Wall flatness evaluation method based on three-dimensional laser scanning technology

Abstract

This invention provides a method for evaluating wall flatness based on three-dimensional laser scanning technology. The method first uses a three-dimensional laser scanner to acquire raw three-dimensional point cloud data of the wall surface. Then, the raw data is preprocessed to improve its quality. Based on the processed point cloud data, a reference plane representing the overall trend of the wall surface is extracted using plane fitting technology. Next, the effective area of ​​the wall surface to be evaluated is determined from the point cloud data, and at least one evaluation reference line is planned within this area. Then, the actual surface point data corresponding to the reference line is extracted, and the deviation value of each point from the reference plane is calculated. Finally, these deviation values ​​are statistically analyzed to calculate evaluation indicators such as standard deviation and range, thereby achieving an objective, comprehensive, and quantitative evaluation of wall flatness. This invention has the advantages of high detection efficiency, accurate and intuitive results, and strong repeatability, and is applicable to fields such as building quality acceptance and decoration project evaluation.

Description

Technical Field

[0001] This invention belongs to the field of building construction quality inspection, specifically relating to a method for evaluating wall flatness based on three-dimensional laser scanning technology. Background Technology

[0002] Wall flatness is a key indicator in building decoration, construction, and quality acceptance. Traditionally, the assessment of wall flatness has relied mainly on manual contact measurement methods, such as using straightedges and wedge gauges for localized inspections. While these methods are simple, they have significant limitations: the results are greatly influenced by the operator's experience and subjective judgment, measurement efficiency is low, and data can only be obtained from discrete sampling points, failing to comprehensively and continuously reflect the flatness of the entire wall surface, and easily overlooking quality issues such as localized depressions or protrusions.

[0003] With the development of 3D measurement technology, equipment such as total stations and 3D laser scanners have been introduced into the field of engineering inspection. High-density point cloud data obtained through 3D laser scanning can completely and accurately record the 3D topographic information of the measured wall surface. Currently, some studies have attempted to use point cloud data for flatness analysis, but existing methods still have shortcomings in processing procedures and evaluation accuracy.

[0004] First, efficiently and accurately separating the point set belonging to the wall to be tested from the overall point cloud data obtained by scanning, and eliminating interference from non-wall targets such as doors, windows, furniture, and equipment, remains a key preprocessing challenge. Manual segmentation is inefficient and unsuitable for large-area automated inspection. Second, when establishing an evaluation benchmark, simple overall plane fitting is easily affected by large-scale local deformation or noise points on the wall surface, leading to distortion of the benchmark surface itself, thus making subsequent deviation calculations inaccurate. In addition, the existing process often relies on multiple independent software or tools for manual processing of point cloud data preprocessing, benchmark fitting, deviation calculation, and evaluation report generation. This fragmented process has a low degree of automation and cannot meet the needs of rapid, batch engineering inspection.

[0005] Therefore, there is an urgent need for a method that can automatically, accurately, and comprehensively assess wall flatness to overcome the shortcomings of the existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a wall flatness assessment method based on three-dimensional laser scanning technology. This method can automatically, accurately, and comprehensively assess wall flatness, meeting the growing demand of modern construction projects for efficient, objective, and digital quality inspection.

[0007] The technical solution of this invention is: a wall surface flatness evaluation method based on three-dimensional laser scanning technology, comprising the following steps: S1: Use a 3D laser scanner to scan the wall surface to obtain the original 3D point cloud data of the wall surface; S2: Preprocess the original 3D point cloud data to obtain the target point cloud data; S3: Based on the target point cloud data, extract the reference plane of the wall surface through plane fitting; S4: In the target point cloud data, determine the effective area of ​​the wall to be evaluated, and plan at least one evaluation reference line within the effective area; S5: Extract the point cloud data corresponding to the evaluation reference line, and calculate the deviation value of each sampling point in the point cloud data from the reference plane; S6: Based on the set of deviation values, perform flatness statistical analysis and calculate one or more flatness evaluation indicators to quantitatively evaluate the flatness of the wall surface.

[0008] Further, in step S4, determining the area of ​​the wall to be evaluated includes: The target point cloud data is transformed into a local coordinate system with the principal normal vector of the wall as the Z-axis through three-dimensional coordinate transformation; In the local coordinate system, the point cloud is spatially clustered based on the KD-Tree nearest neighbor retrieval algorithm, and the subset of point cloud belonging to the wall to be tested is separated and extracted as the effective region.

[0009] Further, in step S4, selecting or generating at least one reference line within the region to be evaluated includes: Obtain the point cloud boundary contour of the effective region; Based on the boundary profile, one or more straight line segments are generated as evaluation reference lines through geometric constraints or user settings.

[0010] Further, S5 includes: S51: On the evaluation reference line, a series of sampling points are determined at a preset sampling interval; S52: For each sampling point, in the point cloud subset of the effective area, obtain its corresponding actual surface point through nearest neighbor search, and use the elevation of the actual surface point as the elevation value of the sampling point. S53: Based on the equation of the reference plane, calculate the difference between the actual elevation value of each sampling point and the ideal elevation value of the reference plane at the corresponding position to obtain the deviation value.

[0011] Furthermore, in step S2, the preprocessing includes noise reduction and filtering operations; the filtering operations include, but are not limited to, at least one of statistical filtering and radius filtering.

[0012] Further, S3 includes: S31: Extract a subset of point clouds belonging to the wall to be tested from the target point cloud data; S32: The least squares method is used to fit the plane equation of the point cloud subset to obtain the mathematical expression of the reference plane.

[0013] Furthermore, in step S6, the flatness evaluation index includes at least one of the following: The standard deviation, maximum positive deviation, maximum negative deviation, range, and percentage of sampling points whose absolute deviation exceeds the preset tolerance threshold.

[0014] Furthermore, S6 is followed by S7; S7: Generate a wall flatness assessment report. The report visualizes the trend of deviation changes on the reference line in the form of charts and graphs, and marks the flatness assessment indicators.

[0015] Furthermore, when planning multiple evaluation reference lines, S6 calculates the flatness evaluation index corresponding to each evaluation reference line, and takes the worst value or average value among all indicators as the final flatness evaluation result of the wall surface to be evaluated.

[0016] Furthermore, in S1, the three-dimensional laser scanner is a ground-based fixed scanner, ensuring that the scanner position and parameter settings can completely cover the wall surface to be measured and obtain a point cloud density that meets the preset accuracy requirements during scanning.

[0017] The beneficial effects of this invention are: (1) This invention automatically collects high-density point cloud data of the entire wall surface through three-dimensional laser scanning, and uses algorithms to automatically complete the entire process from data processing, reference surface extraction to deviation calculation and index generation, eliminating human error, greatly improving detection efficiency, and making the evaluation results have an objective standard that is traceable and reproducible. (2) By combining three-dimensional coordinate transformation with spatial clustering algorithms such as KD-Tree, it is possible to intelligently and accurately separate the subset of point cloud belonging to the wall to be tested from the overall point cloud, effectively eliminating interference, ensuring the pertinence of all subsequent analyses, and overcoming the disadvantages of manual segmentation being inefficient and prone to errors. (3) By using the least squares method to perform plane fitting on the accurately extracted wall point cloud subset, the obtained reference plane can more realistically reflect the overall design trend or ideal state of the wall and is not sensitive to local noise or minor defects. This provides a more reliable reference for subsequent deviation calculation and avoids evaluation deviation caused by the distortion of the reference plane. (4) The analysis mode of continuous sampling along the evaluation reference line is introduced, which can not only capture the flatness change along a specific direction, but also comprehensively evaluate the undulation of the entire wall surface in two dimensions through the comprehensive analysis of multiple lines. This effectively avoids the one-sidedness caused by single-point or single-line evaluation, and the results are more representative. (5) This invention not only calculates statistical indicators such as standard deviation and range, but also introduces practical engineering indicators such as the proportion of out-of-range points. It can quantify the flatness quality from different dimensions. The automatically generated graphic report transforms the abstract deviation data into intuitive curves or heat maps, making the quality problems clear at a glance, which greatly facilitates the judgment of engineering quality, rectification guidance and acceptance archiving. (6) This invention organically integrates point cloud acquisition, preprocessing, algorithm analysis, and result output into a systematic method. This changes the previous situation of relying on multiple software for scattered manual operations, lowers the technical threshold, and is conducive to the standardization and application of this technology in the industry, thus promoting the digital transformation of the building inspection industry. Attached Figure Description

[0018] Figure 1 This is a flowchart of the wall flatness evaluation method based on three-dimensional laser scanning technology in this invention. Detailed Implementation

[0019] Various exemplary embodiments of the invention will now be described in detail with reference to the accompanying drawings. The descriptions of the exemplary embodiments are merely illustrative and are in no way intended to limit the invention or its application or use. The invention can be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to make the invention thorough and complete, and to fully express the scope of the invention to those skilled in the art. It should be noted that, unless otherwise specifically stated, the relative arrangement of components and steps, the composition of materials, numerical expressions, and values ​​set forth in these embodiments should be interpreted as merely exemplary and not as limiting.

[0020] The terms "first," "second," and similar words used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different parts. Words such as "including" or "comprising" mean that the element preceding the word encompasses the element listed after it, without excluding the possibility of encompassing other elements. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] like Figure 1 As shown, a method for evaluating wall flatness based on three-dimensional laser scanning technology is disclosed, characterized by the following steps: S1: Use a 3D laser scanner to scan the wall surface to obtain the original 3D point cloud data of the wall surface; S2: Preprocess the original 3D point cloud data to obtain the target point cloud data; S3: Perform plane fitting on the target point cloud data based on the least squares method to extract the reference plane of the wall surface; S4: In the target point cloud data, determine the effective area of ​​the wall to be evaluated, and plan at least one evaluation reference line within the effective area; S5: Extract the point cloud data corresponding to the evaluation reference line, and calculate the deviation value of each sampling point in the point cloud data from the reference plane; S6: Based on the set of deviation values, perform flatness statistical analysis and calculate one or more flatness evaluation indicators to quantitatively evaluate the flatness of the wall surface.

[0022] In S4, determining the area of ​​the wall to be evaluated includes: The target point cloud data is transformed into a local coordinate system with the principal normal vector of the wall as the Z-axis through three-dimensional coordinate transformation; In the local coordinate system, the point cloud is spatially clustered based on the KD-Tree nearest neighbor retrieval algorithm, and the subset of point cloud belonging to the wall to be tested is separated and extracted as the effective region.

[0023] Specifically, first, load the point set P into the 3D point cloud processing software. total Through human-computer interaction, the user roughly selects an area containing the target wall on the point cloud display interface. The system then uses either Random Sample Consensus (RANSAC) or Principal Component Analysis (PCA) to perform initial plane fitting on all points within this selected area, quickly obtaining an initial plane model. The unit normal vector of this initial plane is calculated and used as the initial estimate n of the principal normal vector of the wall. initial This step is a user-assisted coarse positioning, designed to provide a reliable starting point for subsequent precise calculations.

[0024] The initial estimate n obtained initial Z as the new coordinate system loc The axial direction. Theoretically, this direction should be perpendicular to the ideal wall. We assume the direction of gravity is perpendicular to the Z-axis. loc The projection onto the plane of the axis, after normalization, is used as the Y-axis of the local coordinate system. loc Axis. Calculate X using the cross product of vectors. loc The axis, i.e., X loc = Y loc × Z loc Ensure a right-handed coordinate system is formed. Calculate the rotation-translation transformation matrix T from the original global coordinate system to this local coordinate system. Then, transform the point set P...total All points are transformed by matrix T to obtain the point set P in the new local coordinate system. local In this coordinate system, the Z-axis of a point on an ideal wall is... loc The coordinate values ​​should be close to a constant, while X loc and Y loc The coordinates describe the two-dimensional position of a point on the wall. This transforms the point cloud from a global coordinate system to a local coordinate system based on the target wall, simplifying subsequent processing.

[0025] In the local coordinate system P local In the middle, the wall dot cloud should be mainly distributed in one Z-shape. loc Within thin layers where value fluctuations are minimal. This characteristic is utilized for precise extraction: Constructing a spatial index: for point set P local Construct a three-dimensional KD-Tree data structure to support efficient nearest neighbor search.

[0026] Set clustering parameters: Define the key parameters of the region growing algorithm, including: Seed point search radius R seed This is used to find the initial seed point, for example, 0.05 meters.

[0027] Neighborhood growth radius R grow This is used to find the nearest point to the current point during the growth process, for example, 0.03 meters.

[0028] Planar distance tolerance D tol This is used to determine whether a point is close enough to the current clustering plane, for example, 0.01 meters.

[0029] Tolerance A of the angle between the normal vectors tol This is used to determine whether the normal direction of a point is sufficiently consistent with the average normal of the current cluster, for example, 15 degrees.

[0030] Perform region-based growth clustering: a. P local In the middle, searching for Z loc The point with the smallest absolute coordinate value is selected as the first seed point.

[0031] b. Using this seed point as the core, initiate region growth. Use a KD-Tree query to find its R-value. grow All neighboring points within the radius.

[0032] c. For each queried neighbor point, calculate its distance to the temporary plane fitted by the currently clustered points. If this distance is less than D... tol Furthermore, the angle between the normal vector of this point and the current cluster average normal vector is less than A. tol If the value is true, then add that point to the current cluster.

[0033] d. Use newly added, unprocessed points as new growth points, and repeat steps b-c until no new points meet the addition criteria. At this point, a planar point cloud cluster (Cluster) is formed. i Growth complete.

[0034] e. Among the remaining unvisited points, search for Z again. loc Points with coordinates close to 0 are used as seeds, and the above growth process is repeated until all possible points belonging to the plane have been visited. This ultimately results in multiple planar point cloud clusters: Cluster1, Cluster2,... Cluster n .

[0035] Iterate through all clustering results (Cluster) i Calculate the following characteristics for each cluster: the number of points N i Z loc Standard deviation of coordinates The thickness of the cluster is represented by the thickness along the wall normal; the cluster boundary is at X. loc -Y loc Projected area A on the plane i Choose N i maximum, Minimum, and A i The cluster whose area is closest to the actual area of ​​the target wall is identified as the target wall point cloud cluster. wall .

[0036] Cluster wall All points included in the local coordinate system P local The index below maps back to the original global coordinate system point set P. total In the process, the corresponding 3D points are extracted to form the final output wall point cloud subset P. wall .

[0037] By implementing the above-described S4 specific method, the present invention can automatically and accurately separate the wall area to be evaluated from the complex overall scene point cloud, effectively eliminating interference from non-wall targets, and providing a clean and reliable data foundation for subsequent reference plane fitting and flatness calculation.

[0038] In S4, selecting or generating at least one reference line within the region to be evaluated includes: Obtain the point cloud boundary contour of the effective region; Based on the boundary profile, one or more straight line segments are generated as evaluation reference lines through geometric constraints or user settings.

[0039] Specifically, the 3D wall point cloud Pwall Project the coordinates onto a two-dimensional plane perpendicular to its principal normal vector. Typically, an established local coordinate system (where the Z-axis is the wall normal) is used directly, and the coordinates of each point are taken as its two-dimensional projected coordinates, thus obtaining the two-dimensional point set P. 2D The projection process discards Z. loc Coordinate information, only retaining the planar position on the wall surface.

[0040] For rectangular walls with relatively regular shapes, the convex hull algorithm in computational geometry can be used to process the point set P. 2D The algorithm quickly obtains the convex polygon boundary, which is the outer contour of the wall. For walls with concave structures such as doors and windows, the Alpha Shape algorithm is used to more accurately reflect the actual boundary. By setting an appropriate radius parameter Alpha, the algorithm can generate a boundary contour containing concave points, thereby identifying the inner edges of doors and windows, and finally obtaining one or more closed polygonal loops. The largest outer loop represents the outer boundary of the wall, and the inner loops represent the boundary of the opening.

[0041] The obtained boundary polygons are simplified using the Douglas-Peucker algorithm, which reduces the number of vertices while maintaining the basic shape features, resulting in a simplified ordered vertex sequence, which serves as the mathematical expression C of the wall contour.

[0042] After obtaining contour C, a typical reference line can be automatically generated based on preset engineering evaluation specifications or statistical analysis requirements. Common generation modes include: Center Axis Mode: Calculates the minimum bounding rectangle of the outer boundary of contour C. Two line segments are generated, one horizontal and one vertical, passing through the horizontal and vertical centerlines of this rectangle. These two line segments are intersected with the boundary of contour C, and after trimming, the horizontal center reference line L located within the solid wall area is obtained. h and vertical center reference line L v .

[0043] Equal-spacing grid line mode: Also based on the circumscribed rectangle of contour C, a set of parallel lines is generated in the horizontal and vertical directions at fixed intervals. Each set of parallel lines is trimmed with the wall contour C, retaining all line segments that fall within the wall entity, thus generating a reference line grid covering the wall.

[0044] Feature guide line mode: If the outline C contains the inner ring of the door and window openings, the center point or the upper and lower border lines of the inner ring rectangle of these openings can be calculated; a horizontal or vertical line passing through the center point of the opening can be generated, or the upper and lower borders of the opening can be used directly as reference lines for the evaluation of the flatness of the wall around the opening.

[0045] To meet specific evaluation needs, this implementation also provides a human-computer interaction interface, allowing users to flexibly set reference lines: Graphical selection: On a 3D point cloud or 2D projection view, users can directly click with the mouse to draw straight lines. The system captures the start and end points of the user's drawing and automatically trims them into valid line segments within the wall based on the wall contour C.

[0046] Parametric input: Users can input specific parameters of the reference line through a dialog box. For example, by inputting the Y-coordinate value of a horizontal line, the system automatically generates a horizontal reference line that runs through the wall and is located at that height, and performs the same outline trimming.

[0047] Reference line editing: For any reference line L generated automatically or manually i Users can interactively adjust it, such as dragging the endpoints to change their position and length, or modifying their orientation.

[0048] Each finalized evaluation reference line L i In the system, it is defined as a data structure that contains at least the following information: Geometric definition: The starting coordinates (X, Y, Z) in the local coordinate system start ,Y start ) and endpoint coordinates (X end ,Y end The Z coordinate is 0 by default, which means it is in the reference plane.

[0049] Type label: Indicates whether it is a horizontal line, vertical line, or diagonal line.

[0050] Wall area identifier: Associate it with the wall point cloud subset P where it belongs. wall .

[0051] The set of all generated evaluation reference lines {L i This will be used as the output of this step and directly provided to the subsequent S5 to guide dense sampling and deviation calculation along the line.

[0052] In S4 above, by automatically extracting two-dimensional contours from three-dimensional point clouds and combining geometric constraints with user interaction, this method can efficiently and flexibly generate reference lines that conform to standard evaluation specifications and meet specific customized needs. This ensures the relevance and comprehensiveness of the flatness evaluation, laying a structured foundation for subsequent precise quantitative analysis.

[0053] In some embodiments, S5 includes: S51: On the evaluation reference line, a series of sampling points are determined at a preset sampling interval; S52: For each sampling point, in the point cloud subset of the effective area, obtain its corresponding actual surface point through nearest neighbor search, and use the elevation of the actual surface point as the elevation value of the sampling point. S53: Based on the equation of the reference plane, calculate the difference between the actual elevation value of each sampling point and the ideal elevation value of the reference plane at the corresponding position to obtain the deviation value.

[0054] Specifically, the sampling interval Δs is set according to the required accuracy. For example, for high-precision detection, it can be set to 0.01 meters; for routine detection, it can be set to 0.05 meters. An adaptive interval can also be used, automatically reducing the interval in areas with large curvature changes.

[0055] For the reference line L, calculate its two-dimensional direction vector, and based on the total length and sampling interval Δs, calculate a series of sampling points S at equal intervals starting from the starting point. i The coordinates of (i=1, 2, ..., N) on the XOY plane (X i ,Y i ).

[0056] For efficient searching, the wall point cloud subset P is used. wall Construct a three-dimensional KD-Tree spatial index structure. For each two-dimensional sampling point S... i (X i , Y i ), perform the following operations: a. Define the search column: starting from point (X) i , Y i Centered on (0, 0), define a vertical cylindrical search area in 3D space. The radius of the base of the cylinder is the preset search radius R, and the height of the cylinder in the Z direction should cover the entire possible fluctuation range of the wall point cloud.

[0057] b. Retrieve neighboring points: Utilize the range search function of the KD-Tree to quickly retrieve P. wall All points falling within the cylindrical region are considered as sampling points S. i The corresponding point cloud on the actual surface of the wall is denoted as the neighborhood point set N. i .

[0058] c. Calculate the actual surface elevation: To avoid the influence of noise points, instead of directly taking a single nearest point, we calculate the elevation of the neighborhood point set N. i The elevation (Z coordinate) is statistically processed: If N i If the data is not empty, calculate the median or average of the Z-coordinates of all points and use this value as the actual elevation of the wall surface at that sampling point. .

[0059] If Ni If the value is empty, it may be located in a hole or edge of the wall. In this case, the sampling point is marked as an invalid point and will not be included in the subsequent deviation calculation and statistics. Alternatively, the elevation of the adjacent valid sampling point can be used for interpolation to fill the gap.

[0060] For each valid sampling point S i Its flatness deviation d i This is the vertical distance between the actual surface of the wall at that point and the reference plane. The calculation formula is:

[0061] in, This represents the actual elevation of the wall surface at the sampling point. To convert the plane coordinates of the sampling points (X) i ,Y i Substitute the values ​​into the equation of the datum plane to calculate the ideal plane elevation. In the local coordinate system, if the equation of the datum plane is Z=0, then... The deviation is simplified to .

[0062] The deviation value d i A positive value indicates that the wall surface is raised relative to the reference plane at that point, while a negative value indicates that it is recessed. The magnitude of the absolute value directly reflects the degree of unevenness.

[0063] The calculated deviation sequence {d} can be used to... i Perform simple post-processing to improve data quality: Smoothing filter: Apply a moving average filter (e.g., with a window width of 3 sampling points) to suppress high-frequency fluctuations that may be caused by small noise in the point cloud, making the deviation curve smoother and better reflecting the overall trend; the moving average filter can use a window width of 3 sampling points.

[0064] Gross error removal: Calculate the statistical characteristics (such as mean and standard deviation) of the entire sequence, and consider sampling points with deviation values ​​exceeding the mean ± 3 times the standard deviation as possible gross errors, and mark or remove them; statistical characteristics include mean, standard deviation, etc.

[0065] Finally, the ordered sequence of deviation values ​​{d} i The data, along with the corresponding reference line L and sampling point location information, will be stored as direct input data for the flatness statistical analysis in S6. S5, through equidistant sampling, KD-Tree-based columnar neighborhood search, and statistical elevation calculation, enables this method to reconstruct the continuous wall profile along the specified reference line from discrete point clouds with high accuracy and robustness, and accurately calculate its deviation relative to the ideal reference plane. This method effectively overcomes the calculation errors that may be caused by the uneven spatial distribution of point cloud data, ensuring the accuracy and reliability of the flatness assessment results.

[0066] In some embodiments, the preprocessing in step S2 includes noise reduction and filtering operations; the filtering operations include, but are not limited to, at least one of statistical filtering and radius filtering.

[0067] Statistical filtering is used to remove isolated noise points that are far from the main point cloud group, such as those generated by dust, flying insects, or electronic noise during scanning. Specifically, for each point P in the point cloud... i Use a KD-Tree to search for its K nearest neighbors. Calculate the distance from these K neighbor points to point P. i average distance μ i and its standard deviation σ i Set a global distance threshold multiplier n. For each point P... i If its average distance μ i Greater than the global average distance μ global Add n times the global standard deviation σ global If a point is identified as an outlier, it is then classified as such. The process iterates through all points and removes all data points identified as outliers. This method effectively filters out sparse noise that spatially deviates significantly from the main point cloud distribution.

[0068] Radius filtering, as a supplement to or alternative to statistical filtering, is used to remove points whose density is significantly lower in a local region than in the surrounding region. These points may be residual isolated noise or sparse points located at object boundaries. For each point P in the point cloud... i Use a KD-Tree search to find all neighboring points within a specified radius R centered at the given point. Count the number N neighboring points within this radius. i Set a minimum neighbor number threshold N. min If N i <N min Then we consider point P to be... i If the point cloud density in a given area is too low, the point may be noise or an unreliable point and should be removed. Repeat this process for all points. This method cleans up sparse areas, making the remaining point cloud denser and more uniform.

[0069] By employing a multi-level, multi-mode filtering strategy, various types of noise in the original point cloud can be systematically and adaptively removed, significantly improving data quality while preserving the true geometric details of the wall surface to the maximum extent.

[0070] In some embodiments, S3 includes: S31: Extract a subset of point clouds belonging to the wall to be tested from the target point cloud data; S32: The least squares method is used to fit the plane equation of the point cloud subset to obtain the mathematical expression of the reference plane.

[0071] Specifically, calculate the point set P wall The geometric center (centroid) M ( , , Given the covariance matrix C of all points relative to the centroid M, and the eigenvalue decomposition of the covariance matrix C, take the eigenvector corresponding to the smallest eigenvalue, normalize it, and use it as the unit normal vector n = (A, B, C) of the reference plane. Substitute the normal vector n and the centroid M into the plane equation and calculate the constant term: D = -(A· + B· + C· This process yields a complete mathematical expression for the reference plane. The distances from all points to this plane are calculated, and their standard deviations and maximum values ​​are analyzed to assess the quality of the fit. Through mathematical optimization, the reference plane that best represents the overall trend of the wall surface is automatically determined. The results are objective and accurate, providing a reliable basis for subsequent deviation calculations.

[0072] In some embodiments, in step S6, the flatness evaluation index includes at least one of the following: The deviation values ​​include the standard deviation, maximum positive deviation, maximum negative deviation, range, and the percentage of sampling points whose absolute deviation exceeds the preset tolerance threshold. A smaller standard deviation σ indicates a more uniform wall surface; the maximum positive deviation Max... Pos Maximum negative deviation Max Neg The value of the largest local defect is directly indicated; the range R visually displays the total drop at the most uneven point; the percentage of sampling points where the absolute value of the deviation exceeds the preset tolerance threshold, P. exceed It provides a direct assessment from the perspective of project pass rate. Through multi-dimensional statistical indicators, it achieves a comprehensive and objective quantitative evaluation of wall flatness, from overall dispersion and local extreme defects to project compliance rate, providing accurate data basis for quality acceptance and process improvement.

[0073] In some embodiments, S6 is followed by S7; In S7, a wall flatness assessment report is generated. This report visually displays the deviation trend along the reference line in chart form and labels the flatness assessment indicators. The report is visually intuitive and comprehensive, greatly improving the readability and usability of the assessment results. This allows construction, supervision, or acceptance personnel to quickly and accurately grasp the wall flatness quality status, providing a direct and reliable basis for quality control, rectification decisions, and project acceptance.

[0074] In some embodiments, when planning multiple evaluation reference lines, S6 calculates the flatness evaluation index corresponding to each evaluation reference line, and takes the worst value or average value among all indicators as the final flatness evaluation result of the wall surface to be evaluated.

[0075] Specifically, for extreme value indicators, the worst-case principle is adopted, that is, the most unfavorable value of the indicator among all reference lines is taken as the final value of the wall; including the final maximum positive deviation, the final maximum negative deviation, and the final range.

[0076] For statistical indicators, an average or weighted average principle is used to reflect the overall average condition of the wall surface. This includes the average principle and the weighted average principle; the average principle calculates the average of all reference line indicators, while the weighted average principle calculates the average based on the length (len) of each reference line. k Or the number of sampling points M k Assign different weights.

[0077] The final evaluation index set calculated is compared item by item with the preset engineering acceptance standards.

[0078] If all final indicators are better than or equal to the standard values, the wall surface flatness is deemed acceptable.

[0079] If any final indicator exceeds the standard value, the flatness is deemed unqualified, and the unqualified indicator and its value will be clearly listed in the report.

[0080] Based on the overall results of the final indicators, a graded conclusion will be given, such as excellent, good, qualified, or requiring rectification.

[0081] In S1, the three-dimensional laser scanner is a ground-based fixed scanner. During scanning, it is ensured that the scanner position and parameter settings can completely cover the wall surface to be measured and obtain a point cloud density that meets the preset accuracy requirements.

[0082] This implementation method solves the problem of overall evaluation after local sampling inspection of a large wall surface by defining well-defined multi-line data fusion rules and comprehensive evaluation logic.

[0083] As an example, a Trimble TX6 or Faro Focus S 350 3D laser scanner is used. The scanner positions are planned according to the dimensions of the wall to be measured and the site environment. Typically, 2-3 scanning stations are arranged in a fan shape within a range of 3-8 meters directly in front of the wall, ensuring that the scanning areas of adjacent stations overlap by more than 30% to cover the entire wall surface without blind spots. At least 3 dedicated targets are placed in stable positions near the wall as control points for subsequent automatic point cloud data stitching and coordinate system unification.

[0084] Securely mount the scanner on the tripod and level it. Using the instrument's interface or accompanying software, set the scanning parameters, with a resolution of 1 / 4 or 1 / 5 to control the single-station scanning time while ensuring point cloud density. Adjust the horizontal and vertical scanning angles according to the station distance to ensure that a single-station scan covers the target wall and surrounding targets used for stitching. Enable the instrument's built-in real-time noise reduction or dual-scan comparison function to initially suppress noise caused by moving objects during the acquisition phase.

[0085] Begin scanning from the first planned station. After a single station scan is complete, the instrument automatically saves a point cloud data file containing information such as coordinates and reflection intensity. Carefully move the scanner to the next planned station, ensuring the target remains within the field of view, and repeat the scanning process until all stations have been scanned.

[0086] Import the scan data from all stations into the preprocessing software. Using deployed targets, the software automatically performs precise registration and stitching of the multi-station data to form a complete and unified wall point cloud dataset. Check the stitched point cloud: confirm that there is no missing data in the wall area to be measured and that the point cloud density is uniform. Use software tools to measure and ensure that the point spacing in key areas such as the wall center and four corners meets the preset requirements. Simultaneously, verify the stitching accuracy report of the overall point cloud; typically, the root mean square error should be less than 2 mm to meet the accuracy requirements for subsequent flatness analysis.

[0087] The various embodiments of the present invention have now been described in detail. To avoid obscuring the concept of the invention, some details known in the art have not been described. Those skilled in the art will fully understand how to implement the technical solutions disclosed herein based on the above description.

[0088] The embodiments described above only illustrate some implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for evaluating wall flatness based on three-dimensional laser scanning technology, characterized in that, Includes the following steps: S1: Use a 3D laser scanner to scan the wall surface to obtain the original 3D point cloud data of the wall surface; S2: Preprocess the original 3D point cloud data to obtain the target point cloud data; S3: Based on the target point cloud data, extract the reference plane of the wall surface through plane fitting; S4: In the target point cloud data, determine the effective area of ​​the wall to be evaluated, and plan at least one evaluation reference line within the effective area; S5: Extract the point cloud data corresponding to the evaluation reference line, and calculate the deviation value of each sampling point in the point cloud data from the reference plane; S6: Based on the set of deviation values, perform flatness statistical analysis and calculate one or more flatness evaluation indicators to quantitatively evaluate the flatness of the wall surface.

2. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: In step S4, determining the area of ​​the wall surface to be evaluated includes: The target point cloud data is transformed into a local coordinate system with the principal normal vector of the wall as the Z-axis through three-dimensional coordinate transformation; In the local coordinate system, the point cloud is spatially clustered based on the KD-Tree nearest neighbor retrieval algorithm, and the subset of point cloud belonging to the wall to be tested is separated and extracted as the effective region.

3. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 2, characterized in that: In step S4, selecting or generating at least one reference line within the region to be evaluated includes: Obtain the point cloud boundary contour of the effective region; Based on the boundary profile, one or more straight line segments are generated as evaluation reference lines through geometric constraints or user settings.

4. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that, S5 includes: S51: On the evaluation reference line, a series of sampling points are determined at a preset sampling interval; S52: For each sampling point, in the point cloud subset of the effective area, obtain its corresponding actual surface point through nearest neighbor search, and use the elevation of the actual surface point as the elevation value of the sampling point. S53: Based on the equation of the reference plane, calculate the difference between the actual elevation value of each sampling point and the ideal elevation value of the reference plane at the corresponding position to obtain the deviation value.

5. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: In step S2, the preprocessing includes noise reduction and filtering operations; the filtering operations include, but are not limited to, at least one of statistical filtering and radius filtering.

6. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: S3 includes: S31: Extract a subset of point clouds belonging to the wall to be tested from the target point cloud data; S32: The least squares method is used to fit the plane equation of the point cloud subset to obtain the mathematical expression of the reference plane.

7. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: In step S6, the flatness evaluation index includes at least one of the following: The standard deviation, maximum positive deviation, maximum negative deviation, range, and percentage of sampling points whose absolute deviation exceeds the preset tolerance threshold.

8. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: S6 is followed by S7; S7: Generate a wall flatness assessment report. The report visualizes the trend of deviation changes on the reference line in the form of charts and graphs, and marks the flatness assessment indicators.

9. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that: When planning multiple evaluation reference lines, S6 calculates the flatness evaluation index corresponding to each evaluation reference line, and takes the worst value or average value among all indicators as the final flatness evaluation result of the wall surface to be evaluated.

10. The wall surface flatness evaluation method based on three-dimensional laser scanning technology according to claim 1, characterized in that, In step S1, the three-dimensional laser scanner is a ground-based fixed scanner. During scanning, the scanner position and parameter settings are ensured to completely cover the wall surface to be measured and obtain a point cloud density that meets the preset accuracy requirements.

Images