Duct piece quality detection method and device for concrete duct piece and medium

Abstract

The embodiment of the invention discloses a duct piece quality detection method and device for a concrete duct piece and a medium, and relates to the technical field of point cloud data processing, and the method comprises the steps: obtaining duct piece scanning point cloud data corresponding to a to-be-detected concrete duct piece and duct piece standard part point cloud data corresponding to the to-be-detected concrete duct piece; curved surface point cloud extraction is carried out on the segment scanning point cloud data and the segment standard part point cloud data, and a scanning curved surface point cloud set and a standard curved surface point cloud set are determined; according to the scanning curved surface point cloud set and the standard curved surface point cloud set, determining a curved surface corresponding relation between each scanning curved surface point cloud and the standard curved surface point cloud, and determining a plurality of corresponding curved surface transformation matrixes based on the curved surface corresponding relation; and through the plurality of curved surface transformation matrixes, registering the segment scanning point cloud data with the segment standard part point cloud data, determining segment registration point cloud data, and performing segment quality detection on the to-be-detected concrete segment based on the segment registration point cloud data.

Description

Technical Field

[0001] The present specification relates to the technical field of point cloud data processing, and in particular to a method, device and medium for detecting the quality of concrete segments. Background Art

[0002] Pre-assembly of industrial steel structures is mainly common in large steel structure bridges, wind turbine blades, power towers and other application scenarios. For the pre-assembly technology of steel structure bridges, it is necessary to carry out horizontal assembly in the steel structure factory, horizontal assembly on site after painting, and finally vertical assembly on site. The horizontal assembly and hoisting work is large, and it also requires the cooperation of total stations and levels. If the span of the bridge is large, a larger site support is required to meet the horizontal assembly process. In view of this, the industry is increasingly eager for digital simulation pre-assembly. In the pre-assembly stage, the factory quality and factory accuracy of concrete segments play a vital role, which not only affects the safety and stability of structures such as wind turbine towers, but also directly affects the long-term operation efficiency and safety of wind turbines. At present, when the quality of concrete segments is inspected, they are aligned with standard segments through point cloud registration technology to compare the size and ensure that the two are compared in the same coordinate system. The traditional point cloud registration method focuses on calculating more statistical features, and then solving the required transformation matrix (including rotation and translation parameters) by matching features. If there are missing points in the point cloud, it will seriously interfere with various features in the statistical sense, resulting in inaccurate registration results.

[0003] The large concrete segments that are unique to wind turbine towers are huge in size, with a height of more than four meters and a maximum arc length of more than ten meters. In addition, the weight of each segment is between three and six tons, so it is impossible to accurately obtain the segment size using manual measurement methods. At the same time, after the segments leave the factory, they are only allowed to be placed, stored, and transported upright, resulting in missing and incomplete data for measurements and scans in a vertical state. Therefore, the collected segment point cloud of the concrete segment is missing, resulting in inaccurate registration results of the traditional point cloud matching method, which in turn affects the quality inspection results of the segment. Summary of the invention

[0004] One or more embodiments of the present specification provide a method, device and medium for detecting the quality of concrete segments, which are used to solve the following technical problems: there are missing points in the collected segment point clouds of the concrete segments, resulting in inaccurate registration results of traditional point cloud matching methods, which in turn affects the quality detection results of the segments.

[0005] One or more embodiments of this specification adopt the following technical solutions:

[0006] One or more embodiments of the present specification provide a method for detecting the quality of a concrete segment, the method comprising: obtaining segment scanning point cloud data corresponding to a concrete segment to be detected and segment standard part point cloud data corresponding to the concrete segment to be detected; performing surface point cloud extraction on the segment scanning point cloud data and the segment standard part point cloud data respectively, and determining a scanning surface point cloud set and a standard surface point cloud set, wherein the scanning surface point cloud set includes a plurality of scanning surface point clouds, and the standard surface point cloud set includes a plurality of standard surface point clouds; determining a surface correspondence relationship between each of the scanning surface point clouds and the standard surface point clouds according to the scanning surface point cloud set and the standard surface point cloud set, so as to determine a plurality of corresponding surface transformation matrices based on the surface correspondence relationship; registering the segment scanning point cloud data with the segment standard part point cloud data through the plurality of surface transformation matrices, and determining segment registration point cloud data, so as to perform segment quality detection on the concrete segment to be detected based on the segment registration point cloud data.

[0007] Furthermore, surface point cloud extraction is performed on the segment scanning point cloud data and the segment standard part point cloud data respectively to determine a scanning surface point cloud set and a standard surface point cloud set, specifically including: analyzing each scanning point cloud in the segment scanning point cloud data and each standard point cloud in the segment standard part point cloud data to determine scanning point cloud feature information of each scanning point cloud and standard point cloud feature information of each standard point cloud, wherein the scanning point cloud feature information includes scanning point distance information and scanning normal vector information, and the standard point cloud feature information includes standard point distance information and standard normal vector information; determining a scanning surface point cloud set and a standard surface point cloud set according to the scanning point cloud feature information of each scanning point cloud and the standard point cloud feature information of each standard point cloud.

[0008] Furthermore, according to the scanning point cloud feature information of each of the scanning point clouds and the standard point cloud feature information of each of the standard point clouds, a scanning surface point cloud set and a standard surface point cloud set are determined, specifically including: based on the scanning point distance information of each of the scanning point clouds and the standard point distance information of each of the standard point clouds, noise points are removed from the pipe segment scanning point cloud data and the pipe segment standard part point cloud data respectively to determine the first pipe segment scanning point cloud data and the first pipe segment standard part point cloud data; according to the scanning normal vector information of each of the scanning point clouds, a plurality of target scanning surface point clouds in the first pipe segment scanning point cloud data are determined, and the number of scanning points in each of the target scanning surface point clouds is obtained; through the standard of each of the standard point clouds, Normal vector information is used to determine multiple target standard surface point clouds in the first segment standard part point cloud data, and obtain the number of standard points in each of the target standard surface point clouds; based on a preset first point count threshold and the number of scanned points in each of the target scanned surface point clouds, screening is performed in the multiple target scanned surface point clouds to determine the multiple scanned surface point clouds, wherein the number of scanned points of each of the scanned surface point clouds is not less than the first point count threshold; according to a second point count threshold and the number of standard points in each of the target standard surface point clouds, screening is performed in the multiple target standard surface point clouds to determine the multiple standard surface point clouds, wherein the number of standard points of each of the standard surface point clouds is not less than the second point count threshold.

[0009] Furthermore, based on the scanned surface point cloud set and the standard surface point cloud set, the surface correspondence relationship between each of the scanned surface point cloud and the standard surface point cloud is determined, specifically including: obtaining multiple scanned surface point clouds in the scanned surface point cloud set and the standard surface point cloud set including multiple standard surface point clouds; performing point cloud feature extraction on each of the scanned surface point clouds and each of the standard surface point clouds to determine the scanned surface features corresponding to each of the scanned surface point clouds and the standard surface features corresponding to each of the standard surface point clouds; based on each of the scanned surface features and the standard surface features, performing feature matching on each of the scanned surface point clouds and the standard surface point clouds to determine the specified standard surface point cloud corresponding to each of the scanned surface point clouds, so as to determine the surface correspondence relationship.

[0010] Furthermore, based on the surface correspondence relationship, a corresponding plurality of surface transformation matrices are determined, specifically including: determining a specified scanned surface point cloud and a corresponding specified standard surface point cloud through the surface correspondence relationship; performing standardization processing on the specified scanned surface point cloud and the specified standard surface point cloud respectively, determining a processed scanned surface point cloud corresponding to the specified scanned surface point cloud and a processed standard surface point cloud corresponding to the specified standard surface point cloud, and determining a standardized transformation matrix and a corresponding standardized transformation inverse matrix; determining a flipped scanned surface point cloud and a flipped transformation matrix according to the processed scanned surface point cloud and the processed standard surface point cloud; performing registration according to the flipped scanned surface point cloud and the processed standard surface point cloud through an iterative closest point algorithm, and determining a registration transformation matrix; determining a corresponding plurality of surface transformation matrices based on the standardized transformation matrix, the standardized transformation inverse matrix, the flipped transformation matrix and the registration transformation matrix.

[0011] Furthermore, based on the processed scanning surface point cloud and the processed standard surface point cloud, a flipped scanning surface point cloud and a flip transformation matrix are determined, specifically including: flipping the processed scanning surface point cloud according to a preset flipping method to determine a plurality of alternative flipped scanning surface point clouds, wherein the flipping method includes rotating around the X-axis by a specified angle, rotating around the Y-axis by a specified angle, and rotating around the Z-axis by a specified angle; matching the plurality of alternative flipped scanning surface point clouds and the processed scanning surface point cloud with the processed standard surface point cloud respectively to determine the flipped scanning surface point cloud; and determining a flip transformation matrix based on the flipped scanning surface point cloud.

[0012] Furthermore, the segment scanning point cloud data is registered with the segment standard part point cloud data through the multiple surface transformation matrices to determine the segment registration point cloud data, which specifically includes: determining the standardized transformation matrix, the standardized transformation inverse matrix, the flipping transformation matrix and the registration transformation matrix in the surface transformation matrix; performing point cloud standardization processing on the segment scanning point cloud data through the standardized transformation matrix to determine the standardized segment scanning point cloud data; flipping the standardized segment scanning point cloud data according to the flipping transformation matrix to determine the flipped segment scanning point cloud data; performing registration transformation on the flipped segment scanning point cloud data based on the registration transformation matrix to determine the registered segment scanning point cloud data; performing transformation processing on the registered segment scanning point cloud data through the standardized transformation inverse matrix to determine the segment registration point cloud data.

[0013] Furthermore, based on the segment registration point cloud data, the segment quality inspection is performed on the concrete segment to be inspected, specifically including: performing multi-dimensional feature extraction on the segment registration point cloud data to determine the corresponding size feature data and geometric feature data; performing segment size inspection on the concrete segment to be inspected through the size feature data to determine the size deviation data corresponding to the concrete segment to be inspected; matching the geometric feature data with the standard part geometric feature data corresponding to the segment standard part point cloud data acquired in advance to determine the shape consistency matching data corresponding to the concrete segment to be inspected; based on the size deviation data and the shape consistency matching data, determining the quality assessment information corresponding to the concrete segment to be inspected, so as to perform segment quality inspection on the concrete segment to be inspected.

[0014] One or more embodiments of this specification provide a segment quality inspection device for a concrete segment, including:

[0015] at least one processor; and,

[0016] a memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above method.

[0018] One or more embodiments of the present specification provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the above method.

[0019] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects: through the above technical solution, by extracting the surface point cloud of the pipe segment scanning point cloud data and the pipe segment standard part point cloud data respectively, it is possible to focus more accurately on the actual surface shape of the pipe segment, rather than the noise or irrelevant parts in the entire point cloud data, thereby reducing the registration error caused by data loss, because the surface point cloud extraction can usually retain key shape information while eliminating unnecessary background or noise; determine the surface correspondence between each scanned surface point cloud and the standard surface point cloud, and calculate multiple surface deformations based on this correspondence. Matrix conversion is a more sophisticated and accurate registration method. Compared with the traditional global registration method, it can take into account the local differences between different surfaces of the segment, thereby achieving more accurate local registration and reducing the registration error caused by incomplete data or scanning angle limitations. Quality inspection based on the registered segment registration point cloud data can ensure the accuracy and reliability of the inspection results. Since the registration process has minimized the errors caused by missing data and inaccurate registration, the subsequently extracted dimensional feature data and geometric feature data will be closer to the actual situation, improving the accuracy and credibility of quality inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0021] Figure 1 A flow chart of a method for detecting the quality of a concrete segment provided in an embodiment of this specification;

[0022] Figure 2 A schematic structural diagram of a segment quality inspection device for a concrete segment provided in an embodiment of this specification. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0024] The embodiment of this specification provides a method for detecting the quality of concrete segments. It should be noted that the execution subject in the embodiment of this specification can be a server or any device with data processing capabilities. Figure 1 A flow chart of a method for detecting the quality of a concrete segment provided in an embodiment of this specification is shown in FIG. Figure 1 As shown, it mainly includes the following steps:

[0025] Step S101, obtaining the segment scanning point cloud data corresponding to the concrete segment to be inspected and the segment standard part point cloud data corresponding to the concrete segment to be inspected.

[0026] In one embodiment of the present specification, the scanning device is determined according to the size, shape, surface material and detection accuracy requirements of the pipe segment. Generally, a high-precision three-dimensional laser scanner or other applicable scanning equipment is used to collect the pipe segment scanning point cloud data corresponding to the concrete pipe segment to be detected. During the scanning process, the scanning environment should be kept clean and tidy to avoid debris and unnecessary interference. In addition, the light is uniform during the scanning process to reduce shadows and reflections. The concrete pipe segment to be detected is placed in a suitable scanning position, and the scanning device is used to perform a full-scale scan of the pipe segment. During the scanning process, attention should be paid to keeping the equipment stable and accurate to avoid scanning errors caused by shaking and vibration. After the scanning is completed, the point cloud data of the pipe segment is exported from the scanning device, usually in the form of a point cloud file (such as .las, .pcd, .ply, etc.). To obtain standard parts information, it is first necessary to determine the standard size, shape, material and other information of the pipe segment to be detected. If there is existing standard parts point cloud data, it can be used directly. If there is no standard parts point cloud data, it can be generated by the following method: use CAD software to draw a three-dimensional model of the standard part, import the three-dimensional model into the point cloud generation software, and generate the corresponding point cloud data. Alternatively, use a high-precision 3D printer to print out the standard part entity, and then scan it to obtain point cloud data.

[0027] Step S102, performing surface point cloud extraction on the segment scanning point cloud data and the segment standard part point cloud data respectively, and determining a scanning surface point cloud set and a standard surface point cloud set.

[0028] The scanning surface point cloud set includes a plurality of scanning surface point clouds, and the standard surface point cloud set includes a plurality of standard surface point clouds.

[0029] In the traditional point cloud matching algorithm, a variety of statistical features are calculated, and the specific transformation matrix is ​​determined by matching the features. In the case of missing scanned point clouds (for example, there is a ground contact surface during scanning, which makes it impossible to scan the ground contact surface), various statistical features will be interfered with, resulting in inaccurate registration results. In one embodiment of the present specification, the point cloud is split to obtain a set of several surfaces. Since the matching between surfaces is more accurate and relatively easy than the overall point cloud, the overall calculated features can be converted into surface correspondence. Surface point clouds are extracted from the segment scanning point cloud data and the segment standard part point cloud data respectively, and a scanning surface point cloud set containing several scanning surface point clouds and a standard surface point cloud set containing several standard surface point clouds are determined.

[0030] Surface point cloud extraction is performed on the segment scanning point cloud data and the segment standard part point cloud data respectively to determine a scanning surface point cloud set and a standard surface point cloud set, specifically comprising: analyzing each scanning point cloud in the segment scanning point cloud data and each standard point cloud in the segment standard part point cloud data to determine scanning point cloud feature information of each scanning point cloud and standard point cloud feature information of each standard point cloud, wherein the scanning point cloud feature information includes scanning point distance information and scanning normal vector information, and the standard point cloud feature information includes standard point distance information and standard normal vector information; determining a scanning surface point cloud set and a standard surface point cloud set according to the scanning point cloud feature information of each scanning point cloud and the standard point cloud feature information of each standard point cloud.

[0031] In one embodiment of the present specification, for each scanned point cloud in the segment scanned point cloud data, the distance between it and the neighboring points is calculated, which can be done by finding the nearest K points through the K-nearest neighbor (K-NN) algorithm, and calculating the average distance or the minimum / maximum distance as the point distance information of the point. A local surface fitting method (such as PCA) is used to estimate the normal vector of each scanned point, find the local neighborhood around the point, and use these points to fit a local plane or surface, the normal vector of which is the normal vector of the point. The method for extracting standard point distance information and standard normal vector information is similar to that of the scanned point cloud, but is applied to the segment standard part point cloud data.

[0032] The point distance information and normal vector information of the scanned point cloud are used as features to perform cluster analysis on the scanned point cloud. Clustering algorithms (such as K-means, DBSCAN, etc.) can divide the point cloud into different surface areas based on these features. During the clustering process, clustering parameters (such as the number of clusters, neighborhood size, etc.) can be adjusted as needed to ensure the accuracy and rationality of the surface division. The point cloud set corresponding to each surface area is extracted from the clustering results, that is, the scanned surface point cloud set. The method for determining the standard surface point cloud set is similar to that for the scanned surface point cloud set, but is directly applied to the point cloud data of the standard parts of the segment. Since the point cloud data of the standard parts of the segment has a clear geometric shape and surface definition, the determination of its surface point cloud set may be more direct and accurate.

[0033] According to the scanning point cloud feature information of each of the scanning point clouds and the standard point cloud feature information of each of the standard point clouds, a scanning surface point cloud set and a standard surface point cloud set are determined, specifically including: based on the scanning point distance information of each of the scanning point clouds and the standard point distance information of each of the standard point clouds, noise points are removed from the segment scanning point cloud data and the segment standard part point cloud data respectively to determine the first segment scanning point cloud data and the first segment standard part point cloud data; according to the scanning normal vector information of each of the scanning point clouds, a plurality of target scanning surface point clouds in the first segment scanning point cloud data are determined, and the number of scanning points in each of the target scanning surface point clouds is obtained; through the standard normal vector information of each of the standard point clouds, a plurality of target scanning surface point clouds in the first segment scanning point cloud data are determined, and the number of scanning points in each of the target scanning surface point clouds is obtained; through the standard normal vector information of each of the standard point clouds, a plurality of target scanning surface point clouds in the first segment scanning point cloud data are determined, and a plurality of target scanning surface point clouds in the first segment scanning point cloud data are ... determined; Vector information is used to determine multiple target standard surface point clouds in the first segment standard part point cloud data, and obtain the number of standard points in each of the target standard surface point clouds; based on a preset first point count threshold and the number of scanned points in each of the target scanned surface point clouds, the multiple target scanned surface point clouds are screened to determine the multiple scanned surface point clouds, wherein the number of scanned points of each of the scanned surface point clouds is not less than the first point count threshold; based on a second point count threshold and the number of standard points in each of the target standard surface point clouds, the multiple target standard surface point clouds are screened to determine the multiple standard surface point clouds, wherein the number of standard points of each of the standard surface point clouds is not less than the second point count threshold.

[0034] In one embodiment of the present specification, each point in the segment scanning point cloud data is traversed, and the scanning point distance information (such as the average distance or standard deviation to the neighboring points) is compared with the preset noise threshold. If the point distance information of a certain point exceeds the noise threshold, the point is considered to be a noise point and is removed from the point cloud data. Repeat the above process until all noise points are removed to obtain the first segment scanning point cloud data. Use the same method to remove noise points from the segment standard part point cloud data to obtain the first segment standard part point cloud data. It should be noted that the noise threshold can be set based on the experience of historical processing of segment point cloud data, and should be reasonably set according to the characteristics of the actual point cloud data and the detection requirements to avoid mistakenly removing valid points or retaining too many noise points.

[0035] Traverse each point in the first segment scanning point cloud data, and divide the point cloud into different potential surface areas according to its scanning normal vector information (clustering algorithm or normal vector similarity-based method can be used). For each potential surface area, calculate the number of scanning points therein. Determine multiple target scanning surface point clouds, which represent the main surfaces in the scanning data. Use a similar method to divide the first segment standard part point cloud data into different potential surface areas according to the standard normal vector information. Calculate the number of standard points in each potential surface area. According to the scanning accuracy of the scanning device, preset a first point count threshold, which represents the minimum number of points required for a valid surface point cloud. Traverse all target scanning surface point clouds, and check whether the number of scanning points in each point cloud is not less than the point count threshold. Filter out the scanning surface point clouds that meet the conditions as the final multiple scanning surface point clouds. According to the scanning parameters or construction parameters in the standard part construction process, set a second point count threshold, and use the second point count threshold to filter the target standard surface point cloud. Filter out the standard surface point clouds that meet the conditions as the final multiple standard surface point clouds.

[0036] Through the above technical solution, the removal of noise points can significantly improve the quality of point cloud data. Noise points are usually caused by random errors in the scanning process, environmental factors or irregularities on the surface of the object. They may interfere with subsequent data analysis and processing. By setting a reasonable noise threshold and removing these points, errors can be reduced and the data can be made more accurate and reliable. Through surface division based on normal vector information and screening based on point count thresholds, point cloud data representing the main surfaces of the pipe segment can be extracted. These point cloud data have sufficient points and clear surface features, and can more accurately represent the shape and structure of the pipe segment. The removal of noise points and the filtering of invalid surface point clouds can reduce the amount of data that needs to be processed, thereby improving the efficiency of data processing. In subsequent steps, such as surface reconstruction, matching or comparative analysis, using more streamlined and accurate data sets can significantly reduce computing time and resource consumption.

[0037] Step S103, determining the surface correspondence relationship between each scanned surface point cloud and the standard surface point cloud according to the scanned surface point cloud set and the standard surface point cloud set, so as to determine a plurality of corresponding surface transformation matrices based on the surface correspondence relationship.

[0038] According to the scanning surface point cloud set and the standard surface point cloud set, the surface correspondence relationship between each of the scanning surface point cloud and the standard surface point cloud is determined, specifically including: obtaining multiple scanning surface point clouds in the scanning surface point cloud set and the standard surface point cloud set including multiple standard surface point clouds; performing point cloud feature extraction on each of the scanning surface point cloud and each of the standard surface point cloud to determine the scanning surface features corresponding to each of the scanning surface point clouds and the standard surface features corresponding to each of the standard surface point clouds; according to each of the scanning surface features and the standard surface features, performing feature matching on each of the scanning surface point clouds and the standard surface point clouds to determine the specified standard surface point clouds corresponding to each of the scanning surface point clouds, so as to determine the surface correspondence relationship.

[0039] In one embodiment of the present specification, feature extraction is performed on each scanned surface point cloud and each standard surface point cloud, and the extracted features are used to describe the shape, geometric characteristics and local structure of the point cloud. Common point cloud features include local surface descriptors (such as FPFH, SHOT, PFH, etc.), curvature, normal vector distribution, point cloud density, etc. For each point cloud, the feature values ​​of all its points are calculated to form a feature vector or feature map, and the extracted features are used to match the scanned surface point cloud with the standard surface point cloud.

[0040] The feature vector or feature map of each point cloud is converted into a common feature space for comparison. The similarity between the features of two point clouds is evaluated by similarity metrics such as Euclidean distance and cosine similarity. Matching algorithms (such as nearest neighbor search, FLANN, KD tree, etc.) are used to find the most similar corresponding point cloud in the standard surface point cloud set for each scanned surface point cloud. Based on the results of feature matching, for each scanned surface point cloud, the standard surface point cloud with the highest similarity is selected as the corresponding point cloud, and the specified standard surface point cloud corresponding to each scanned surface point cloud is determined, thereby establishing a surface correspondence relationship. For example, if the point cloud data of the standard part of the pipe segment is split into six surfaces a, b, c, d, e, and f, and the point cloud data of the pipe segment scanning is split into four surfaces a', c', d', and f', the corresponding relationships are a' and a, c' and c, d' and d, and f' and f, where a' and a are point clouds of the same type and size, c' and c are point clouds of the same type and size, d' and d are point clouds of the same type and size, and f' and f are point clouds of the same type and size. The transformation operation of the point cloud will be calculated later based on these corresponding relationships.

[0041] Based on the surface correspondence, multiple corresponding surface transformation matrices are determined, specifically including: determining a specified scanned surface point cloud and a corresponding specified standard surface point cloud through the surface correspondence; performing standardization processing on the specified scanned surface point cloud and the specified standard surface point cloud respectively, determining a processed scanned surface point cloud corresponding to the specified scanned surface point cloud and a processed standard surface point cloud corresponding to the specified standard surface point cloud, and determining a standardized transformation matrix and a corresponding standardized transformation inverse matrix; determining a flipped scanned surface point cloud and a flipped transformation matrix according to the processed scanned surface point cloud and the processed standard surface point cloud; performing registration according to the flipped scanned surface point cloud and the processed standard surface point cloud through an iterative closest point algorithm, and determining a registration transformation matrix; determining multiple corresponding surface transformation matrices based on the standardized transformation matrix, the standardized transformation inverse matrix, the flipped transformation matrix and the registration transformation matrix.

[0042] According to the processed scanning surface point cloud and the processed standard surface point cloud, a flipped scanning surface point cloud and a flip transformation matrix are determined, specifically including: flipping the processed scanning surface point cloud according to a preset flipping method to determine a plurality of candidate flipped scanning surface point clouds, wherein the flipping method includes rotating around an X-axis by a specified angle, rotating around a Y-axis by a specified angle, and rotating around a Z-axis by a specified angle; matching the plurality of candidate flipped scanning surface point clouds and the processed scanning surface point cloud with the processed standard surface point cloud respectively to determine a flipped scanning surface point cloud; and determining a flip transformation matrix according to the flipped scanning surface point cloud.

[0043] In one embodiment of the present specification, the specified scanning surface point cloud and the corresponding specified standard surface point cloud are determined through the surface correspondence relationship. The specified scanning surface point cloud and the corresponding specified standard surface point cloud here refer to two surfaces with a surface correspondence relationship. The specified standard surface point cloud is represented by pcd_std_major, and pcd_scan_major represents the specified scanning surface point cloud. The point cloud pcd_std_major is standardized to obtain the processed standard surface point cloud pcd_std_major_cent, as well as two transformation matrices std_cmtx and std_dcmtx. The above two transformation matrices refer to the standard transformation matrices corresponding to the standard surface point cloud. The point cloud pcd_scan_major is standardized to obtain the processed scanning surface point pcd_scan_major_cent, as well as two transformation matrices, the standardized transformation matrix scan_cmtx and the corresponding standardized transformation inverse matrix scan_dcmtx.

[0044] It should be noted that the standardization of point clouds can be achieved through the following steps: The first step is to calculate the centroid of the point cloud. The centroid (also called the center point or center of gravity) is the average position of all points in the point cloud, providing a reference point for subsequent transformations. The centroid can be obtained by calculating the average coordinates of all points in the point cloud. The second step is to perform PCA principal component analysis on the point cloud. Principal component analysis (PCA) is a statistical method used to analyze patterns in data and extract the most important features (i.e., principal components). In point cloud processing, PCA is used to find the three main directions of point cloud data (i.e., three orthogonal bases). These directions represent the three dimensions with the largest data changes. These three directions are usually the "length", "width" and "height" of point cloud data. The third step is to obtain the standardized transformation matrix and inverse matrix through the centroid and the three orthogonal bases obtained by PCA. A transformation matrix can be constructed using the centroid and the three orthogonal bases obtained by PCA. This transformation matrix can transform the original point cloud from the original coordinate system to a new coordinate system, where the origin of the new coordinate system is located at the centroid and the coordinate axes are aligned with the three orthogonal bases obtained by PCA. This transformation process is called standardization or normalization. At the same time, the inverse matrix of this transformation also needs to be calculated so that the normalized point cloud can be transformed back to the original coordinate system when necessary.

[0045] Flip pcd_scan_major_cent 180° relative to the three coordinate axes, and a total of four alternative solutions can be obtained, including the point cloud itself pcd_scan_major_cent, and three alternative flipped scan surface point clouds corresponding to the point cloud rotated 180° around the X axis, the point cloud rotated 180° around the Y axis, and the point cloud rotated 180° around the Z axis. Match the three alternative flipped scan surface point clouds and the processed scan surface point cloud with the processed standard surface point cloud, respectively, and find the solution closest to the processed standard surface point cloud pcd_std_major_cent, which is determined to be the flipped scan surface point cloud pcd_scan_major_cent_flipped. Generate two transformation matrices through pcd_scan_major_cent and pcd_scan_major_cent_flipped, including the flip transformation matrix flip_cmtx and the flip transformation inverse matrix flip_dcmtx corresponding to the flip transformation matrix. Through the iterative closest point (ICP) algorithm,

[0046] pcd_scan_major_cent_flipped is registered to pcd_std_major_cent to obtain two transformation matrices: the registration transformation matrix reg_cmtx and the inverse registration transformation matrix reg_dcmtx corresponding to the registration transformation matrix.

[0047] Through standardization processing, the differences in scale, position, etc. between the scanned surface point cloud and the standard surface point cloud can be eliminated, providing a more consistent basis for subsequent alignment and registration; flipping processing can ensure that the scanned surface point cloud is consistent with the target standard surface point cloud in a specific direction, reducing the registration error caused by inconsistent directions. The ICP algorithm finds the optimal transformation matrix through iterative optimization, which can further improve the accuracy of registration, so that the scanned surface point cloud can be as close to the standard surface point cloud as possible; standardization and flipping processing increase the robustness of the algorithm, allowing the algorithm to process point cloud data of different scales, positions and directions; by determining the surface correspondence relationship, fine registration processing can be performed on each local surface area.

[0048] Step S104, registering the segment scanning point cloud data with the segment standard part point cloud data through multiple surface transformation matrices, determining the segment registration point cloud data, and performing segment quality inspection on the concrete segment to be inspected based on the segment registration point cloud data.

[0049] The segment scanning point cloud data is registered with the segment standard part point cloud data through the multiple surface transformation matrices to determine the segment registration point cloud data, specifically including: determining the standardized transformation matrix, the standardized transformation inverse matrix, the flipping transformation matrix and the registration transformation matrix in the surface transformation matrix; performing point cloud standardization processing on the segment scanning point cloud data through the standardized transformation matrix to determine the standardized segment scanning point cloud data; flipping the standardized segment scanning point cloud data according to the flipping transformation matrix to determine the flipped segment scanning point cloud data; performing registration transformation on the flipped segment scanning point cloud data based on the registration transformation matrix to determine the registered segment scanning point cloud data; performing transformation processing on the registered segment scanning point cloud data through the standardized transformation inverse matrix to determine the segment registration point cloud data.

[0050] In one embodiment of the present specification, the following operations are performed on the segment scanning point cloud data pcd_scan point cloud: the standardized transformation matrix scan_cmtx is used to standardize the pcd_scan point cloud to obtain the standardized segment scanning point cloud data pcd_scan_cent; the flip transformation matrix flip_cmtx is used to flip the standardized segment scanning point cloud data pcd_scan_cent to obtain the flipped segment scanning point cloud data pcd_scan_cent_flipped; the registration transformation matrix reg_cmtx is used to transform pcd_scan_cent_flipped to obtain the registered segment scanning point cloud data pcd_scan_cent_reg; the standardized transformation inverse matrix std_dcmtx is used to transform it to the standard part point cloud position to obtain the pcd_scan_reg segment registration point cloud data, that is, the point cloud to be registered after being registered with the standard part point cloud. It should be noted that the registration process in the above process is the registration of each surface, and the registration of the complete point cloud can be achieved through the registration of each surface. In addition, the pcd_scan point cloud in the above scheme refers to any surface point cloud belonging to the segment scanning point cloud data, and its corresponding multiple surface transformation matrices are the set of transformation matrices corresponding to this surface point cloud.

[0051] The application of the standardized transformation matrix eliminates the differences in scale, position, etc. between the scanned point cloud and the standard part point cloud, providing a unified basis for subsequent registration; the flip transformation matrix ensures the consistency of the scanned point cloud with the target standard part point cloud in a specific direction, reducing the registration error caused by inconsistent directions; the registration transformation matrix is ​​calculated through optimization methods such as the ICP algorithm, which can further reduce the deviation between the scanned point cloud and the standard part point cloud and achieve high-precision registration; by determining the surface correspondence and calculating multiple surface transformation matrices, fine registration of each local surface area can be achieved.

[0052] Based on the segment registration point cloud data, the segment quality inspection is performed on the concrete segment to be inspected, specifically including: performing multi-dimensional feature extraction on the segment registration point cloud data to determine the corresponding size feature data and geometric feature data; performing segment size inspection on the concrete segment to be inspected through the size feature data to determine the size deviation data corresponding to the concrete segment to be inspected; matching the geometric feature data with the standard part geometric feature data corresponding to the pre-acquired segment standard part point cloud data to determine the shape consistency matching data corresponding to the concrete segment to be inspected; based on the size deviation data and the shape consistency matching data, determining the quality assessment information corresponding to the concrete segment to be inspected, so as to perform segment quality inspection on the concrete segment to be inspected.

[0053] In one embodiment of the present specification, using tools in a point cloud processing library (such as PCL, Open3D, etc.), key dimension information such as length, width, height, diameter, etc. is extracted from the registered point cloud data, which can be obtained by fitting geometric shapes such as planes and cylinders and calculating their parameters. The extracted dimension features are compared with the corresponding dimensions in the standard part point cloud data, and the dimension deviation is calculated. The dimension deviation data can be expressed as the difference or percentage between the actual size and the standard size. The geometric features such as the curvature and normal vector of the point cloud are extracted, and the extracted geometric feature data are matched with the geometric features of the standard part point cloud data. Feature matching algorithms (such as FLANN, KD-Tree, etc.) can be used to find the most similar feature point set. The similarity or distance between the matching feature point sets is calculated as a measure of shape consistency, that is, shape consistency matching data. Based on the dimension deviation data and shape consistency matching data, a quality assessment standard is formulated. A threshold can be set to determine whether the dimension deviation and shape consistency are within an acceptable range. According to the evaluation results, a quality assessment report is generated, including the dimension deviation details and the shape consistency score, to achieve quality inspection of the concrete pipe segment to be inspected. Multi-dimensional feature extraction can more comprehensively reflect the geometric characteristics of the segments, including size and shape details, and reduce detection errors caused by insufficient single feature extraction. By accurately matching size feature data and geometric feature data, it is possible to achieve fine detection of segment size and shape, thereby improving detection accuracy and reliability. The size feature data covers the key size information of the segment, such as length, width, height and diameter, and can comprehensively evaluate whether the size of the segment meets the standards. The geometric feature data focuses on the shape and surface characteristics of the segment, such as curvature, normal vector, etc., and can reflect the overall shape consistency and surface quality of the segment. Combining size and geometric feature data, a comprehensive evaluation of segment quality can be achieved.

[0054] Through the above technical scheme, by extracting the surface point cloud of the pipe segment scanning point cloud data and the pipe segment standard part point cloud data respectively, it is possible to focus more accurately on the actual surface shape of the pipe segment, rather than the noise or irrelevant parts in the entire point cloud data, and reduce the registration error caused by data missing, because the surface point cloud extraction can usually retain key shape information while eliminating unnecessary background or noise; determining the surface correspondence between each scanned surface point cloud and the standard surface point cloud, and calculating multiple surface transformation matrices based on this correspondence is a more sophisticated and accurate registration method. Compared with the traditional global registration method, it can take into account the local differences between different surfaces of the pipe segment, thereby achieving more accurate local registration and reducing the registration error caused by incomplete data or scanning angle limitation; quality inspection based on the registered pipe segment registration point cloud data can ensure the accuracy and reliability of the inspection results. Since the registration process has minimized the errors caused by missing data and inaccurate registration, the size feature data and geometric feature data extracted subsequently will be closer to the actual situation, improving the accuracy and credibility of quality inspection.

[0055] The embodiment of this specification also provides a segment quality detection device for a concrete segment, such as Figure 2 As shown, the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above method.

[0056] The embodiments of the present specification also provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the above method.

[0057] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0058] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0059] The devices and media provided in the embodiments of this specification correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0060] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take 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.) that contain computer-usable program code.

[0061] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0062] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0063] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0064] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0065] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0066] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0067] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0068] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A method for detecting the quality of a concrete segment, characterized in that: The method comprises: Acquire the segment scanning point cloud data corresponding to the concrete segment to be inspected and the segment standard part point cloud data corresponding to the concrete segment to be inspected; Performing surface point cloud extraction on the segment scanning point cloud data and the segment standard part point cloud data respectively, and determining a scanning surface point cloud set and a standard surface point cloud set, wherein the scanning surface point cloud set includes a plurality of scanning surface point clouds, and the standard surface point cloud set includes a plurality of standard surface point clouds; Determine, according to the scanned surface point cloud set and the standard surface point cloud set, a surface correspondence relationship between each of the scanned surface point clouds and the standard surface point cloud, so as to determine a plurality of corresponding surface transformation matrices based on the surface correspondence relationship; The segment scanning point cloud data is registered with the segment standard part point cloud data through the multiple surface transformation matrices to determine the segment registration point cloud data, so as to perform segment quality inspection on the concrete segment to be inspected based on the segment registration point cloud data.

2. A method for detecting the quality of a concrete segment according to claim 1, characterized in that: The scanning point cloud data of the pipe segment and the standard part point cloud data of the pipe segment are respectively subjected to surface point cloud extraction to determine a scanning surface point cloud set and a standard surface point cloud set, specifically including: Analyze each scanning point cloud in the segment scanning point cloud data and each standard point cloud in the segment standard part point cloud data to determine scanning point cloud feature information of each scanning point cloud and standard point cloud feature information of each standard point cloud, wherein the scanning point cloud feature information includes scanning point distance information and scanning normal vector information, and the standard point cloud feature information includes standard point distance information and standard normal vector information; A scanning surface point cloud set and a standard surface point cloud set are determined according to the scanning point cloud feature information of each of the scanning point clouds and the standard point cloud feature information of each of the standard point clouds.

3. A method for detecting the quality of a concrete segment according to claim 2, characterized in that: Determining a scanned surface point cloud set and a standard surface point cloud set according to the scanned point cloud feature information of each of the scanned point clouds and the standard point cloud feature information of each of the standard point clouds specifically includes: Based on the scanning point distance information of each of the scanning point clouds and the standard point distance information of each of the standard point clouds, respectively, the segment scanning point cloud data and the segment standard part point cloud data are removed for noise points, to determine the first segment scanning point cloud data and the first segment standard part point cloud data; Determine a plurality of target scanning surface point clouds in the first segment scanning point cloud data according to the scanning normal vector information of each of the scanning point clouds, and obtain the number of scanning points in each of the target scanning surface point clouds; Determine a plurality of target standard surface point clouds in the first segment standard part point cloud data through standard normal vector information of each of the standard point clouds, and obtain the number of standard points in each of the target standard surface point clouds; Based on a preset first point count threshold and the number of scanned points in each of the target scanned surface point clouds, screening is performed in the multiple target scanned surface point clouds to determine the multiple scanned surface point clouds, wherein the number of scanned points in each of the scanned surface point clouds is not less than the first point count threshold; According to the second point count threshold and the number of standard points in each of the target standard surface point clouds, screening is performed among the multiple target standard surface point clouds to determine the multiple standard surface point clouds, wherein the number of standard points in each of the standard surface point clouds is not less than the second point count threshold.

4. The method for detecting the quality of a concrete segment according to claim 1, characterized in that: Determining the surface correspondence relationship between each of the scanned surface point clouds and the standard surface point cloud according to the scanned surface point cloud set and the standard surface point cloud set specifically includes: Acquire a plurality of scanned surface point clouds in the scanned surface point cloud set and the standard surface point cloud set includes a plurality of standard surface point clouds; Performing point cloud feature extraction on each of the scanned surface point clouds and each of the standard surface point clouds to determine the scanned surface features corresponding to each of the scanned surface point clouds and the standard surface features corresponding to each of the standard surface point clouds; According to each of the scanned surface features and the standard surface features, feature matching is performed on each of the scanned surface point clouds and the standard surface point clouds, and the designated standard surface point clouds corresponding to each of the scanned surface point clouds are determined to determine the surface correspondence relationship.

5. The method for detecting the quality of a concrete segment according to claim 1, characterized in that: Based on the surface correspondence relationship, a plurality of corresponding surface transformation matrices are determined, specifically including: Determine the specified scanned surface point cloud and the corresponding specified standard surface point cloud through the surface correspondence relationship; Standardizing the designated scanned surface point cloud and the designated standard surface point cloud respectively, determining a processed scanned surface point cloud corresponding to the designated scanned surface point cloud and a processed standard surface point cloud corresponding to the designated standard surface point cloud, and determining a standardized transformation matrix and a corresponding standardized transformation inverse matrix; Determine a flip scanned surface point cloud and a flip transformation matrix according to the processed scanned surface point cloud and the processed standard surface point cloud; By iterating the closest point algorithm, registering the flipped scanned surface point cloud and the processed standard surface point cloud to determine a registration transformation matrix; Based on the normalized transformation matrix, the inverse normalized transformation matrix, the flip transformation matrix and the registration transformation matrix, a plurality of corresponding surface transformation matrices are determined.

6. A method for detecting the quality of a concrete segment according to claim 5, characterized in that: Determining a flip scanned surface point cloud and a flip transformation matrix according to the processed scanned surface point cloud and the processed standard surface point cloud, specifically includes: According to a preset flipping method, the processed scanned surface point cloud is flipped to determine a plurality of candidate flipped scanned surface point clouds, wherein the flipping method includes rotating a specified angle around an X-axis, rotating a specified angle around a Y-axis, and rotating a specified angle around a Z-axis; Matching the multiple candidate flipped scanning surface point clouds and the processed scanning surface point cloud with the processed standard surface point cloud respectively to determine the flipped scanning surface point cloud; A flip transformation matrix is ​​determined according to the flip scanned surface point cloud.

7. The method for detecting the quality of a concrete pipe segment according to claim 1, characterized in that: By using the plurality of surface transformation matrices, the segment scanning point cloud data is registered with the segment standard part point cloud data to determine the segment registration point cloud data, specifically including: Determine a normalized transformation matrix, an inverse normalized transformation matrix, a flip transformation matrix, and a registration transformation matrix in the surface transformation matrix; By using the standardized transformation matrix, the segment scanning point cloud data is subjected to point cloud standardization processing to determine the standardized segment scanning point cloud data; According to the flip transformation matrix, the standardized segment scanning point cloud data is flipped to determine the flipped segment scanning point cloud data; Based on the registration transformation matrix, the registration transformation is performed on the flipped segment scanning point cloud data to determine the registration segment scanning point cloud data; The registered segment scanning point cloud data is transformed by using the standardized inverse transformation matrix to determine the segment registration point cloud data.

8. The method for detecting the quality of a concrete segment according to claim 1, characterized in that: Based on the segment registration point cloud data, segment quality inspection is performed on the concrete segment to be inspected, specifically including: Performing multi-dimensional feature extraction on the segment registration point cloud data to determine corresponding size feature data and geometric feature data; Performing segment size detection on the concrete segment to be detected based on the size characteristic data, and determining size deviation data corresponding to the concrete segment to be detected; Matching the geometric feature data with the standard part geometric feature data corresponding to the pre-acquired point cloud data of the segment standard part to determine shape consistency matching data corresponding to the concrete segment to be tested; Based on the size deviation data and the shape consistency matching data, quality assessment information corresponding to the concrete segment to be inspected is determined to perform segment quality inspection on the concrete segment to be inspected.

9. A segment quality inspection device for a concrete segment, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method according to any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to execute the method according to any one of claims 1 to 8.

Images