WO2024099007A1 - 一种适用于长钢筋笼的自动化检测方法及装置 - Google Patents

一种适用于长钢筋笼的自动化检测方法及装置 Download PDF

Info

Publication number
WO2024099007A1
WO2024099007A1 PCT/CN2023/123479 CN2023123479W WO2024099007A1 WO 2024099007 A1 WO2024099007 A1 WO 2024099007A1 CN 2023123479 W CN2023123479 W CN 2023123479W WO 2024099007 A1 WO2024099007 A1 WO 2024099007A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
cloud data
long steel
data
point
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2023/123479
Other languages
English (en)
French (fr)
Inventor
舒江鹏
张晓武
张从广
肖文楷
段元锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University
Original Assignee
Zhejiang University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University filed Critical Zhejiang University
Publication of WO2024099007A1 publication Critical patent/WO2024099007A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/08—Measuring arrangements characterised by the use of optical techniques for measuring diameters
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the present invention relates to the technical field of steel cage detection, and in particular to an automatic detection method and device suitable for long steel cages.
  • the steel cage In the current production of reinforced concrete prefabricated components and the process of steel bar binding on the construction site, the steel cage needs to be inspected.
  • the quality inspection of the steel skeleton of the component is completely dependent on manual work.
  • the quality inspector checks the number of steel bars in the steel skeleton by counting, measures the spacing and diameter of the steel bars by steel rulers and calipers, and determines whether the quality of the steel bars meets the design requirements after comparing with the drawings.
  • This method that relies on manual inspection has problems such as low efficiency and easy errors.
  • the method based on three-dimensional laser scanning point cloud can obtain three-dimensional point cloud data of steel cage.
  • Existing research on steel cage point cloud is mostly suitable for situations with good data integrity and small data volume.
  • Patent document CN112902800A discloses a steel bar diameter measuring device, including a box body, a detachable box cover is installed on the upper surface of the box body, a first baffle and a second baffle are symmetrically installed on the upper surface of the box cover for limiting the position of the steel bar, a side of the box body is installed with a plurality of indicator lights from top to bottom, a vertically arranged carrier is fixed on the inner surface of the box body, a side of the carrier is installed with a plurality of travel switches electrically connected to an external power supply from top to bottom according to the diameter of the steel bar, a through hole is opened in the middle of the upper surface of the box cover, a sleeve is provided on the upper side of the through hole, a rubber sleeve is pasted in the sleeve, and a pressure block for squeezing the travel switch is installed at the lower end of the pressure rod.
  • the device can only detect a single unformed steel bar, and cannot be used
  • Patent document CN110779421A discloses a detection device and method for steel bar size, comprising a rectangular placement plate, on which at least three measuring pieces are evenly spaced along the length direction of the placement plate, the measuring piece comprises a measuring base, on which a fixing plate and a measuring plate are relatively arranged, the fixing plate and the measuring plate are arranged in abutment with each other, the measuring plate is slidably connected to the measuring base in a direction perpendicular to the length direction of the placement plate, a driving plate is arranged on one side of the placement plate located on the measurement plate, the driving plate is arranged along the length direction of the placement plate, a spring is arranged between the measurement plate and the driving plate of each measuring piece, the two ends of the spring are respectively fixedly connected to the measurement plate and the driving plate, and a driving device for driving the driving plate to move in a direction perpendicular to the length direction of the placement plate is arranged on the placement plate.
  • the device can only detect a small part of the steel bar structure, and is not
  • the present invention provides an automated detection method, which only needs to collect point cloud data of the long steel cage for analysis and processing, so as to quickly and accurately obtain the dimensional data of each part of the long steel cage.
  • An automated detection method for long steel cages comprising:
  • Step 1 collecting point cloud data of the long steel cage, determining the position of the long steel cage in space based on the point cloud data, and taking the length direction of the longest side of the long steel cage as the main axis direction, adjusting the long steel cage in space so that the main axis direction is parallel to any coordinate axis of the space;
  • Step 2 projecting the point cloud data in the main axis direction according to the adjusted long steel cage to obtain the corresponding one-dimensional point density and coordinate relationship, and cutting the point cloud data according to the one-dimensional point density to obtain the point cloud data of the stirrup slices;
  • Step 3 According to the stirrup slice point cloud data obtained in step 2, the improved RANSAC method is used to perform stirrup fitting to obtain the point cloud data of the corresponding stirrup;
  • Step 4 After removing the point cloud data of the stirrups and the point cloud data of the corrugated pipe part in the long steel cage space, the point cloud data of the longitudinal reinforcement is obtained by using a statistical method;
  • Step 5 Based on the point cloud data obtained in Steps 3 and 4, the length and diameter data of the long steel cage part are calculated by principal component analysis.
  • the present invention divides and slices the point cloud data according to the density in the length direction of the point cloud, detects stirrups for each point cloud slice, and then detects longitudinal reinforcement for the remaining overall point cloud.
  • the principal component analysis method is used to determine the projection direction
  • the RANSAC method with limited parameters is used to improve the detection accuracy
  • the statistical method is used to enhance the robustness to missing data.
  • step 2 the specific process in step 2 is as follows:
  • Step 2-1 delete the coordinate components in the non-main axis direction in the point cloud data of each long steel cage to obtain a one-dimensional data point;
  • Step 2-2 count the maximum and minimum values of the one-dimensional data points, divide a certain number of equal-width intervals between the two, and count the number of coordinate points in each interval;
  • Step 2-3 smoothing the data of the number of coordinate points obtained by statistics, generating a corresponding image curve according to the processed data, and extracting the minimum value corresponding to the fluctuation feature in the image curve as a reference point for cutting;
  • Step 2-4 Obtain the reference point according to step 2-3, cut the long steel cage, and obtain a slice of each point cloud data containing a stirrup section.
  • the improved RANSAC method introduces shape feature constraint rules based on the traditional RANSAC method, thereby improving the cutting efficiency of point cloud data.
  • the shape feature constraint rules include limiting the direction of the straight line connecting two sampling points to be close to the coordinate axis direction, the slenderness ratio of the inner point bounding box is relatively large, and the standard deviation of the point cloud coordinates along the length direction is relatively small.
  • step 3 also includes a review operation: according to the spatial position of the point cloud data in the fitting result, the position distribution area where the corresponding stirrups are located is divided, and on the basis of reducing the set threshold of the number of internal points, a statistical method is used to re-search for the extracted stirrups in the position distribution area to avoid missing the point cloud data of the stirrups.
  • step 4 the point cloud data of the longitudinal reinforcement is obtained by using a statistical method, and the specific process is as follows:
  • the remaining point cloud data it is first projected along the main axis direction to obtain the two-dimensional point density value, and the area with a two-dimensional point density value higher than a certain threshold is used as the candidate area for the regional longitudinal reinforcement.
  • the steel bars in the candidate area are screened to obtain the longitudinal reinforcement and the corresponding point cloud data.
  • step 5 the specific steps are as follows:
  • Step 5-1 remove outliers based on the acquired point cloud data acquisition radius filtering, and determine the steel bar length based on the difference between the maximum and minimum coordinates in the main axis direction;
  • Step 5-2 Project the coordinates of the point cloud data along the main axis direction to obtain a two-dimensional point cloud, and perform circle fitting on the two-dimensional point cloud to obtain the steel bar diameter.
  • the present invention also provides an automated detection device, comprising a memory and one or more processors, wherein the processors store executable codes, and when the one or more processors execute the executable codes, they are used to implement the above-mentioned automated detection method, and the specific steps are as follows: scanning to obtain point cloud data of the long steel cage, analyzing the point cloud data through an automated detection method, and outputting the length and diameter data of each part of the long steel cage.
  • the present invention has the following beneficial effects:
  • the present invention adopts a PCA-based method to obtain the main direction of the point cloud, projects the point cloud into one dimension, and then cuts the point cloud data according to the density, so that the sampling object of RANSAC when detecting stirrups is adjusted from the complete point cloud to the local slice, which greatly speeds up the data processing speed.
  • the present invention determines the possible positions of stirrups by statistical methods, thereby preventing missed detection due to poor point cloud data and having good robustness.
  • the RANSAC algorithm adopted in the present invention adds the direction of the limited sampling line, the slenderness ratio of the internal points, and the uniformity on the basis of the commonly used limited number of internal points, so that the result is more consistent with the actual steel bars and the accuracy of finding steel bars is higher.
  • FIG1 is a flow chart of an automated detection method for long steel cages based on point clouds in the present invention
  • FIG2 is a projection statistical diagram of the original point cloud data of the embodiment of the present invention, showing the effect of curve smoothing
  • FIG3 is a cross section of two stirrups extracted from an embodiment of the present invention.
  • FIG4 is a schematic diagram of long stirrup identification of two stirrup cross sections extracted in an embodiment of the present invention, in which black points are extracted points and gray points are remaining points;
  • FIG5 is a schematic diagram of a process of completely extracting a stirrup cross section in an embodiment of the present invention.
  • FIG6 is a schematic diagram of extracting a portion of a bellows in an embodiment of the present invention, wherein the gray points are bellows points and the black points are remaining external points;
  • Figure 7 is a thermal map obtained by merging the point clouds after the stirrups are extracted and projecting and counting along the long axis of the point clouds;
  • FIG8 is a schematic diagram of measuring the size of the extracted steel bar point cloud.
  • the present invention provides a point cloud-based long steel cage automatic detection method, comprising the following steps:
  • Step 1 Preprocess the point cloud data, use the principal component analysis method to determine the position of the steel cage, and rotate it so that its main direction is along the axial direction.
  • the first component of the right singular value V is the length direction of the steel cage. Take the unit direction vector along the length direction, whose components in the directions of the three coordinate axes of x, y, and z are x 0 , y 0 , and z 0 respectively. Rotate the steel cage to the length direction along a certain coordinate axis to facilitate subsequent projection operations. Assuming that the length direction is along the y-axis after rotation, the rotation matrix is as follows:
  • the main direction of the point cloud can be along the y-axis.
  • the point cloud is projected to the long axis direction of the steel cage, the relationship between the one-dimensional point density and the coordinates is counted, and the point cloud is cut according to the point density to obtain the stirrup slices of the point cloud.
  • the coordinate components in the x and z directions of the three-dimensional coordinates are removed to obtain one-dimensional data points, and a certain number of intervals are divided at equal intervals, and the number of points in each coordinate interval is counted.
  • the data is divided into 5000 intervals. Due to the possible unevenness of the data points along the length direction, the original data is first fitted with the least squares of a larger moving window to obtain a set of smoother data, and then the original data is divided by the smoothed data to make the numerical values evenly distributed.
  • the moving least squares fitting parameters with a window length of 201 data and an order of 1 are used to obtain smooth data, and uniform ratio data is obtained after removing the original data. The ratio data is then fitted to make the curve smooth.
  • two moving window least squares fittings with a window length of 5 and an order of 1 and a window length of 3 and an order of 1 are performed successively, and the results are shown in FIG2 .
  • the maximum point of the statistical curve is the location of the stirrup section, and the point surrounded by two adjacent minimum points is a stirrup section.
  • the minimum points are traversed to find the minimum points.
  • the minimum coordinates on the coordinate components of the projection axis are used as the cutting points to cut the three-dimensional point cloud into several point cloud section slices.
  • the cut point cloud slices are shown in Figure 3, where (a) is a section with better data and (b) is a section with worse data.
  • Step 3 According to the stirrup slice point cloud data obtained in step 2, the stirrups are fitted using the improved RANSAC method;
  • the common RANSAC method for straight line segments is: first randomly select two sampling points in the point cloud slice. If the number of point clouds within a certain radius around the straight line (hereinafter referred to as "inliers") exceeds a given threshold, it is considered that a straight line is found, otherwise re-sample. Based on this, the present invention adds the following restrictions according to the actual situation of steel bars: the direction of the straight line connecting the two sampling points is limited to be close to the direction of the coordinate axis, the slenderness ratio of the bounding box of the inner point is relatively large, and the standard deviation of the point cloud coordinates along the length direction is relatively small. If these three restrictions are not met, re-sample.
  • the parameters given in the embodiment are: the ratio of the maximum directional component along the coordinate axis exceeds 0.9995, the minimum slenderness ratio exceeds 30, and the maximum standard deviation of the coordinates is 1.5. Using this method, most of the long stirrups can be found, as shown in Figure 4, where (a) is a section with better data and (b) is a section with poorer data.
  • the present invention also proposes a solution to the problem of missed stirrup detection due to sparse data points in some areas: when the data points are sparse, the inner points extracted by the above-mentioned RANSAC method are reduced in dimension according to the main axis direction, and then clustered to obtain several cluster centers. The coordinates of the cluster centers are projected into each point cloud slice, and the corresponding straight line is obtained after limiting the direction. By limiting only a small threshold value of the number of inner points, long stirrups that are easily missed can be found.
  • Step 4 Merge the point cloud after extracting stirrups, delete the bellows part according to the coordinate range, and use statistical methods to find and determine the longitudinal reinforcement;
  • the range of the middle area of the stirrups is calculated first.
  • the coordinates of the corner points of the stirrups are used to determine the coordinate range of the internal area, and then the range is reduced by a certain proportion to avoid accidental deletion of part of the point cloud of the longitudinal reinforcement.
  • the coordinates of the corner points of the stirrups are reduced by 97.5% to determine the range of the middle area, and the points within the range are removed as bellows.
  • the effect of the embodiment is shown in Figure 6, where the gray points are bellows points and the black points are the remaining external points.
  • the position of the longitudinal reinforcement is determined by a statistical method.
  • the point cloud without the corrugated tube and stirrups is projected along the long axis of the point cloud to obtain two-dimensional points. After rasterization, the number of points in the grid is counted. Since the data density in different areas varies greatly, the statistical results are more obvious after taking the logarithm. The statistical results obtained in the embodiment are shown in Figure 7 (the darker the color, the denser the points).
  • the center point of the area in the thermal map where the data density exceeds a certain threshold is used as an alternative point. If the number of points within a certain radius of a straight line along the length direction of the steel cage starting from the alternative point exceeds the threshold, it is considered that a longitudinal reinforcement has been found.
  • Step 5 For each detected steel bar point cloud, the length direction is determined by principal component analysis to determine the length, and the diameter is determined by circle fitting.
  • radius filtering is first performed to remove outliers, and then the length direction of the steel bar is determined through principal component analysis.
  • the steel bar point cloud is rotated so that the length direction is along a certain coordinate axis for easy projection.
  • the steel bar length is determined by the difference between the maximum and minimum coordinates along the length axis; the coordinates of the point cloud are projected along the length direction to obtain a two-dimensional point cloud, and the two-dimensional point cloud is fitted with a circle to obtain the steel bar diameter.
  • the measurement method is shown in Figure 8.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Image Analysis (AREA)

Abstract

一种适用于长钢筋笼的自动化检测方法，包括：步骤1、预处理点云数据，使用主成分分析方法确定钢筋笼位置，旋转调整使其主方向沿轴向；步骤2、将点云投影到钢筋笼长轴方向，统计一维点密度与坐标的关系，根据点密度对点云进行切割，得到点云的箍筋切片；步骤3、对每个切片中的点云，使用改进RANSAC方法拟合箍筋；步骤4、剔除箍筋的点云数据和长钢筋笼空间内波纹管部分的点云数据后，获取纵筋的点云数据；步骤5、根据获得的点云数据，计算获得长钢筋笼的长度与直径数据。还提供一种自动化检测装置。

Description

一种适用于长钢筋笼的自动化检测方法及装置 技术领域
本发明涉及钢筋笼的检测技术领域，尤其涉及一种适用于长钢筋笼的自动化检测方法及装置。
背景技术
在当前的钢筋混凝土预制构件生产以及工地现场的钢筋绑扎过程中，都需要对钢筋笼进行检查。目前构件的钢筋骨架质量检查完全依赖人工，质量检查人员通过计数检查钢筋骨架中钢筋的数量，通过钢尺和卡尺测量钢筋的间距以及直径等参数，与图纸对比后确定钢筋的质量是否符合设计要求。这种依赖人工检测的方法存在效率低、容易出错等问题。
基于三维激光扫描点云的方法能够获得钢筋笼的三维点云数据，现有的关于钢筋笼点云的研究多适用于数据完整性好、数据量小的情况，尚未有针对长钢筋笼点云检测的研究。
专利文献CN112902800A公开了一种钢筋测径装置，包括盒体，所述盒体上表面安装有可拆卸的盒盖，所述盒盖上表面对称安装有用于限制钢筋位置的第一挡板和第二挡板，所述盒体一侧面从上至下一次安装有若干指示灯，所述盒体内表面固定有竖向布置的载板，所述载板一侧面依据钢筋直径从上至下安装若干与外接电源电性连接的行程开关，所述盒盖上表面中部位置开设通孔，所述通孔上侧设有筒套，所述筒套内粘贴有橡胶套，所述压杆下端安装有用于挤压行程开关的压块。该装置只能对单条未成型的钢筋进行检测，对于已捆扎成型的钢筋笼无法使用。
专利文献CN110779421A公开了一种用于钢筋尺寸的检测装置及检测方法，包括长方形的放置板，放置板上沿着放置板的长度方向均匀间隔设置有至少三个以上的测量件，测量件包括测量底座，测量底座上相对设置有固定板和测量板，固定板与测量板抵接设置，测量板沿垂直于放置板长度方向的方向滑动连接于测量底座上，放置板上位于测量板的一侧设置有驱动板，驱动板沿着放置板的长度方向设置，每个测量件的测量板与驱动板之间均设置有弹簧，弹簧的两端分别与测量板和驱动板固定连接，放置板上设置有用于带动驱动板沿垂直于放置板长度方向的方向移动的驱动装置。该装置仅能对小部分钢筋结构进行检测，对于长钢筋笼结构效果不佳。
发明内容
为了解决上述问题，本发明提供了一种自动化检测方法，该方法仅需采集长钢筋笼的点云数据进行分析处理，从而快速、准确获得长钢筋笼各部位的尺寸数据。
一种适用于长钢筋笼的自动化检测方法，包括：
步骤1、采集长钢筋笼的点云数据，基于所述点云数据确定长钢筋笼在空间中的位姿，并以长钢筋笼的最长边长度方向作为主轴方向，在空间中对长钢筋笼调整使得所述主轴方向与空间的任一坐标轴平行；
步骤2、根据调整后的长钢筋笼，将点云数据投影在所述主轴方向上，获得对应的一维点密度与坐标关系，并根据所述一维点密度对点云数据进行切割，获得箍筋切片的点云数据；
步骤3、根据步骤2获得的箍筋切片点云数据，采用改进的RANSAC方法进行箍筋拟合，获得对应箍筋的点云数据；
步骤4、剔除箍筋的点云数据和长钢筋笼空间内波纹管部分的点云数据后，采用统计方法获取纵筋的点云数据；
步骤5、根据步骤3和步骤4获得的点云数据，通过主成分分析法计算获得长钢筋笼部位的长度和直径数据。
本发明根据点云长度方向的密度对点云数据进行分割切片，对每个点云切片检测箍筋，后对剩下的整体点云检测纵筋，采用主成分分析的方法确定投影方向，采用限定参数的RANSAC的方法提高检测精度，采用统计的方法增强对缺失数据的鲁棒性。
具体的，所述采集长钢筋笼的点云数据时，需要对采集获得的点云数据进行中心化处理。
具体的，在步骤2中的具体过程如下：
步骤2-1、删去每个长钢筋笼的点云数据中非主轴方向的坐标分量，得到一维数据点；
步骤2-2、统计一维数据点的最大值与最小值，二者之间划分一定数量的等宽区间，统计每一区间中的坐标点数量；
步骤2-3、对统计得到的坐标点数量的数据进行平滑处理，根据处理后的数据生成对应的图像曲线，并提取所述图像曲线中所述波动特征对应的极小值作为切割的参考点；
步骤2-4、根据步骤2-3得到参考点，对长钢筋笼进行切割，获得每个点云数据的切片都包含一个箍筋截面。
优选的，在步骤3中，所述改进的RANSAC方法在传统的RANSAC方法基础上引入了外形特征约束规则，从而提高点云数据的切割效率。
具体的，所述外形特征约束规则包括限定两采样点连成的直线方向为接近坐标轴方向，内点包围盒的长细比较大，点云坐标沿长度方向标准差较小。
优选的，在步骤3中，还包括复查操作：根据拟合结果中点云数据所在的空间位置，划分获得对应箍筋所在的位置分布区域，在减小设定的内点数量阈值基础上，采用统计法在所述位置分布区域内重新查找为被提取的箍筋，避免遗漏箍筋的点云数据。
具体的，在步骤4中，所述采用统计方法获取纵筋的点云数据，具体过程如下：
根据剩余的点云数据，先沿主轴方向投影，统计得到二维点密度值，并将二维点密度值高于一定阈值的区域作为区域纵筋的备选区域，在所述备选区域内的钢筋进行筛选，获得纵筋与对应的点云数据。
具体的，在步骤5中，具体步骤如下：
步骤5-1、根据获得的点云数据采集半径滤波去除离群点，根据主轴方向上坐标最大值与最小值之差确定钢筋长度；
步骤5-2、将点云数据的坐标沿主轴方向投影得到二维点云，对二维点云进行圆拟合即可得到钢筋直径。
本发明还提供了一种自动化检测装置，包括存储器和一个或多个处理器，所述处理器中存储有可执行代码，所述一个或多个处理器执行所述可执行代码时，用于实现上述的自动化检测方法，具体步骤如下：扫描获得长钢筋笼的点云数据，通过自动化检测方法对点云数据进行分析，输出长钢筋笼各部位的长度与直径数据。
与现有技术相比，本发明的有益效果：
(1)本发明采用基于PCA的方法得到点云主方向，将点云投影到一维后根据密度切割点云数据，使检测箍筋时RANSAC的采样对象从完整的点云调整到局部切片，大大加快数据处理速度。
(2)本发明通过统计的方法确定箍筋的可能位置，防止由于点云数据较差而出现的漏检情况，有较好的鲁棒性。
(3)本发明所采用的RANSAC算法在常见的限定内点数的基础上增加了限定采样直线的方向、内点的长细比、均匀程度，使得结果更符合实际钢筋，寻找钢筋的准确度更高。
附图说明
图1为本发明中基于点云的长钢筋笼自动化检测方法流程图；
图2为本发明中对实施例原始点云数据的投影统计图，展示了曲线平滑的效果；
图3为本发明中对实施例提取出的某两个箍筋截面；
图4为本发明中对实施例提取出的某两个箍筋截面的长箍筋识别示意图，图中黑色点为提取出的点，灰色点为剩余点；
图5为本发明中对实施例某一箍筋截面完整提取的过程示意图；
图6为本发明中对实施例部分段的波纹管进行提取的示意图，图中灰色点为波纹管点，黑色点为剩余外点；
图7为提取完箍筋后的点云合并后，沿点云长轴方向投影并统计后得到的热力图；
图8为对提取到的钢筋点云进行测量尺寸的示意图。
具体实施方式
为了使本发明的目的、技术方案及优点更加明白清楚，结合附图和实施例，对本发明进一步的详细说明，应当理解，此处所描述的具体实施例仅仅用以解释本发明，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，均在本发明保护范围。
如图1所示，本发明提供一种基于点云的长钢筋笼自动化检测方法，包括以下步骤：
步骤1：预处理点云数据，使用主成分分析方法确定钢筋笼位置，旋转调整使其主方向沿轴向。
具体实现上，先将每个点云的坐标减去点云坐标均值，使调整后的点云中心点在坐标原点，以调整后的点云数据作为原始数据进行处理。对该调整后的n行3列的三维点云坐标数据进行奇异值分解，得到左奇异值U，奇异值S，右奇异值V，则右奇异值V的第一个分量即钢筋笼长度方向。取沿该长度方向的单位方向向量，其在x、y、z三坐标轴方向的分量分别为x0、y0、z0，将钢筋笼旋转至长度方向沿某一坐标轴，以便于后续投影操作。假定旋转后长度方向沿y轴，则旋转矩阵如下：

其中依次通过这两个旋转矩阵旋转即可使点云主方向沿y轴。
将点云投影到钢筋笼长轴方向，统计一维点密度与坐标的关系，根据点密度对点云进行切割，得到点云的箍筋切片。
具体实现上，对已经调整到主轴方向的点云，以步骤一已获得的沿y轴方向的点云为例，去掉三维坐标中x、z方向的坐标分量，得到一维的数据点，并等间隔地划分一定数量区间，每个坐标区间内统计点的数量。在本实施例中，将数据切分成5000个区间段。由于数据点可能存在的沿长度方向的不均匀，先对原数据采用较大移动窗口的最小二乘拟合得到一组较为平滑的数据，再用原数据除以这个平滑后的数据，使得到数值大小散布均匀。本实施例中， 采用窗口长度为201个数据，阶数为1的移动最小二乘拟合参数得到平滑数据，除原数据后得到均匀的比值数据。之后对该比值数据再进行拟合使曲线平滑，本实施例中，连续进行了窗口为5、阶数为1，窗口为3，阶数为1的两次移动窗口最小二乘拟合，结果如图2所示。
由于箍筋所在截面的点云是相对密集的，因此统计曲线的极大值点附近就是箍筋截面所在位置，相邻两极小值点所包围的点就是一个箍筋截面。对拟合后的数据，遍历找出极小值点，在原三维点云数据中，将投影轴的坐标分量上的各极小值坐标作为切分点，将三维点云切割成若干点云截面切片，切割得到的点云切片如图3所示，其中(a)是数据较好的截面，(b)是数据较差的截面。
步骤3：根据步骤2获得的箍筋切片点云数据，使用改进RANSAC方法拟合箍筋；
具体实现上，常见的针对直线段的RANSAC做法为：先随机取点云切片中的两个采样点，若直线周围一定半径范围内的点云(下称“内点”)数量超过给定阈值，则认为找到一条直线，否则重新采样。本发明在此基础根据钢筋的实际情况增加了如下限定：限定两采样点连成的直线方向为接近坐标轴方向，内点的包围盒的长细比较大，点云坐标沿长度方向标准差较小，若不满足这三个限定要求，则重新采样。实施例中给出的参数为：最大方向分量沿坐标轴比例超过0.9995，最小长细比超过30，坐标最大标准差1.5，使用这样的方法能找出大部分长箍筋，如图4所示，其中(a)是数据较好的截面，(b)是数据较差的截面。
本发明还提出了解决由于部分区域数据点稀疏而造成的箍筋漏检情况：在数据点稀疏的情况下，将采用上述RANSAC方法提取出的内点按主轴方向降维，再进行聚类，可以得到几个聚类中心，将聚类中心的坐标投射到每一个点云切片中，限定方向后得到对应的直线，只限定较小的内点数量阈值，就能找出容易被漏检的长箍筋。
实际情况中短箍筋由于数据扫描角度原因较稀疏，难以直接被找到，因此先找出长箍筋，再根据最外侧两长箍筋的两端点的坐标相连确定短箍筋的位置，将一定范围内的点作为内点，若内点数超过一定阈值，则认为是短箍筋。经过以上操作后即可找出完整箍筋，实施例某一截面的完整检测结果如图5所示，其中图(a)为原始截面，图(b)为提取长箍筋的效果，图(c)为提取所有箍筋的效果。
步骤4：合并提取箍筋后的点云，根据坐标范围删去波纹管部分，使用统计的方法查找并确定纵筋；
具体实现上，考虑到波纹管位于箍筋中间，因此先计算出箍筋中间区域范围。用箍筋各角点的坐标确定出内部区域坐标范围，再按一定比例缩小范围，避免误删纵筋部分点云。在实施例中将箍筋角点坐标以97.5％的比例缩小来确定中间区域范围，范围内的点作为波纹管除去，实施例的效果如图6所示，图中灰色点为波纹管点，黑色点为剩余外点。
采用统计的方法确定纵筋位置。对去除了波纹管与箍筋的点云，将其沿点云长轴方向投影，得到二维点，栅格化后统计网格内点数，由于不同区域数据密度相差较大，取对数后统计结果对比更明显，实施例得到统计结果如图7所示(颜色越深的地方点越密集)，将热力图中数据密度超过一定阈值的区域中心点作为备选点，若以备选点为起点沿钢筋笼长度方向的直线一定半径范围内的点数量超过阈值，则认为找到了一根纵筋。
步骤5：对于检测出的每个钢筋点云，通过主成分分析确定长度方向以确定长度，通过圆拟合确定直径。
具体实现上，对于每根提取出的钢筋点云，先进行半径滤波去除离群点，再通过主成分分析确定钢筋长度方向，将钢筋点云旋转至使长度方向沿某一坐标轴向，以便于投影。通过沿长度轴向的坐标最大值与最小值之差确定钢筋长度；将点云的坐标沿长度方向投影，得到二维点云，对二维点云进行圆拟合即可得到钢筋直径。测量方法如图8所示。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本发明保护的范围之内。

Claims (9)

  1. 一种适用于长钢筋笼的自动化检测方法，其特征在于，包括：
    步骤1、采集长钢筋笼的点云数据，基于所述点云数据确定长钢筋笼在空间中的位姿，并以长钢筋笼的最长边长度方向作为主轴方向，在空间中对长钢筋笼调整使得所述主轴方向与空间的任一坐标轴平行；
    步骤2、根据调整后的长钢筋笼，将点云数据投影在所述主轴方向上，获得对应的一维点密度与坐标关系，并根据所述一维点密度对点云数据进行切割，获得箍筋切片的点云数据；
    步骤3、根据步骤2获得的箍筋切片点云数据，采用改进的RANSAC方法进行箍筋拟合，获得对应箍筋的点云数据；
    步骤4、剔除箍筋的点云数据和长钢筋笼空间内波纹管部分的点云数据后，采用统计方法获取纵筋的点云数据；
    步骤5、根据步骤3和步骤4获得的点云数据，通过主成分分析法计算获得长钢筋笼各部位的长度与直径数据。
  2. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤1中，所述采集长钢筋笼的点云数据时，需要对采集获得的点云数据进行中心化处理。
  3. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤2的具体过程如下：
    步骤2-1、删去每个长钢筋笼的点云数据中非主轴方向的坐标分量，得到一维数据点；
    步骤2-2、统计一维数据点的最大值与最小值，二者之间划分一定数量的等宽区间，统计每一区间中的坐标点数量；
    步骤2-3、对统计得到的坐标点数量的数据进行平滑处理，根据处理后的数据生成对应的图像曲线，并提取所述图像曲线中所述波动特征对应的极小值作为切割的参考点；
    步骤2-4、根据步骤2-3得到参考点，对长钢筋笼进行切割，获得每个点云数据的切片都包含一个箍筋截面。
  4. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤3中，所述改进的RANSAC方法在传统的RANSAC方法基础上引入了外形特征约束规则。
  5. 根据权利要求4所述的适用于长钢筋笼的自动化检测方法，其特征在于，所述外形特征约束规则包括限定两采样点连成的直线方向为接近坐标轴方向，内点包围盒的长细比较大，点云坐标沿长度方向标准差较小。
  6. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤3中， 还包括复查操作：根据拟合结果中点云数据所在的空间位置，划分获得对应箍筋所在的位置分布区域，在减小设定的内点数量阈值基础上，采用统计法在所述位置分布区域内重新查找未被提取的箍筋。
  7. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤4中，所述采用统计方法获取纵筋的点云数据，具体过程如下：
    根据剩余的点云数据，先沿主轴方向投影，统计得到二维点密度值，并将二维点密度值高于一定阈值的区域作为纵筋的备选区域，在所述备选区域内的钢筋进行筛选，获得纵筋与对应的点云数据。
  8. 根据权利要求1所述的适用于长钢筋笼的自动化检测方法，其特征在于，在步骤5中，具体步骤如下：
    步骤5-1、根据获得的点云数据采用半径滤波去除离群点，根据主轴方向上坐标最大值与最小值之差确定钢筋长度；
    步骤5-2、将点云数据的坐标沿主轴方向投影得到二维点云，对二维点云进行圆拟合即可得到钢筋直径。
  9. 一种自动化检测装置，其特征在于，包括存储器和一个或多个处理器，所述存储器中存储有可执行代码，所述一个或多个处理器执行所述可执行代码时，用于实现权利要求1～8任一所述的自动化检测方法，其具体步骤如下：扫描获得长钢筋笼的点云数据，通过自动化检测方法对点云数据进行分析，输出长钢筋笼各部位的长度与直径数据。
PCT/CN2023/123479 2022-11-09 2023-10-09 一种适用于长钢筋笼的自动化检测方法及装置 Ceased WO2024099007A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211397044.3A CN115854883B (zh) 2022-11-09 2022-11-09 一种适用于长钢筋笼的自动化检测方法及装置
CN202211397044.3 2022-11-09

Publications (1)

Publication Number Publication Date
WO2024099007A1 true WO2024099007A1 (zh) 2024-05-16

Family

ID=85662849

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/123479 Ceased WO2024099007A1 (zh) 2022-11-09 2023-10-09 一种适用于长钢筋笼的自动化检测方法及装置

Country Status (2)

Country Link
CN (1) CN115854883B (zh)
WO (1) WO2024099007A1 (zh)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118456446A (zh) * 2024-07-09 2024-08-09 湖南大学 基于三维重建的钢筋笼智能绑扎方法、装置、设备及介质
CN118918361A (zh) * 2024-07-12 2024-11-08 重庆大学 基于rgbd图像的预制构件钢筋间距智能检测方法
CN119217382A (zh) * 2024-11-27 2024-12-31 湖南大学 一种钢筋绑扎机器人的自动作业方法及钢筋绑扎机器人
CN119295544A (zh) * 2024-09-27 2025-01-10 中铁十四局集团装备有限公司 基于点云数据分析的盾构机刀具紧固螺栓的间接定位方法
CN119321233A (zh) * 2024-09-04 2025-01-17 保利长大工程有限公司 一种智能控制箍筋尺寸的检测设备
CN120838966A (zh) * 2025-06-30 2025-10-28 中建三局第三建设工程有限责任公司 一种新型智能化钢筋自动绕笼机
CN120949345A (zh) * 2025-03-21 2025-11-14 同济大学 一种装配式剪力墙边缘构件的钢筋检测方法和设备
CN121383938A (zh) * 2025-12-24 2026-01-23 湖南大学 一种现场多钢筋建筑构件的端到端钢筋间距测量方法及系统
CN121552394A (zh) * 2026-01-22 2026-02-24 湖南大学 一种钢筋绑扎机器人的控制方法、系统、设备及存储介质
RU2856855C1 (ru) * 2025-06-16 2026-02-25 Федеральное государственное бюджетное образовательное учреждение высшего образования "Рязанский государственный радиотехнический университет имени В.Ф. Уткина" Способ контроля серповидности листового материала при наличии некорректных измерений
CN121658919A (zh) * 2026-02-06 2026-03-13 山东大学 一种基于无人机激光点云数据的钢筋网自动化识别方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115854883B (zh) * 2022-11-09 2023-11-03 浙江大学 一种适用于长钢筋笼的自动化检测方法及装置
CN120403431B (zh) * 2025-04-29 2026-02-24 长江武汉通信管理局 船舶通航尺度及交通流检测方法、装置、设备和介质

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392476A (zh) * 2014-12-04 2015-03-04 上海岩土工程勘察设计研究院有限公司 基于最小包围盒算法提取隧道三维轴线的方法
JP2021060199A (ja) * 2019-10-02 2021-04-15 株式会社大林組 鉄筋推定システム、鉄筋推定方法及び鉄筋推定プログラム
CN113487722A (zh) * 2021-06-18 2021-10-08 浙江大学 一种基于三维激光扫描方法的混凝土构件自动化检测方法
CN113776452A (zh) * 2021-07-27 2021-12-10 中国铁道科学研究院集团有限公司标准计量研究所 一种圆柱形筒体的圆度测量方法及装置
CN114234819A (zh) * 2021-12-23 2022-03-25 中交第二航务工程局有限公司 基于三维激光扫描的钢筋部品质量自动化检测方法
JP2022094386A (ja) * 2020-12-15 2022-06-27 三井住友建設株式会社 情報処理装置、検出方法およびプログラム
CN115014198A (zh) * 2022-05-23 2022-09-06 西南石油大学 一种基于三维激光扫描的钢筋安装检测方法
US20220343629A1 (en) * 2019-09-20 2022-10-27 Nec Corporation Processing device, processing method, and computerreadable medium
CN115854883A (zh) * 2022-11-09 2023-03-28 浙江大学 一种适用于长钢筋笼的自动化检测方法及装置

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9286538B1 (en) * 2014-05-01 2016-03-15 Hrl Laboratories, Llc Adaptive 3D to 2D projection for different height slices and extraction of robust morphological features for 3D object recognition
CN110849498B (zh) * 2018-08-21 2025-06-20 深圳市敏杰电子科技有限公司 温度传感器
CN110728689B (zh) * 2019-11-29 2023-04-07 中铁第六勘察设计院集团有限公司 一种基于点云数据的既有铁路中线提取方法
WO2022226588A1 (en) * 2021-04-28 2022-11-03 Aerologix Group Pty Ltd Method and system for generating a model of a structure
CN113400460B (zh) * 2021-06-29 2022-09-23 江苏图知天下科技有限公司 一种钢筋的绑扎方法及绑扎装置
CN114119488B (zh) * 2021-10-29 2024-05-10 成都建工第一建筑工程有限公司 一种面向工厂化的预制叠合板尺寸质量智能检测方法

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104392476A (zh) * 2014-12-04 2015-03-04 上海岩土工程勘察设计研究院有限公司 基于最小包围盒算法提取隧道三维轴线的方法
US20220343629A1 (en) * 2019-09-20 2022-10-27 Nec Corporation Processing device, processing method, and computerreadable medium
JP2021060199A (ja) * 2019-10-02 2021-04-15 株式会社大林組 鉄筋推定システム、鉄筋推定方法及び鉄筋推定プログラム
JP2022094386A (ja) * 2020-12-15 2022-06-27 三井住友建設株式会社 情報処理装置、検出方法およびプログラム
CN113487722A (zh) * 2021-06-18 2021-10-08 浙江大学 一种基于三维激光扫描方法的混凝土构件自动化检测方法
CN113776452A (zh) * 2021-07-27 2021-12-10 中国铁道科学研究院集团有限公司标准计量研究所 一种圆柱形筒体的圆度测量方法及装置
CN114234819A (zh) * 2021-12-23 2022-03-25 中交第二航务工程局有限公司 基于三维激光扫描的钢筋部品质量自动化检测方法
CN115014198A (zh) * 2022-05-23 2022-09-06 西南石油大学 一种基于三维激光扫描的钢筋安装检测方法
CN115854883A (zh) * 2022-11-09 2023-03-28 浙江大学 一种适用于长钢筋笼的自动化检测方法及装置

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HODGE GABRIELLE, GATTAS JOSEPH M: "Geometric and semantic point cloud data for quality control of bridge girder reinforcement cages", AUTOMATION IN CONSTRUCTION, ELSEVIER, AMSTERDAM, NL, vol. 140, 20 May 2022 (2022-05-20), AMSTERDAM, NL , pages 104334 - 104334-16, XP093170243, ISSN: 0926-5805, DOI: 10.1016/j.autcon.2022.104334 *
LIU SHI-LONG, MA ZHI-LIANG: "Automatic Inspection Algorithm of Quantity and Spacing of Reinforcement Based on Point Cloud", JOURNAL OF ARCHITECTURE AND CIVIL ENGINEERING., vol. 39, no. 4, 31 July 2022 (2022-07-31), pages 90 - 99, XP093170261, ISSN: 1673-2049 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118456446A (zh) * 2024-07-09 2024-08-09 湖南大学 基于三维重建的钢筋笼智能绑扎方法、装置、设备及介质
CN118918361A (zh) * 2024-07-12 2024-11-08 重庆大学 基于rgbd图像的预制构件钢筋间距智能检测方法
CN119321233A (zh) * 2024-09-04 2025-01-17 保利长大工程有限公司 一种智能控制箍筋尺寸的检测设备
CN119321233B (zh) * 2024-09-04 2026-05-05 保利长大工程有限公司 一种智能控制箍筋尺寸的检测设备
CN119295544A (zh) * 2024-09-27 2025-01-10 中铁十四局集团装备有限公司 基于点云数据分析的盾构机刀具紧固螺栓的间接定位方法
CN119217382A (zh) * 2024-11-27 2024-12-31 湖南大学 一种钢筋绑扎机器人的自动作业方法及钢筋绑扎机器人
CN120949345A (zh) * 2025-03-21 2025-11-14 同济大学 一种装配式剪力墙边缘构件的钢筋检测方法和设备
RU2856855C1 (ru) * 2025-06-16 2026-02-25 Федеральное государственное бюджетное образовательное учреждение высшего образования "Рязанский государственный радиотехнический университет имени В.Ф. Уткина" Способ контроля серповидности листового материала при наличии некорректных измерений
CN120838966A (zh) * 2025-06-30 2025-10-28 中建三局第三建设工程有限责任公司 一种新型智能化钢筋自动绕笼机
CN121383938A (zh) * 2025-12-24 2026-01-23 湖南大学 一种现场多钢筋建筑构件的端到端钢筋间距测量方法及系统
CN121552394A (zh) * 2026-01-22 2026-02-24 湖南大学 一种钢筋绑扎机器人的控制方法、系统、设备及存储介质
CN121658919A (zh) * 2026-02-06 2026-03-13 山东大学 一种基于无人机激光点云数据的钢筋网自动化识别方法

Also Published As

Publication number Publication date
CN115854883A (zh) 2023-03-28
CN115854883B (zh) 2023-11-03

Similar Documents

Publication Publication Date Title
CN115854883B (zh) 一种适用于长钢筋笼的自动化检测方法及装置
Shu et al. Point cloud and machine learning-based automated recognition and measurement of corrugated pipes and rebars for large precast concrete beams
CN110728689B (zh) 一种基于点云数据的既有铁路中线提取方法
CN114119488B (zh) 一种面向工厂化的预制叠合板尺寸质量智能检测方法
WO2024066463A1 (zh) 一种基于双目视觉的钢筋间距的测量方法和系统
CN109461133B (zh) 桥梁螺栓脱落检测方法及终端设备
CN118258329B (zh) 基于三维扫描技术的墙面垂直度及平整度检测方法
CN114898118A (zh) 基于多源点云的输电线路房屋拆迁量自动统计方法及系统
CN113487555A (zh) 一种基于点云网格化的输电线路隐患点快速检测方法
CN114140466B (zh) 基于图像处理的植物根系测量方法、系统及装置
CN117854061B (zh) 基于三维点云的隧道掌子面迹线裂隙提取方法及系统
CN116432052A (zh) 一种新能源汽车模具用夹具质量检测方法
CN109543236A (zh) 基于变异系数级比分析的岩体结构面粗糙度统计样本数确定方法
CN111060922A (zh) 基于机载激光雷达点云空间分布特征的树木点云提取方法
CN112766685A (zh) 一种基于大数据分析和机器视觉的道路工程质量监理方法及云计算监理平台
CN116740060A (zh) 基于点云几何特征提取的装配式预制构件尺寸检测方法
CN121073865A (zh) 一种基于三维激光点云的桥梁结构线形变化检测方法
CN115841634A (zh) 一种基于输电线路激光点云的电力塔自动提取和类型识别方法及装置
CN115143936A (zh) 一种基于激光点云的输电工程杆塔倾斜度测量方法
CN111813775B (zh) 一种隧道点云数据处理方法、装置及存储介质
CN118482658B (zh) 连铸圆坯低倍试样外形检测方法、装置及设备
CN104573339A (zh) 页岩气储层的地质参数确定方法和装置
CN114240946B (zh) 一种定位器异常检测方法、系统、存储介质及计算设备
CN117408894A (zh) 一种适用于煤矿的点云数据的处理方法及系统
CN114862661A (zh) 一种基于点云数据提取圆柱面的方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23887699

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 23887699

Country of ref document: EP

Kind code of ref document: A1