0. This straight line represents the overall road surface in the point cloud coordinate system after removing obvious random factors such as unevenness and concavity. Ay+Bz+C=0 represents the fitted straight line The equation Av + ​​Bz + C = 0 is always satisfied, where A, B, and C represent the coefficients of the straight line equation, and z and y represent the vertical and horizontal coordinates in the road surface point cloud data.

[0053] Furthermore, after obtaining the fitted straight line of the scanning section corresponding to the current frame... Subsequently, a road surface anomaly identification mechanism was proposed based on the acquired linear point cloud characteristics. This mechanism utilizes the acquired first frame linear point cloud data P. i {(x i y i，1 , z i，1 ), (x i y i，2 , z i，2 )...,(x i y i，n , z i，n For example, let's take} as an example. Figure 3 As shown, calculate the coordinates of all points p in the linear point cloud. i，1 (x i y i，1 , z i，1 )~p i，n (x i y i，n , z i，n Fitting straight lines from each point in the curve to the scanning section The distance, for example, p i，j (x i y i，j , z i,j )arrive The deviation distance is

[0054]

[0055] If d i，j ≥T 3σ Then p i，j This is an outlier. Threshold T 3σ Based on the Laida criterion, sampling is conducted on the defect-free pavement of the road section to be tested. The deviation distance of each sampling point is calculated according to equation (1), and then the sample standard deviation σ is calculated. According to statistical principles, under defect-free conditions, d i，j <T 3σ The probability is 99.7%, that is, when d i，j ≥T 3σ At that time, p i，j The probability of being misidentified as an anomaly is only 0.3%. Based on this threshold method, the current frame linear point cloud P can be identified. i All anomalies in the data.

[0056] Furthermore, this invention proposes a segmented detection mechanism that divides the road segment into several detection units and achieves precise location of road surface defects. For example... Figure 4 As shown, signs I, II, III... are arranged at equal intervals of Δx on the road shoulder. Figure 1 The labels shown are I301, II302, III303, etc. The labels can be RFID tags, barcodes, QR codes, color codes, etc. A label reading device 101 is fixed on the carrier 1, and the signal receiving direction of this device is coplanar with the laser scanning surface. When 101 is facing the label, it can read the corresponding serial number of the label. The current reading time corresponds to the starting point of the current detection unit, and the next label reading time is the ending point of that detection unit. Assume that the reading time of a certain frame of linear array point cloud data is at the time of reading label I and reading label II (t...). I and t II The data frame is between 1 and 2, and this frame is the m-th frame after the identifier I was read, with the reading time being t. m Based on the real-time speed information, the location of the anomaly can be accurately pinpointed.

[0057]

[0058] In the formula v m x represents the real-time speed of vehicle 1 when the data of frame m is acquired. m_I v represents the x-axis distance from the outlier to the identifier I. i This represents the real-time speed of vehicle l when it starts calculating and acquiring the i-th frame of linear point cloud data from the moment it reads identifier I.

[0059] Furthermore, based on the above segmentation mechanism and combined with the aforementioned anomaly identification mechanism (see...), Figure 3 It can identify all abnormalities in each detection unit. Figure 5(a) shows all the anomalies in the detection units divided by identifier I and identifier II. Next, all the anomalies in each detection unit are clustered, that is, anomalies belonging to different disease areas are grouped into different sets.

[0060] The clustering process is as follows:

[0061] 1. Randomly select one anomaly point p from the current detection unit, and construct a set C using this point as the seed point. k All unvisited outliers within a region centered at p and with radius r are grouped into set C. k Mark point p as visited; (e.g.) Figure 5 (b) As shown, all outliers in the neighborhood of point p are grouped into set C1.

[0062] 2. Traverse all outliers in the seed point region described in step 1, repeating step 1 until no new outliers are added to C. k Therefore, set C k Construction complete; (e.g.) Figure 5 (c) shows that set C1 has been constructed.

[0063] 3. From the remaining unvisited points, randomly select one more anomaly point as a new seed point to construct a new set. Repeat steps 1 and 2 until all points in the detection unit have been visited, meaning all anomalies have been assigned to their corresponding sets based on proximity. For example... Figure 5 As shown in (d), all abnormal points in the current detection unit are assigned to sets (C1, C2, C3...), and each set of abnormal points represents a diseased area.

[0064] Furthermore, quantitative analysis is performed on the clustered sets to determine the length, width, and maximum fluctuation of each diseased region. Specifically: such as... Figure 6 As shown, principal component analysis is performed on the aforementioned sets (C1, C2, C3...) to obtain the first principal component direction e1 and the second principal component direction e2. Projecting each point in the set onto these two principal component directions, the distance between the first and last projection points of e1 is the length L of the diseased area, and the distance between the first and last projection points of e2 is the width W of the diseased area. The maximum deviation distance of each point in the set, determined by equation (1), is the maximum fluctuation of the diseased area at that location.

[0065] Furthermore, based on the aforementioned quantitative analysis of road defects, abnormalities and variations in road surface defects are detected through Mode A and Mode B:

[0066] Mode A: Perform the aforementioned clustering and quantitative analysis on the detection units to determine the number of defects (the number of sets obtained from the aforementioned clustering process) and the size of the defect area (length, width, and maximum undulation). Use thresholds to identify whether the current detection unit is abnormal. If the number of defects in a detection unit exceeds the maximum allowable value, or the length, width, or maximum undulation of a defect area exceeds the allowable value, it indicates that the current detection unit has abnormal road surface defects. The detection unit needs to be inspected, and the location of the defects should be determined by combining the distance from the abnormal point to the marker, and targeted repairs should be carried out.

[0067] Mode B: To accurately and effectively determine abnormal changes in pavement defects, the changes in defects before and after are determined by comparing with a pre-scanning baseline. Specifically, firstly, when the pavement is completed or if the pavement condition is compliant, vehicle 1 pre-scans the entire inspection section, using the aforementioned defect area quantitative analysis method to determine the size of each defect area in the inspection unit. During subsequent actual inspections, each inspection unit is scanned again and processed in the same manner. For example... Figure 7 As shown, taking the detection units divided by markers I and II as an example, the set C corresponding to each disease is calculated during actual detection and pre-scanning. k The centroid position (i.e., C) k The average three-dimensional coordinates of all coordinate points in the image are used to determine the extent of proximity of the centroids. This allows for the identification of identical disease areas from previous and subsequent scans. By comparing the changes in length, width, and maximum undulation of the same disease area before and after the scans, a significant change in a particular value indicates a substantial expansion of the disease at that location. For example, Figure 7 During the rescan, both the length L and width W of the defect area C2 were significantly increased compared to the pre-scan, requiring timely inspection and repair to prevent further deterioration and potential safety hazards. Furthermore, the aforementioned clustering process allows for comparison of the number of defects between the pre- and post-scan phases. A change in the number indicates the emergence of new defect areas, and timely repair will effectively control the defects to a lesser extent, reducing maintenance costs.

[0068] The above-mentioned road surface defect detection line scanning method can selectively or simultaneously operate in modes A and B: mode A is used to identify road surface defect anomalies, while mode B is used to detect changes in road surface defects before and after.

[0069] This invention discloses a line-scanning method for detecting pavement defects based on three-dimensional point clouds, the main technical effects and advantages of which are as follows:

[0070] 3D point cloud data acquisition: By installing a scanning device at the front of the vehicle, road surface point cloud data is acquired, including the y and z coordinates of the coordinate points, and the x coordinate is obtained by combining the vehicle speed and timestamp. This method can acquire 3D information about the road surface accurately, in real-time, and economically, providing a data foundation for damage detection.

[0071] Two-step scan section fitting: By least squares method, the point cloud data is fitted with a straight line, and the obvious concave and convex points are removed in two steps to obtain a more accurate road surface fitting straight line. This method can reduce the influence of accidental factors on the detection results and improve the accuracy of detection.

[0072] Abnormal point identification: By calculating the distance of each point in the point cloud to the fitted straight line and comparing it with the threshold, the abnormal points are intelligently identified. This method can effectively identify the disease area of the road surface.

[0073] Marked segmented detection mechanism: By reading the marks on the road shoulder, the road surface is divided into multiple detection units to realize segmented detection. This method can accurately locate the position of the disease area, which is convenient for subsequent repair and maintenance.

[0074] Disease area quantitative analysis: The abnormal points are clustered to obtain the collection of different disease areas, and then the size of the disease area is quantified by principal component analysis. This method can provide detailed information of the disease area to support the repair decision.

[0075] Disease detection mode: Combined with threshold discrimination method and reference template comparison method, the road surface disease anomaly and abnormality are detected. This method can comprehensively evaluate the health status of the road surface and monitor the development trend of the disease.

[0076] In summary, the present application provides an efficient and accurate road surface disease detection method, which can timely discover and evaluate the road surface disease, provide important reference for road maintenance, improve the road life, reduce the maintenance cost, and ensure the driving safety.

Claims

1. A method for detecting road surface defects by scanning lines, characterized in that, include: Vehicle 1: Motor vehicle, with a sign reading device (101) fixed at its front end for reading signs that are equidistantly arranged on the shoulder; Scanning device 2: Installed at the front end of the carrier (1), the laser source (201) on it projects laser light vertically downward to form a laser scanning surface; The scanning device (2) acquires the y and z coordinates of the dense points distributed on the "one" laser line along the road width direction, while the x coordinate is obtained by combining the real-time speed of the vehicle (1) and the timestamp of each frame of linear array point cloud data. Thus, the linear array three-dimensional point cloud data representing road information is acquired by line scanning. The testing steps are as follows: (1) The least squares method is used to fit the line of the scan section by the linear array three-dimensional point cloud data, and the abnormal points of the disease area are identified based on the deviation distance between each coordinate point on the linear array and the fitted line. (2) The segmented detection is achieved by reading the identifier through the reading device (101), and the abnormal points in each detection unit are clustered and quantitatively analyzed to determine the size of the disease area and the number of diseases based on the number of sets obtained by the clustering process. (3) Detect pavement distress anomalies and changes using two methods: threshold method and benchmark template comparison method; In step (2), the location of the anomaly is determined by a segmented detection mechanism: the corresponding serial number of the marker is read by the marker reading device (101) fixed on the vehicle (1), and the current detection location is determined between which two markers. Combined with the anomaly frame and its corresponding real-time speed information, the distance from the anomaly location to the marker can be accurately determined. Specifically, markers are arranged at equal intervals of Δx on the road shoulder. When the reading device (101) is facing the marker, the corresponding serial number of the marker is read, and the specific location of the anomaly is found according to the serial number. Anomaly detection at time t m And at that moment t m At time t when reading identifier I (301) and reading identifier II (302) I and t II Between these two, it can be concluded that the anomaly must occur between identifier I (301) and identifier II (302), and if t m At times t when reading identifier Ⅰ (302) and reading identifier III (303) II and t III From this, it can be deduced that the anomaly must occur between identifier Ⅱ (302) and identifier III (303), and so on; Let the anomaly frame occur between identifier s and identifier s+1, where s = I, II, III..., from t s The calculation begins when the identifier 's' is read. If an anomaly occurs in the linear array data of the m-th frame, the anomaly identification time is 't'. m Corresponding to the m-th frame of linear point cloud data read after the identifier s, the location of the anomaly can be accurately located based on the real-time velocity information: (2) In the formula This represents the real-time speed of vehicle (1) when acquiring data in frame m. t1 and v1 represent the time of the first frame from when tag s is read and the real-time speed of vehicle 1, respectively. m_s This represents the x-axis distance from the anomaly point to the identifier 's'. This represents the real-time speed of vehicle 1 when it starts calculating and acquiring the i-th frame of linear point cloud data from the moment it reads the identifier s; In step (3), both modes A and B are used to detect road surface defects and abnormalities; Mode A: Threshold method judgment, that is, when the number of defects or the size of the defect area in the detection unit exceeds the normal allowable value, it is determined that there is an abnormality of road defects in the current detection unit. The detection unit needs to be inspected, and the location of the defect is located by combining the distance from the abnormal point to the marker for targeted repair. The size of the defect area includes length, width and maximum undulation. Mode B: Baseline template comparison. Specifically, a pre-scan of all detection units in the entire inspection section is first completed. Based on the number of defects and the size of each defect area in the detection unit, the same processing is performed on each detection unit again during the subsequent actual inspection scan. The same defect areas in the previous and subsequent scans are matched by the centroid position relationship, and the changes in length, width, maximum undulation, and number of defects corresponding to the same defect area are compared. This allows for the determination of the changes in road surface defects over time, so as to make corresponding maintenance decisions.

2. The method for detecting road surface defects by scanning according to claim 1, characterized in that: In step (1), the road surface fitting line is solved in two steps: The first step is to perform a least-squares-based line fitting on all coordinate points of each frame's linear point cloud to obtain the fitted line. ; The second step is to calculate the distance from each coordinate point to the fitted line. If the distance is greater than a certain threshold, the point and its neighboring points are filtered out, and the remaining points are refitted with a straight line to obtain the fitted straight line representing the current scanning section. To determine the representation of the overall road surface in the point cloud coordinate system after removing obvious unevenness and random factors, where, Ay+Bz+C=0 represents the fitted straight line. The equation Ay+Bz+C=0 is always satisfied, where A, B, and C represent the coefficients of the straight line equation, and z and y represent the vertical and horizontal coordinates in the road surface point cloud data.

3. The method for detecting road surface defects by scanning according to claim 2, characterized in that: In step (1), a straight line is fitted based on the deviation of the coordinate point from the scanned section. Identify road surface anomalies by size: When a certain coordinate point p in the linear point cloud i,j (x i , y i,j , z i,j (Fit line to scan section) If the distance exceeds the deviation threshold, the coordinate point is considered an abnormal point of pavement distress. p i,j (x i , y i,j , z i,j )arrive The deviation distance is: （1） If d i,j ≥T 3σ Then p i,j An outlier is identified based on the deviation from the threshold T. 3σ Identify the current frame linear point cloud P i All anomalies in the data.

4. The method for detecting road surface defects by scanning according to claim 3, characterized in that: Deviation from threshold T 3σ Determined by the Laida criterion: Samples are taken from the pavement of the section to be inspected that is free of defects. The distance from the coordinate point to the fitted line of the scanning section is calculated, and the standard deviation σ is calculated. The deviation from the threshold T is determined. 3σ It is 3 times the standard deviation σ.

5. The method for detecting pavement defects by scanning according to any one of claims 1 or 3, characterized in that: In step (2), the abnormal points in each detection unit are clustered, and the abnormal points belonging to different disease areas are assigned to different sets. The sets are then quantitatively analyzed to determine the length, width and maximum fluctuation of each disease area in the detection unit.

6. The method for detecting road surface defects by scanning according to claim 5, characterized in that: The clustering process is as follows: a1. Randomly select one anomaly point p from the current detection unit, and construct a set C using this point as the seed point. k All unvisited outliers within a region centered at p and with radius r are grouped into set C. k Mark point p as visited; a2. Traverse all outliers in the seed point region described in step 1, repeating step 1 until no new outliers are added to C. k Therefore, set C k Construction complete; a3. Randomly select one more abnormal point from the remaining unvisited points as a new seed point, construct a new set, and repeat steps a1 and a2 until all points in the detection unit have been visited, that is, all abnormal points are assigned to the corresponding set according to their proximity relationship. All abnormal points in the current detection unit are assigned to sets {C1, C2, C3...}, where each abnormal point set represents a diseased area.

7. The method for detecting road surface defects by scanning according to claim 6, characterized in that: Quantitative analysis is performed on the clusters to determine the length, width, and maximum fluctuation of each diseased region. Specifically, principal component analysis is performed on the aforementioned clusters {C1, C2, C3...} to obtain the first principal component direction e1 and the second principal component direction e2. Each point in the cluster is projected onto these two principal component directions. The distance between the first and last projection points of e1 is the length L of the diseased region, and the distance between the first and last projection points of e2 is the width W of the diseased region. The deviation distance of each point in the cluster is also determined. The maximum value is the maximum fluctuation of the diseased area.

Images