A municipal engineering road pavement detection method, device and system

Abstract

This application relates to the field of road surface texture depth detection technology, specifically to a method, equipment, and system for detecting road surface texture in municipal engineering projects. The method includes: analyzing the similarity between feature descriptors of all data points in each cluster and combining this with noise feature values ​​to denoise each cluster and obtain denoised clusters; analyzing the average distribution and dispersion of point cloud density of all data points in each denoised cluster and combining this with brightness feature values ​​to obtain data missing clusters; and filling in missing values ​​for all data missing clusters to detect the texture depth of the road surface texture in the municipal engineering project to be tested. This application solves the problems of missing point cloud data and noise interference in the three-dimensional point cloud of the road surface texture in the municipal engineering project to be tested, thus improving the detection accuracy of the texture depth of the road surface texture in municipal engineering projects.

Description

Technical Field

[0001] This application relates to the field of road pavement texture depth detection technology, specifically to a method, equipment and system for detecting road pavement in municipal engineering. Background Technology

[0002] Municipal roads are an important part of urban transportation, and their construction quality directly affects driving safety. Asphalt is one of the most commonly used materials in municipal road construction, mainly used for paving asphalt pavements. This is because asphalt pavements have advantages such as wear resistance, anti-skid properties, short construction period, and low cost. During the acceptance and service of municipal road pavements, the depth of the pavement structure is an important pavement inspection indicator, providing a reference for road maintenance departments regarding the anti-skid performance of the pavement.

[0003] Among existing methods for measuring pavement texture depth, line structure light 3D scanning technology using intelligent 3D laser profile sensors based on the principle of laser triangulation can measure pavement texture depth more efficiently, quickly, and comprehensively compared to traditional sand-laying methods and digital image methods. However, since asphalt pavement is composed of asphalt, mineral powder, and material particles, it usually appears black or dark brown. When the color of the asphalt pavement is too dark, it can easily lead to data loss in the 3D point cloud data collected by the intelligent 3D laser profile sensor, and can also generate noisy data, thereby reducing the detection accuracy of pavement texture depth in municipal engineering projects. Summary of the Invention

[0004] In a first aspect, embodiments of this application provide a method for detecting road surface in municipal engineering projects, the method comprising the following steps:

[0005] A three-dimensional point cloud of the road surface of the municipal engineering project under test is obtained by using an intelligent three-dimensional laser contour sensor. Each data point in the three-dimensional point cloud contains three-dimensional coordinates and brightness values.

[0006] Cluster all data points in the 3D point cloud, analyze the difference between the mean brightness value of all data points in each cluster and the maximum brightness value in the 3D point cloud, and determine the brightness feature value of each cluster; perform plane fitting on all data points in each cluster, and filter feature points from all data points in each cluster by analyzing the distance of all data points to their fitting plane; based on the distance between all feature points in each cluster and the fitting error in the plane fitting process, and combined with the brightness feature value, determine the noise feature value of each cluster;

[0007] By analyzing the similarity between the feature descriptors of all data points in each cluster and combining them with the noise feature values, denoising is performed on each cluster to obtain a denoised cluster.

[0008] The average distribution and dispersion of point cloud density of all data points in each denoised cluster are analyzed to determine the density feature value of each denoised cluster. Combined with the brightness feature value, the missing data clusters are screened out from all denoised clusters of the 3D point cloud.

[0009] Missing values ​​are filled into all missing clusters in the 3D point cloud to obtain a complete 3D point cloud of the road surface of the municipal engineering project to be tested, so as to detect the structural depth of the road surface of the municipal engineering project to be tested.

[0010] Preferably, the brightness feature value of each cluster is the ratio of the average brightness value of all data points in each cluster to the maximum brightness value in the three-dimensional point cloud.

[0011] Preferably, the step of filtering feature points from all data points in each cluster includes:

[0012] The distance from all data points in each cluster to the fitted plane is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. Data points whose distance is greater than the segmentation threshold are recorded as feature points.

[0013] Preferably, the method for determining the noise feature values ​​of each cluster is as follows:

[0014] Calculate the mean distance between all feature points in each cluster, calculate the product of the normalized value of the mean distance and the normalized value of the fitting error, and use the ratio of the product to the normalized value of the corresponding brightness feature value as the noise feature value of each cluster.

[0015] Preferably, obtaining the filtered clusters includes:

[0016] The filter window length W of cluster a a The expression is: W a =f[norm(S) a ×Q a [)×W1]+W2; where S a Q represents the normalized value of the noise eigenvalues ​​of cluster a; a W1 represents the normalized mean of the similarity between the feature descriptors of all data points in cluster a; W2 represents the preset base window length; norm() represents the normalization function; f[] is used to output the largest even number not greater than the input value.

[0017] All data points in each cluster are used as input to the filtering algorithm, where the filtering window length of each data point is set to the filtering window length of its respective cluster, and the filtered clusters are output.

[0018] Preferably, the density characteristic value of each denoised cluster is the result of the normalized value of the point cloud density dispersion of all data points in each denoised cluster divided by the normalized value of the mean point cloud density.

[0019] Preferably, the step of filtering out missing data clusters from all denoised clusters of the 3D point cloud includes:

[0020] The ratio of the normalized density feature value to the normalized brightness feature value of each denoised cluster is used as the missing data feature value of each denoised cluster.

[0021] The missing data feature values ​​of all denoised clusters in the 3D point cloud of the road surface of the municipal engineering project to be tested are used as the input of the threshold segmentation algorithm. The output segmentation threshold is recorded as the missing threshold. The denoised clusters with missing data feature values ​​greater than the missing threshold are regarded as missing data clusters.

[0022] Preferably, the detection of the structural depth of the road surface in the municipal engineering project to be tested includes:

[0023] Choose any point on the road surface of the municipal engineering project to be measured as the origin, take the direction perpendicular to the plane of the road surface of the municipal engineering project to be measured as the Z axis, and arbitrarily determine the positive directions of the X axis and Y axis by right hand;

[0024] Calculate the area of ​​the projection plane of the complete 3D point cloud onto the XOY plane; perform 3D reconstruction of the complete 3D point cloud to obtain a triangular mesh model; calculate the volume of the convex polyhedron formed by all triangular facets in the triangular mesh model and the projection plane of the complete 3D point cloud onto the XOY plane; and use the ratio of the volume to the area as the construction depth of the road surface of the municipal engineering project to be measured.

[0025] Secondly, embodiments of this application also provide a municipal engineering road surface testing device, wherein the device stores a computer program, and when the computer program is executed by a processor, it implements a municipal engineering road surface testing method as described above.

[0026] Thirdly, embodiments of this application provide a municipal engineering road surface detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement any of the aforementioned municipal engineering road surface detection methods.

[0027] As can be seen from the above embodiments, the municipal engineering road surface detection method provided in this application has at least the following characteristics:

[0028] Beneficial effects:

[0029] This application accurately assesses the noise interference experienced by different point cloud data in the acquired 3D point cloud data and the richness of the pavement texture in the municipal engineering road surface area to be measured. This allows for the selection of an appropriate filter window size in the mean filtering algorithm for different data points in the acquired 3D point cloud data. This avoids the situation where traditional filtering algorithms apply the same filtering process to all data points in the acquired 3D point cloud data, resulting in over-smoothing of normal point cloud data and loss of the true texture details of the asphalt pavement. This reduces the impact of noise data on subsequent 3D reconstruction, thereby improving the detection accuracy of the pavement texture depth in municipal engineering roads. Furthermore, this application analyzes various... By analyzing the average distribution and dispersion of point cloud density of all data points in the denoised clusters, combined with brightness feature values, missing data clusters were selected from all denoised clusters of the 3D point cloud. Missing data processing methods were then used to supplement the missing data in the selected point cloud data, improving the detection accuracy of the pavement structure depth in municipal engineering projects. Furthermore, this application uses the 3D point cloud after filtering and missing data filling processing to detect the pavement structure depth of the municipal engineering project under test. This effectively reduces the impact of missing data and noise data caused by pavement materials in the collected 3D point cloud on the detection of pavement structure depth, thus improving the detection accuracy of the pavement structure depth in municipal engineering projects. Attached Figure Description

[0030] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 A flowchart illustrating the steps of a method for detecting road surface in municipal engineering, as provided in one embodiment of this application;

[0032] Figure 2 This is a schematic diagram illustrating the process of obtaining the filter window length according to an embodiment of this application. Detailed Implementation

[0033] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a municipal engineering road surface testing method, equipment, and system proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0035] The following description, in conjunction with the accompanying drawings, details a specific scheme for a method, equipment, and system for detecting road surface in municipal engineering provided in this application.

[0036] Please see Figure 1 The diagram illustrates a flowchart of a method for detecting road surface in municipal engineering according to an embodiment of this application. The method includes the following steps:

[0037] S1: Use an intelligent 3D laser profile sensor to acquire the 3D point cloud of the road surface of the municipal engineering project to be tested. Each data point in the 3D point cloud contains 3D coordinates and brightness values.

[0038] This embodiment uses a smart 3D laser profile sensor to acquire 3D point cloud data and brightness map data of the road surface of the municipal engineering project to be measured. The data points in the 3D point cloud data correspond one-to-one with the data points in the brightness map. The data points in the 3D point cloud data are transformed from the coordinate system of the smart 3D laser profile sensor to the world coordinate system. A point on the road surface of the municipal engineering project to be measured is randomly selected as the origin, and the direction perpendicular to the plane of the road surface is taken as the Z-axis. The positive directions of the X-axis and Y-axis are arbitrarily determined by right-hand rule. The unit length in the world coordinate system is set to 1mm, which can be set by the implementer.

[0039] The world coordinate system is a well-known technology. The process of transforming the coordinate system of data points in a 3D point cloud to the world coordinate system is also a well-known technology, and its specific transformation process will not be described in detail.

[0040] It should be noted that coordinate system transformation is used to align point cloud data with the world coordinate system. Coordinate system alignment does not change the geometry of the object, but only changes the object's reference relationship in space. Therefore, the 3D point cloud after coordinate system transformation is still the 3D point cloud of the road surface of the municipal engineering project to be measured.

[0041] Thus, a three-dimensional point cloud of the road surface of the municipal engineering project to be tested was obtained. The data points in the three-dimensional point cloud contain their spatial three-dimensional coordinates and brightness values.

[0042] S2: Cluster all data points in the 3D point cloud, analyze the difference between the mean brightness value of all data points in each cluster and the maximum brightness value in the 3D point cloud, and determine the brightness feature value of each cluster; perform plane fitting on all data points in each cluster, and filter feature points from all data points in each cluster by analyzing the distance of all data points to their fitting plane; based on the distance between all feature points in each cluster and the fitting error in the plane fitting process, and combined with the brightness feature value, determine the noise feature value of each cluster.

[0043] Because the acquired 3D point cloud data may contain missing data and noise due to differences in road surface color in the municipal engineering road surface under test, in order to reduce the interference of missing data and noise data on the true road surface texture of the municipal engineering road surface under test, the acquired 3D point cloud data is usually filtered and missing data is supplemented sequentially, such as by using mean filtering and linear interpolation. However, these methods usually apply the same processing to all data points in the acquired 3D point cloud data, which can easily cause the point cloud data that has not been introduced with noise and has not had missing data to be over-smoothed, thereby losing the true texture details of the asphalt pavement, and thus affecting the accuracy of subsequent detection of the surface texture depth of the municipal engineering road surface under test. Therefore, to avoid this situation, the following processing is performed:

[0044] (1) Cluster all data points in the 3D point cloud, analyze the difference between the mean brightness value of all data points in each cluster and the maximum brightness value in the 3D point cloud, and determine the brightness characteristic value of each cluster.

[0045] Since the main component of the road surface of the municipal engineering project to be tested is asphalt, the darker the asphalt, the greater the degree to which the incident laser of the intelligent 3D laser profile sensor is absorbed by the road surface, which makes it more likely that there will be missing data and noise data in the point cloud data collected by the intelligent 3D laser profile sensor. The darker the asphalt surface, the lower its brightness usually is.

[0046] Therefore, based on the above analysis, this embodiment determines the brightness feature value of each cluster by clustering all data points in the 3D point cloud and analyzing the difference between the mean brightness value of all data points in each cluster and the maximum brightness value in the 3D point cloud. Specifically:

[0047] In this embodiment, all data points in the 3D point cloud of the road surface of the municipal engineering project to be tested are used as input to the clustering algorithm, and multiple clusters are output to represent the set of all data points corresponding to each road surface area with different brightness in the world coordinate system.

[0048] It should be noted that there are many commonly used clustering algorithms. In this embodiment, the DBSCAN clustering algorithm is used to cluster the data points. In this embodiment, when using this clustering algorithm, the metric distance is set to the Euclidean distance between the three-dimensional coordinates of the data points, the neighborhood radius is 2 units in the world coordinate system of the three-dimensional point cloud, and MinPts is set to 10. In actual applications, as other implementation methods, the parameters in the above clustering algorithm can be set by the implementer according to the specific situation. This embodiment does not impose any special restrictions. In addition, the implementer can also use other clustering methods such as the DPC density peak clustering algorithm according to the specific situation. This embodiment does not impose any special restrictions on the selection of clustering algorithms.

[0049] The DBSCAN clustering algorithm and the Euclidean distance calculation method are well-known techniques, and their specific principles will not be elaborated here.

[0050] Furthermore, in this embodiment, the ratio of the average brightness value of all data points in each cluster to the maximum brightness value in the 3D point cloud is used as the brightness feature value of each cluster.

[0051] Based on the brightness feature values ​​of each cluster, it can be understood that the brightness feature value is used to characterize the brightness of the average brightness of each cluster relative to the brightest area in the entire 3D point cloud. It is a normalized measure that eliminates the influence of absolute brightness differences under different scenes. If the brightness feature value of the current cluster is larger, it means that the average brightness of the cluster is relatively high, which means that the possibility of missing point cloud data and noise interference in the cluster is smaller, and the data quality is relatively high. Conversely, if the brightness feature value of the current cluster is smaller, the possibility of missing point cloud data and noise interference is greater.

[0052] (2) Perform plane fitting on all data points in each cluster. By analyzing the distance of all data points to their fitting plane, feature points are selected from all data points in each cluster. Based on the distance between all feature points in each cluster and the fitting error in the plane fitting process, and combined with the brightness feature value, the noise feature value of each cluster is determined.

[0053] Under normal circumstances, the brightness change of the road surface of the municipal engineering project under test is relatively gradual. Therefore, the data points in the collected 3D point cloud have a relatively consistent brightness distribution. However, due to the randomness of noise data, the brightness data of the point cloud data will usually be abnormal. Therefore, if noise data appears in some areas of the 3D point cloud of the road surface of the municipal engineering project under test, the data points corresponding to that area will usually have inconsistent brightness distribution characteristics compared with the normal municipal engineering road surface. Moreover, these abnormal brightness data points will usually also have relatively discrete data distribution characteristics.

[0054] Therefore, based on the above analysis, this embodiment first performs plane fitting on all data points in each cluster, and analyzes the distance of each data point to its fitting plane to screen feature points from all data points in each cluster, that is, to screen out data points that may be affected by noise interference. Specifically:

[0055] As one implementation method, in this embodiment, for the three-dimensional point cloud of the municipal road surface to be measured, all data points in each cluster are used as input to a plane fitting algorithm based on the least squares method, and the fitting plane is output. In practical applications, as other implementation methods, implementers may also use other plane fitting methods such as principal component analysis algorithm according to specific circumstances. This embodiment does not impose any special restrictions on the selection of plane fitting algorithm.

[0056] Among them, the plane fitting algorithm based on the least squares method is a well-known technique, and the specific process of using it to perform plane fitting on data points will not be described in detail.

[0057] Furthermore, the distance from all data points in each cluster to the fitting plane is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. Data points whose distance is greater than the segmentation threshold are recorded as feature points, which are used to characterize data points in the cluster that may have abnormal brightness, i.e. data points that may be affected by noise interference.

[0058] Based on the above analysis, plane fitting is performed on all data points in each cluster of the 3D point cloud to obtain the fitting plane and fitting error, which is used to evaluate whether the data points in the cluster have relatively incomprehensible brightness distribution characteristics relative to the asphalt pavement in which they are located.

[0059] The distance from all data points in each cluster to the fitted plane is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. All data points whose distance is greater than the segmentation threshold are recorded as feature points, which are used to characterize data points in the cluster that may have abnormal brightness. The mean distance between all data points in each cluster is calculated as the brightness feature value of each cluster, which is used to evaluate whether the data points with abnormal brightness in the cluster have relatively discrete data distribution characteristics.

[0060] It should be noted that there are many commonly used threshold segmentation algorithms. In this embodiment, the maximum inter-class variance algorithm is used to classify the data points in each cluster. In practical applications, as other implementation methods, implementers may also choose other threshold segmentation algorithms according to specific circumstances. This embodiment does not impose any special restrictions.

[0061] Among them, the Otsu's inter-class variance algorithm is a well-known technique, and its specific principle will not be elaborated here.

[0062] It should be noted that, in this embodiment, the maximum inter-class variance algorithm is used for all threshold segmentation algorithms.

[0063] Furthermore, to further determine the degree of noise interference affecting the point clouds corresponding to each cluster, this embodiment determines the noise feature value of each cluster based on the distance between all feature points in each cluster and the fitting error during the plane fitting process, combined with the brightness feature value. Specifically:

[0064] As one implementation method, in this embodiment, the mean distance between all feature points in each cluster is calculated, the normalized value of the mean distance is calculated and the product of the normalized value of the fitting error is calculated, and the ratio of the product to the normalized value of the corresponding brightness feature value is used as the noise feature value of each cluster.

[0065] It should be noted that in this embodiment, all normalization processes are performed using the maximum-minimum normalization method. In practical applications, as other implementation methods, implementers may also use other normalization methods such as z-score normalization, depending on the specific circumstances. This embodiment does not impose any special restrictions.

[0066] The methods for obtaining the fitting error and the maximum / minimum value normalization method are well-known techniques, and their specific methods and normalization principles will not be elaborated here.

[0067] Based on the noise characteristic values ​​of each cluster, it can be understood that if the average distance between all feature points in the current cluster is larger, it reflects that the feature points identified as brightness anomalies are more dispersed within the current cluster, meaning that the influence range of noise points or anomalies is wider, the point cloud corresponding to the cluster is more affected by noise, and the corresponding noise characteristic value is larger. At the same time, the larger the fitting error of the current cluster, the greater the brightness inconsistency of the point cloud data within the current cluster relative to the fitting plane. This indicates that there are more noise points in the point cloud corresponding to the current cluster, and therefore, the corresponding noise characteristic value is larger. In addition, if the brightness characteristic value of the current cluster is smaller, it indicates that the average brightness within the current cluster is lower. Since the above content indicates that the possibility of data loss and noise interference in darker areas is greater, the smaller the brightness distribution value, the worse the data quality of the point cloud corresponding to the cluster may be, and the higher the risk of noise interference or data loss.

[0068] Conversely, if the average distance between all feature points in the current cluster is smaller, it reflects that the feature points identified as brightness anomalies are more concentrated in the current cluster, meaning that the influence range of noise points or anomalies is smaller, the point cloud corresponding to the cluster is less affected by noise, and the corresponding noise feature value is smaller. At the same time, the smaller the fitting error of the current cluster, the better the brightness consistency of the point cloud data in the current cluster relative to the fitting plane, indicating that there are fewer noise points in the point cloud corresponding to the current cluster, and therefore, the corresponding noise feature value is smaller. In addition, if the brightness feature value of the current cluster is larger, it indicates that the average brightness in the current cluster is higher. Since the above content indicates that the possibility of data loss and noise interference in brighter areas is smaller, the larger the brightness distribution value, the better the data quality of the point cloud corresponding to the cluster is likely, and the lower the risk of noise interference or data loss.

[0069] Thus, this embodiment constructs noise feature values ​​by analyzing the brightness characteristics and spatial distribution of three-dimensional point cloud data, which can more effectively identify and suppress noise, preserve the real texture details of the road surface of the municipal engineering project under test, and improve the accuracy of subsequent road surface structure depth detection.

[0070] S3: By analyzing the similarity between the feature descriptors of all data points in each cluster and combining them with the noise feature values, denoising is performed on each cluster to obtain a denoised cluster.

[0071] However, the richer the texture of the road surface area corresponding to the data points in the cluster, the greater the change in local geometry. Therefore, in order to preserve the detailed information of the road surface texture as much as possible, the smoothness of the data in the point cloud corresponding to the cluster should be smaller. Based on the above analysis, this embodiment analyzes the similarity between the feature descriptors of all data points in each cluster and combines the noise feature values ​​to denoise each cluster to obtain a denoised cluster, specifically:

[0072] As one implementation method, in this embodiment, the filter window length W of cluster a is... a The expression is: W a =f[norm(S) a ×Q a [)×W1]+W2; where S a Q represents the normalized value of the noise eigenvalues ​​of cluster a; a W1 represents the normalized mean of the similarity between the feature descriptors of all data points in cluster a; W2 represents the preset base window length; W2 represents the preset adjustment window length; norm() represents the normalization function; f[] is used to output the largest even number not greater than the input value.

[0073] Preferably, the schematic diagram of the filter window length acquisition process provided in this embodiment is as follows: Figure 2 As shown.

[0074] It should be noted that the preset base window length and the preset adjustable window length are both manually set. In this embodiment, the preset base window length is 6 and the preset adjustable window length is 3. In actual applications, as other implementation methods, implementers can also set them according to specific circumstances. This embodiment does not impose any special restrictions.

[0075] Furthermore, it should be noted that the feature descriptor in this embodiment is the VFH descriptor. In practical applications, as other implementation methods, implementers may also use other feature descriptors such as the SVM descriptor depending on the specific circumstances. This embodiment does not impose any special restrictions on the selection of feature descriptors.

[0076] As one implementation method, in this embodiment, the cosine similarity between the feature descriptors of all data points in cluster a is taken as the similarity between the feature descriptors of all data points in cluster a. In practical applications, as other implementation methods, implementers may also use other methods such as the reciprocal of Euclidean distance to measure the similarity between feature descriptors, depending on the specific circumstances. This embodiment does not impose any special restrictions on the selection of methods for measuring the similarity between feature descriptors.

[0077] The VHF descriptor and the calculation method for cosine similarity are well-known techniques. The specific concepts of VHF descriptors and the specific calculation process of cosine similarity will not be elaborated here.

[0078] Based on the filter window length of each cluster, it can be understood that the larger the noise feature value of the current cluster, the more severe the noise interference in the point cloud data of that cluster. In order to effectively suppress this noise, stronger smoothing processing is required. Therefore, a larger filter window is needed. At the same time, if the normalized result of the mean similarity between the feature descriptors of all data points in the current cluster is larger, it indicates that the texture of the municipal engineering road surface corresponding to the data points in the cluster is simpler. Therefore, the filter window can be appropriately increased. Conversely, if the noise feature value of the current cluster is smaller, it indicates that the noise interference in the point cloud data of that cluster is less severe. In order to retain more detailed features, weaker smoothing processing is required. Therefore, a smaller filter window is needed. At the same time, if the normalized result of the mean similarity between the feature descriptors of all data points in the current cluster is smaller, it indicates that the texture of the municipal engineering road surface corresponding to the data points in the cluster is more complex. Therefore, the filter window can be appropriately decreased.

[0079] Furthermore, all data points in each cluster are used as input to the filtering algorithm, wherein the filtering window length of each data point is set to the filtering window length of its respective cluster, and the filtered clusters are output.

[0080] It should be understood that there are many commonly used filtering algorithms. In this embodiment, the mean filtering algorithm is used to denoise the point cloud data. In practical applications, as other implementation methods, implementers may also use other filtering methods according to specific circumstances. This embodiment does not impose any special restrictions.

[0081] The mean filtering algorithm is a well-known technique, and the process of using the mean filtering algorithm to denoise the data will not be described in detail here.

[0082] Thus, this embodiment analyzes the similarity of feature descriptors and noise feature values ​​of data points within clusters, and dynamically adjusts the filter window length. For areas with severe noise and simple textures, a larger filter window is used to effectively suppress noise; for areas with mild noise and complex textures, a smaller filter window is used to preserve details. This method can better balance noise suppression and detail preservation, and improve the accuracy of road surface detection in municipal engineering.

[0083] S4: Analyze the average distribution and dispersion of the point cloud density of all data points in each denoised cluster, determine the density characteristic value of each denoised cluster, and combine it with the brightness characteristic value to screen out the missing data clusters from all denoised clusters of the 3D point cloud.

[0084] Because the height change of the road surface in the municipal engineering project under test is relatively gentle, the continuity of the road surface structure makes the collected three-dimensional point cloud data exhibit a certain continuity in spatial distribution. This continuity results in a relatively uniform point cloud density within the normal point cloud data. Therefore, if a certain road surface area in the municipal engineering project under test has missing data in some point cloud areas in the collected three-dimensional point cloud data, then this point cloud area will usually have a sparser and more uneven point cloud density data distribution characteristics compared to the normal point cloud area, thus causing a spatial discontinuity in this part of the point cloud data.

[0085] Therefore, based on the above analysis, this embodiment determines the density characteristic value of each denoising cluster by analyzing the average distribution and dispersion of the point cloud density of all data points in each denoising cluster, and combines it with the brightness characteristic value to screen out the data missing clusters from all denoising clusters of the 3D point cloud, specifically:

[0086] The normalized value of the point cloud density dispersion of all data points in each denoising cluster is divided by the normalized value of the mean point cloud density, and the result is used as the density feature value of each denoising cluster.

[0087] It should be noted that this embodiment uses a radius-based point cloud density calculation method to calculate the point cloud density of each data point. In practical applications, as other implementation methods, implementers may also use other point cloud density calculation methods according to specific circumstances. This embodiment does not impose any special restrictions.

[0088] Among them, the radius-based point cloud density calculation method is a well-known technology, and its specific calculation principle and process will not be elaborated here.

[0089] Furthermore, the ratio of the normalized density feature value to the normalized brightness feature value of each denoised cluster is used as the missing data feature value of each denoised cluster.

[0090] Based on the missing data feature values ​​of each denoised cluster, it can be understood that the missing data feature values ​​are used to assess the probability of data missingness in the point cloud corresponding to the denoised cluster. If the density feature value of the current denoised cluster is larger, it indicates that the data points are sparser and the point cloud density is more uneven within the denoised cluster. This sparsity and unevenness are usually caused by missing data. Under normal circumstances, the 3D point cloud of municipal engineering road surfaces should be relatively continuous in spatial distribution. Therefore, the larger the density feature value, the greater the probability of missing data, and the larger the missing data feature value. At the same time, if the brightness feature value of the current denoised cluster is smaller, it indicates that the average brightness of the current denoised cluster is lower, and darker areas are more likely to have missing data. Therefore, the smaller the brightness feature value, the greater the probability of missing data, and the smaller the missing data feature value.

[0091] Conversely, the smaller the density feature value of the current denoised cluster, the denser the data points are distributed within the cluster, and the more uniform the point cloud density. This density and uniformity usually indicate data integrity because, under normal circumstances, the 3D point cloud of municipal engineering road surfaces should be relatively continuous in spatial distribution. Therefore, the smaller the density feature value, the lower the probability of missing data, and the smaller the missing data feature value. At the same time, the larger the brightness feature value of the current denoised cluster, the higher the average brightness of the current denoised cluster. Brighter areas are less likely to have missing data. Therefore, the larger the brightness feature value, the lower the probability of missing data, and the smaller the missing data feature value.

[0092] Furthermore, the missing data feature values ​​of all denoised clusters in the 3D point cloud of the road surface of the municipal engineering project to be tested are used as the input of the threshold segmentation algorithm. The output segmentation threshold is recorded as the missing threshold. The denoised clusters with missing data feature values ​​greater than the missing threshold are regarded as missing data clusters.

[0093] Thus, this embodiment identifies regions that may have missing data by analyzing the point cloud density distribution and brightness characteristics of each cluster after denoising. This method helps to accurately identify and process missing regions in point cloud data, providing a more reliable data foundation for subsequent road surface structure depth detection.

[0094] S5: Fill in the missing values ​​of all missing clusters in the 3D point cloud to obtain a complete 3D point cloud of the road surface of the municipal engineering project to be tested, so as to detect the structural depth of the road surface of the municipal engineering project to be tested.

[0095] Based on the analysis and missing data clusters obtained in step S4, this embodiment fills in the missing values ​​of all missing data clusters in the 3D point cloud, and denotes the 3D point cloud after missing value filling as the complete 3D point cloud of the road surface of the municipal engineering project to be tested. Further, the area of ​​the projection plane of the complete 3D point cloud onto the XOY plane is calculated; 3D reconstruction is performed on the complete 3D point cloud. In this embodiment, the Delaunay triangulation method is used to perform 3D punching on the complete 3D point cloud to obtain a triangular mesh model. The volume of the convex polyhedron formed by all triangular facets in the triangular mesh model and the projection plane of the complete 3D point cloud onto the XOY plane is calculated, which is used to characterize the construction depth of the area where the road surface of the municipal engineering project to be tested is located. Further, the ratio of the volume to the area is taken as the construction depth of the road surface of the municipal engineering project to be tested.

[0096] Among them, the Delaunay triangulation method and triangular facets are well-known technologies. The specific process of using the Delaunay triangulation method to reconstruct a complete 3D point cloud and the specific concept of triangular facets will not be elaborated here.

[0097] Based on the same inventive concept as the above method, this application also provides a municipal engineering road surface testing device, which stores a computer program. When the computer program is executed by a processor, it implements the municipal engineering road surface testing method described above.

[0098] Based on the same inventive concept as the above method, this application embodiment also provides a municipal engineering road surface detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described municipal engineering road surface detection methods.

[0099] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0100] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0101] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for detecting road surface in municipal engineering projects, characterized in that, The method includes the following steps: A three-dimensional point cloud of the road surface of the municipal engineering project under test is obtained by using an intelligent three-dimensional laser contour sensor. Each data point in the three-dimensional point cloud contains three-dimensional coordinates and brightness values. Cluster all data points in the 3D point cloud, analyze the difference between the mean brightness value of all data points in each cluster and the maximum brightness value in the 3D point cloud, and determine the brightness feature value of each cluster; perform plane fitting on all data points in each cluster, and filter feature points from all data points in each cluster by analyzing the distance of all data points to their fitting plane; based on the distance between all feature points in each cluster and the fitting error in the plane fitting process, and combined with the brightness feature value, determine the noise feature value of each cluster; By analyzing the similarity between the feature descriptors of all data points in each cluster and combining them with the noise feature values, denoising is performed on each cluster to obtain a denoised cluster, including: calculating the filtering window length of cluster a. The expression is: In the formula, The normalized value representing the noise eigenvalues ​​of cluster a; This represents the normalized mean value of the similarity among the feature descriptors of all data points in cluster a; Indicates the preset base window length; This indicates the preset adjustment window length; Represents the normalization function; Used to output the largest even number not greater than the input value; all data points in each cluster are used as input to the filtering algorithm, wherein the filtering window length of each data point is set to the filtering window length of its respective cluster, and the output is a denoised cluster; The average distribution and dispersion of point cloud density of all data points in each denoised cluster are analyzed to determine the density feature value of each denoised cluster. Combined with the brightness feature value, the missing data clusters are screened from all denoised clusters in the 3D point cloud. This includes: using the ratio of the normalized density feature value to the normalized brightness feature value of each denoised cluster as the missing data feature value of each denoised cluster; using the missing data feature values ​​of all denoised clusters in the 3D point cloud of the road surface of the municipal engineering project to be tested as the input of the threshold segmentation algorithm, recording the output segmentation threshold as the missing threshold, and identifying denoised clusters with missing data feature values ​​greater than the missing threshold as missing data clusters. Missing values ​​are filled into all missing clusters in the 3D point cloud to obtain a complete 3D point cloud of the road surface of the municipal engineering project to be tested, so as to detect the structural depth of the road surface of the municipal engineering project to be tested.

2. The method for detecting road surface in municipal engineering as described in claim 1, characterized in that, The brightness characteristic value of each cluster is the ratio of the average brightness value of all data points in each cluster to the maximum brightness value in the 3D point cloud.

3. The method for detecting road surface in municipal engineering as described in claim 1, characterized in that, The step of filtering feature points from all data points in each cluster includes: The distance from all data points in each cluster to the fitted plane is used as the input of the threshold segmentation algorithm, and the output is the segmentation threshold. Data points whose distance is greater than the segmentation threshold are recorded as feature points.

4. The method for detecting road surface in municipal engineering as described in claim 1, characterized in that, The method for determining the noise feature values ​​of each cluster is as follows: Calculate the mean distance between all feature points in each cluster, calculate the product of the normalized value of the mean distance and the normalized value of the fitting error, and use the ratio of the product to the normalized value of the corresponding brightness feature value as the noise feature value of each cluster.

5. The method for detecting road surface in municipal engineering as described in claim 1, characterized in that, The density feature value of each denoised cluster is the result of the normalized value of the point cloud density dispersion of all data points in each denoised cluster being divided by the normalized value of the mean point cloud density.

6. The method for detecting road surface in municipal engineering as described in claim 1, characterized in that, The process of detecting the structural depth of the road surface in the municipal engineering project to be tested includes: Choose any point on the road surface of the municipal engineering project to be measured as the origin, take the direction perpendicular to the plane of the road surface of the municipal engineering project to be measured as the Z axis, and arbitrarily determine the positive directions of the X axis and Y axis by right hand; Calculate the area of ​​the projection plane of the complete 3D point cloud onto the XOY plane; perform 3D reconstruction of the complete 3D point cloud to obtain a triangular mesh model; calculate the volume of the convex polyhedron formed by all triangular facets in the triangular mesh model and the projection plane of the complete 3D point cloud onto the XOY plane; and use the ratio of the volume to the area as the construction depth of the road surface of the municipal engineering project to be measured.

7. A municipal engineering road surface testing device, wherein the device stores a computer program, characterized in that, When the computer program is executed by the processor, it implements a method for detecting road surface in municipal engineering as described in any one of claims 1-6.

8. A municipal engineering road surface detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements a method for detecting road surface in municipal engineering as described in any one of claims 1-6.

Images