Part surface topography measurement method in laser directional energy process

Abstract

The invention discloses a method for measuring the surface topography of a part in a laser directional energy process, and the method comprises the following steps: collecting the height information of the surface of the part, and reconstructing the point cloud of the surface of the part; extracting a high-density distribution region of the point cloud height values; fitting a reference plane based on improved RANSAC (Random Sample Consensus); removing an inner point of the reference plane as a background point cloud; projecting the point cloud to an XOY plane to generate a mask pattern; the mask pattern keeps the maximum connected domain, and noise point cloud is removed; calculating a transformation matrix T1 of transforming the reference plane to the XOY plane; the printing reference plane of the point cloud is made to coincide with the XOY plane based on the transformation matrix T1; extracting a first main direction of the point cloud; calculating a transformation matrix T2 from the first main direction to the printing direction, so that the direction of the point cloud is consistent with the deposition direction of the current printing layer, and obtaining the surface appearance of the part; and extracting geometric features of the thin wall through a random sampling statistical method. The accuracy and the reliability of the surface appearance of the part can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of part surface topography measurement, and particularly relates to a method for measuring the surface topography of a part in a laser directed energy process. Background Art

[0002] Laser directed energy deposition technology, also known as laser near-net shaping technology, is a metal additive manufacturing technology. It uses a high-power laser to melt metal powder and deposits it directionally along a pre-set printing path on the surface of a substrate, layer by layer to form a part. Compared with traditional manufacturing processes, LPBF has many advantages such as high design freedom, short product cycle, and high material utilization rate. However, during the forming process of LDED, there are problems such as large forming size deviation and poor surface quality, which seriously reduce the forming ability and forming quality of LDED. Especially the surface quality problem will seriously affect the printing quality of subsequent layers and the fatigue life of the part, and even cause printing failure. Once this happens, it will cause a huge waste of manpower and materials. A line laser scanning camera can quickly measure the height information of the part surface, and combine the laser line scanning speed and resolution to obtain the point cloud of the part surface. Then, through the segmentation, noise reduction, and pose adjustment of the point cloud, the surface topography of the part is finally obtained. Compared with traditional surface quality detection means, such as 3D profilers and optical microscopes, this method can speed up the measurement of the part surface topography during and after the printing process, and has a larger field of view. It can improve the monitoring ability and post-characterization level of the LDED process and promote the development of intelligent manufacturing.

[0003] Chen et al. installed a line laser scanning camera at the end of a robotic arm to collect the surface topography of the part during printing and segmented the substrate point cloud and the target point cloud through the RANSAC algorithm. Based on the RANSAC point cloud segmentation algorithm, the convergence speed is slow and the iteration is time-consuming. Binega et al. fitted the point cloud data of the scanned substrate surface during the scanning process to obtain a straight line and calculated the angle alpha between the straight line and the XOY plane. Binega et al. calculated alpha for each scanned profile to correct the pose of the point cloud. However, applying different transformation angles to each scanned profile will cause deformation and distortion of the surface topography. In addition, this method ignores the influence of substrate deformation. Experiments show that the influence of substrate deformation is much greater than the influence of camera vibration during the scanning process of the line laser scanning camera. In the studies of Chen et al. and Binega et al., the problem that the camera laser line is not perpendicular to the printing direction caused by camera installation error is ignored. Therefore, it is necessary to correct the point cloud direction to coincide with the printing direction. Summary of the Invention

[0004] In order to overcome the defects and deficiencies of the existing technology, the present invention provides a method for measuring the surface topography of parts in a laser directed energy process. The present invention is based on a line laser scanning camera to collect the height information of the part surface, and combines the scanning speed and resolution of the line laser to generate the surface point cloud of the part. Through the segmentation, noise reduction, and pose adjustment of the point cloud, the surface topography of the part is finally obtained. The present invention can improve the accuracy and reliability of the part surface topography.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] The present invention provides a method for measuring the surface topography of parts in a laser directed energy process, including the following steps:

[0007] Based on a line laser scanning camera, collect the height information of the part surface, and reconstruct the surface point cloud of the part in combination with the resolution and scanning speed of the line laser;

[0008] Extract the high-density distribution area of the point cloud height value;

[0009] Based on an improved RANSAC algorithm, fit to obtain the optimal reference plane;

[0010] Remove the inliers of the reference plane as background point clouds. The inliers of the reference plane are caused by the substrate surface;

[0011] Project the point cloud onto the XOY plane to generate a mask image;

[0012] Retain the largest connected component in the mask image and remove the noise point cloud to obtain the noise-reduced point cloud;

[0013] Calculate the transformation matrix T1 for transforming the reference plane to the XOY plane;

[0014] Apply the transformation matrix T1 to the noise-reduced point cloud to make the printing reference plane of the point cloud coincide with the XOY plane and correct the height value of the point cloud;

[0015] Extract the first principal direction of the point cloud;

[0016] Calculate the transformation matrix T2 for transforming the first principal direction to the printing direction and apply it to the point cloud so that the direction of the point cloud is consistent with the deposition direction of the current printing layer, and obtain the finally pose-corrected point cloud as the surface topography of the part;

[0017] Extract the geometric features of the thin wall by the random sampling and statistical method.

[0018] As a preferred technical solution, extracting the high-density distribution area of the point cloud height value specifically includes:

[0019] Perform a histogram statistics on the point cloud height values, divide the data intervals, obtain the number of data points in each data interval, and extract the data intervals where the number of data points exceeds the threshold as the high-density distribution areas of the point cloud height values.

[0020] As a preferred technical solution, an optimal reference plane is fitted based on the improved RANSAC algorithm, specifically including:

[0021] Set the number of iterations and the distance threshold. In the high-density distribution area of the point cloud height values, randomly select 3 points to fit a plane, calculate the normal vector between the three points, obtain the normal vector through cross product, and calculate the distance from each point in the point cloud to this plane;

[0022] Mark the points with a distance less than the set distance threshold as inliers, and through iterative loops, obtain the optimal fitted plane as the reference plane.

[0023] As a preferred technical solution, set the number of iterations and the distance threshold, specifically including:

[0024] Set the number of iterations and the distance threshold through preliminary experiments. In the preliminary experiments, determine the number of iterations through the relationship graph of the number of inliers versus the number of iterations to make the fitting reach the convergence condition, mark the background point cloud, extract the height value distribution characteristics of the background point cloud, and determine the distance threshold through the 3σ principle.

[0025] As a preferred technical solution, randomly select 3 points to fit a plane, and the plane equation is expressed as:

[0026] ax + by + cz + d = 0

[0027] where a, b, and c are the components of the plane normal vector, and d is the constant term of the plane equation;

[0028] Calculate the normal vector between 3 points p 1 , p 2 , p 3 as:

[0029] n 1 = p 2 - p 1

[0030] n 2 = p 3 - p 1

[0031] Obtain the normal vector through cross product: n = n 1 × n 2 ;

[0032] Calculate the distance from each point in the point cloud to this plane, expressed as:

[0033]

[0034] Among them, distance represents the distance from each point in the point cloud to the plane.

[0035] As a preferred technical solution, extracting the first principal direction of the point cloud specifically includes:

[0036] Performing maximum principal component analysis on the point cloud, and taking the eigenvector corresponding to the maximum eigenvalue as the first principal direction, that is, the direction of the current layer of the point cloud.

[0037] As a preferred technical solution, performing maximum principal component analysis on the point cloud, and taking the eigenvector corresponding to the maximum eigenvalue as the first principal direction, specifically includes:

[0038] Centering the point cloud:

[0039] p i =(P i -μ)

[0040] Among them, μ represents the centroid of the point cloud, P i represents the original coordinate of the i-th point in the point cloud, and pi represents the coordinate of the i-th point after centering processing;

[0041] For the centered point cloud, find the covariance matrix:

[0042]

[0043] Among them, N represents the number of points in the point cloud;

[0044] Calculate the eigenvalues and eigenvectors of the covariance matrix:

[0045] Cv = λv

[0046] Among them, C represents the covariance matrix, v represents the eigenvector of the covariance matrix, and λ represents the eigenvalue of the covariance matrix;

[0047] The eigenvector corresponding to the maximum eigenvalue is the first principal direction of the point cloud.

[0048] As a preferred technical solution, extracting the geometric features of the thin wall by the random sampling and statistical method specifically includes:

[0049] Set the number of sampling times and the distance threshold. Randomly select two planes within the height range of 1 / 5 to 4 / 5 of the point cloud, calculate the distances from the point cloud to the two planes, and the points with distances less than the distance threshold are the inliers of the plane. Use the least squares method to fit the inliers in the upper and lower half planes to obtain the upper and lower boundary lines, calculate the wall thickness and center of the thin wall in the current plane, obtain a set of wall thickness change rates and center change rates. Through multiple sampling experiments, multiple sets of wall thickness change rates and center change rates are obtained, and then through histogram statistics, take the average values of the wall thickness change rates and center change rates in the maximum distribution interval as the final wall thickness change rates and center change rates.

[0050] As a preferred technical solution, the formulas for the wall thickness change rate and the center change rate are as follows:

[0051] Width 1 =mean(upperBoundary 1 -lowerBoundary 1 )

[0052] center 1 =mean((upperBoundary 1 +lowerBoundary 1 ) / 2)

[0053] Width 2 =mean(upperBoundary 2 -lowerBoundary 2 ))

[0054] center 2 =mean((upperBoundary 2 +lowerBoundary 2 ) / 2)

[0055]

[0056] Among them, upperBoundary 1 、upperboundary 2 、lowerBoundary 1 、lowerBoundary 2 respectively represent the upper and lower boundary lines of the thin wall in the random planes 1 and 2, Width 1 、Width 2 respectively represent the average distances between the upper and lower boundary lines of the thin wall in the random planes 1 and 2, which are used as the thickness of the thin wall in the current plane, center 1 、center 2respectively represent the average values of the center lines of the thin walls in the random planes 1 and 2, serving as the center positions of the thin walls in the current plane. Width variation rate represents the change rate of the wall thickness of the thin wall, and center variation rate represents the change rate of the center of the thin wall, h 1 、h 2 respectively represent the heights of the random planes 1 and 2.

[0057] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0058] (1) Compared with the existing methods for measuring the surface topography of parts, such as profilometers, etc., the present invention is based on a line laser scanning camera for surface measurement of parts, with advantages such as large depth of field, wide measurement range, fast measurement speed, small and easy-to-integrate equipment, etc., and can achieve online or offline measurement.

[0059] (2) Compared with the existing surface topography measurement and extraction algorithms based on line laser scanning cameras, the present invention proposes an improved RANSAC algorithm, which speeds up the convergence rate and improves the noise resistance of the algorithm.

[0060] (3) Aiming at the problem that the pose of the point cloud is not considered in the prior art, the present invention proposes a method for correcting the pose of the reference plane (printing plane) of the point cloud and a method for correcting the pose of the printing direction of the point cloud, which can improve the accuracy and reliability of the surface topography of the part.

[0061] (4) The present invention uses the random sampling and statistical method to extract the geometric features of the part, reducing the random error of the geometric feature values. Brief Description of the Drawings

[0062] Figure 1 is a schematic flow chart of the method for measuring the surface topography of parts in the laser directed energy process of the present invention;

[0063] Figure 2 is a schematic diagram of the original point cloud of the part surface of the present invention;

[0064] Figure 3 is a schematic histogram of the height values of the point cloud of the present invention;

[0065] Figure 4 is a schematic flow chart of the improved RANSAC fitting of the reference plane of the present invention;

[0066] Figure 5 is a schematic diagram of the fitting result of the reference plane of the present invention;

[0067] Figure 6 is a schematic diagram of the result of removing the background point cloud of the present invention;

[0068] Figure 7Schematic diagram of the mask of the point cloud of the present invention in the XOY plane;

[0069] Figure 8 Schematic diagram of the point cloud noise reduction and pose correction process of the present invention;

[0070] Figure 9 Schematic diagram of the result of removing the noise point cloud of the present invention;

[0071] Figure 10 Schematic diagram of the comparison before and after the pose correction of the reference plane of the present invention;

[0072] Figure 11 Schematic diagram of the comparison before and after the pose correction of the single melt channel and thin wall point cloud of the present invention in the printing direction;

[0073] Figure 12 Flow chart of the random sampling and statistical method for feature extraction of the present invention. Detailed implementation manners

[0074] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0075] Embodiment

[0076] As Figure 1 shown, this embodiment provides a method for measuring the surface topography of a part in a laser directed energy process, including the following steps:

[0077] S1: During or after the processing, through a line laser scanning camera, collect the height information of the part surface, and combine the resolution and scanning speed of the line laser, as Figure 2 shown, reconstruct the point cloud of the part surface;

[0078] In this embodiment, the original data collected by the line laser scanning camera is a (3000*n) height matrix. In the embodiment, the used line laser scanning camera has a laser line length of 30 mm in the scanning plane, including 3000 points. Therefore, in the vertical scanning direction, the resolution is 10 microns. The sampling frequency of the line laser scanning camera is 2500 hz, and the scanning speed is 25 mm / s. Therefore, in the scanning direction, the resolution is 10 microns. Taking the (1, 1) of the height matrix as the coordinate origin, the (x, y) coordinates of each point in the point cloud are as follows:

[0079] x i = 0.010 * i;

[0080] y i = 0.010 * j;

[0081] S2: Perform a histogram statistics on the point cloud height values to extract the high-density distribution region of the point cloud height values;

[0082] As Figure 3 shown, perform a histogram statistics on the point cloud height values, automatically divide the data interval, obtain the number of data points in each data interval, take the 75% quantile of the number of data points as the threshold, and extract the data interval where the number of data points exceeds the threshold as the high-density distribution region of the point cloud height values;

[0083] S3: As Figure 4 shown, perform an improved RANSAC fitting on the printed reference plane to obtain the optimal reference plane;

[0084] In this embodiment, randomly select 3 points in the high-density distribution region to solve the plane equation, determine the distance between the complete point cloud and the plane, calculate the number of inliers, as Figure 5 shown, and obtain the optimal reference plane equation through iteration;

[0085] In this embodiment, perform an improved RANSAC fitting on the printed reference plane to obtain the optimal reference plane. First, set the number of iterations and the distance threshold to ensure that the fitting can converge and meet the fitting accuracy requirements within the number of iterations. Then, randomly select three points in the high-density distribution region of the point cloud height values to fit a plane, and the plane equation can be expressed as:

[0086] ax + by + cz + d = 0

[0087] where a, b, and c are the components of the plane normal vector, and d is the constant term of the plane equation;

[0088] Calculate the normal vector between the three points:

[0089] n 1 = p 2 - p 1

[0090] n 2 = p 3 - p 1

[0091] Obtain the normal vector by cross product: n = n 1 × n 2 ;

[0092] Calculate the distance from each point in the point cloud to this plane, expressed as:

[0093]

[0094] Points with a distance less than a pre - set distance threshold are marked as inliers. The number of inliers is counted. The more inliers there are, the better the fitting result. Through iterative loops, an optimal fitted plane is obtained as the reference plane. The pre - set number of iterations and distance threshold are determined through preliminary experiments. In the preliminary experiments, the number of iterations is determined by the relationship graph of the number of inliers versus the number of iterations to make the fitting reach the convergence condition. In the preliminary experiments, by manually annotating the background point cloud, the height value distribution characteristics of the background point cloud are extracted, and the distance threshold is determined through the 3σ principle;

[0095] S4: As Figure 6 shown, the inliers of the reference plane are removed as the background point cloud. The inliers of the reference plane are caused by the substrate surface. In this embodiment, the background point cloud is removed to avoid interference from the background point cloud in the subsequent extraction of the first principal direction of the target point cloud;

[0096] S5: As Figure 7 shown, the point cloud is projected onto the XOY plane to generate a mask image;

[0097] S6: As Figure 8 shown, the mask image retains the largest connected component, removes the noise point cloud caused by splashes attached to the substrate surface, and then acts on the point cloud. As Figure 9 shown, the denoised point cloud is obtained;

[0098] S7: Calculate the transformation matrix T1 for the reference plane transformed to the XOY plane;

[0099] S8: The transformation matrix T1 is applied to the denoised point cloud to make the printing reference plane of the point cloud coincide with the XOY plane and correct the height value of the point cloud;

[0100] As Figure 10 shown, in the figure, (a) is the front view of the point cloud with the reference plane uncorrected. As can be seen from the red dashed line marked in the front view, the point cloud as a whole has a pitch angle in the front view, and the z - coordinate of the point cloud cannot correctly reflect the height value of the point cloud. In the figure, (b) is the front view of the point cloud after the reference plane correction. The bottom of the point cloud coincides with the horizontal line, indicating that the reference plane of the point cloud coincides with the XOY plane.

[0101] S9: Extract the first principal direction of the point cloud;

[0102] Due to the installation error of the line - laser scanning camera, the laser line cannot be perfectly perpendicular to the printing direction, and the direction of the scanned point cloud will have a certain deviation from the actual printing direction. In this embodiment, the point cloud is subjected to principal component analysis (PCA) to obtain the eigenvector corresponding to the largest eigenvalue as the first principal direction, that is, the direction of the current - layer point cloud. Specifically, it includes:

[0103] First, centralize the point cloud: p i =(P i- μ), where μ represents the centroid of the point cloud, and P i represents the original coordinates of the i-th point in the point cloud, and p i represents the coordinates of the i-th point after centering;

[0104] For the centered point cloud, calculate the covariance matrix:

[0105]

[0106] where N represents that there are N points in the point cloud;

[0107] Eigenvalue decomposition: Calculate the eigenvalues and eigenvectors of the covariance matrix through the following formula:

[0108] Cv = λv

[0109] where C represents the covariance matrix, v represents the eigenvector of the covariance matrix, and λ represents the eigenvalue of the covariance matrix;

[0110] The eigenvector corresponding to the largest eigenvalue is the first principal direction of the point cloud;

[0111] S10: Calculate the transformation matrix T2 for transforming the first principal direction to the printing direction, and apply it to the point cloud so that the direction of the point cloud is consistent with the printing direction of the current deposition layer, and obtain the finally pose-corrected point cloud as the surface topography of the part;

[0112] As Figure 11 shown, in the figure, (a) and (c) are the side views of the single-pass point cloud and the thin-wall point cloud without correction of the printing direction. It can be seen from the figure that there is an inclination angle between the direction of the point cloud and the printing direction; (b) and (d) in the figure are the side views of the single-pass point cloud and the thin-wall point cloud after correction of the printing direction. In the side view, the boundary line of the point cloud is clear.

[0113] S11: As Figure 12 shown, extract the geometric features of the thin wall by the random sampling and statistical method;

[0114] In this embodiment, the sampling times (preferably 1000) and the distance threshold (preferably 0.02) are first set. The larger the sampling times, the better the noise resistance of the result, but the longer the operation time. The smaller the distance threshold, the higher the accuracy, but the lower the noise resistance. Within the height range of 1 / 5 to 4 / 5 of the point cloud, two planes (random planes 1 and 2) are randomly selected, and the distances from the point cloud to the two planes are calculated. The points with distances less than the distance threshold are the inliers of the plane. The inliers within the upper and lower half planes are fitted by the least squares method to obtain the upper and lower boundary lines, and then the wall thickness and the center of the thin wall in the current plane are calculated. Further, a set of wall thickness change rates and center change rates are obtained. Through multiple sampling experiments, multiple sets of wall thickness change rates and center change rates are obtained. Then, through histogram statistics, the average values of the wall thickness change rate and the center change rate in the maximum distribution interval are taken as the final wall thickness change rate and center change rate.

[0115] Among them, the formulas for the wall thickness change rate and the center change rate are as follows:

[0116] Width 1 =mean(upperBoundary 1 -lowerBoundary 1 )

[0117] center 1 =mean((upperBoundary 1 +lowerBoundary 1 ) / 2)

[0118] Width 2 =mean(upperBoundary 2 -lowerBoundary 2 )

[0119] center 2 =mean((upperBoundary 2 +lowerBoundary 2 ) / 2)

[0120]

[0121] Among them, upperBoundary 1 , upperboundary 2 , lowerBoundary 1 , lowerBoundary 2 respectively represent the upper and lower boundary lines of the thin wall in random planes 1 and 2, Width 1 , Width 2respectively represent the average distance between the upper and lower boundary lines of the thin wall in the random planes 1 and 2, which is used as the thickness of the thin wall in the current plane, center 1 、center 2 respectively represent the average values of the center lines of the thin wall in the random planes 1 and 2, which are used as the center positions of the thin wall in the current plane. Width variation rate represents the variation rate of the wall thickness of the thin wall, and center variation rate represents the variation rate of the center of the thin wall, h 1 、h 2 respectively represent the heights of the random planes 1 and 2.

[0122] The above embodiments are the preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A method for measuring the surface topography of a part using a laser directed energy process, characterized in that: The steps include: The height information of the part surface is collected based on the line laser scanning camera, and the point cloud of the part surface is reconstructed by combining the resolution and scanning speed of the line laser; Extract high-density distribution areas of point cloud height values; The optimal reference plane is obtained by fitting based on the improved RANSAC algorithm; The inner points of the reference plane are removed as background point clouds, and the inner points of the reference plane are caused by the substrate surface; Project the point cloud onto the XOY plane to generate a mask image; The mask image retains the largest connected domain, removes the noise point cloud, and obtains the denoised point cloud; Calculate the transformation matrix T1 from the reference plane to the XOY plane; The transformation matrix T1 is applied to the denoised point cloud to make the printing reference plane of the point cloud coincide with the XOY plane and correct the height value of the point cloud; Extract the first main direction of the point cloud; Calculate the transformation matrix T2 from the first principal direction to the printing direction, and apply it to the point cloud so that the direction of the point cloud is consistent with the deposition direction of the current printing layer, and obtain the final pose-corrected point cloud as the surface morphology of the part; The geometric features of thin walls are extracted by random sampling statistics.

2. The method for measuring the surface topography of a part in a laser directed energy process according to claim 1, characterized in that: Extract high-density distribution areas of point cloud height values, including: Perform histogram statistics on the point cloud height values, divide the data intervals, obtain the number of data points in each data interval, and extract the data intervals where the number of data points exceeds the threshold as the high-density distribution area of ​​the point cloud height values.

3. The method for measuring the surface topography of a part in a laser directed energy process according to claim 1, characterized in that: The optimal reference plane is obtained by fitting based on the improved RANSAC algorithm, including: Set the number of iterations and the distance threshold, randomly select three points in the high-density distribution area of ​​the point cloud height value to fit a plane, calculate the normal vector between the three points, obtain the normal vector by cross multiplication, and calculate the distance from each point in the point cloud to the plane; Points whose distance is less than the set distance threshold are recorded as internal points. Through loop iteration, the optimal fitting plane is obtained as the reference plane.

4. The method for measuring the surface topography of a part in a laser directed energy process according to claim 3, characterized in that: Set the number of iterations and distance threshold, including: The number of iterations and the distance threshold are set through preliminary experiments. In the preliminary experiment, the number of iterations is determined through the relationship diagram between the number of internal points and the number of iterations, so that the fitting reaches the convergence condition, the background point cloud is annotated, the height value distribution characteristics of the background point cloud are extracted, and the distance threshold is determined by the 3σ principle.

5. The method for measuring the surface topography of a part in a laser directed energy process according to claim 3, characterized in that: Randomly select 3 points to fit a plane, and the plane equation is expressed as: ax+by+cz+d=0 Among them, a, b, c are the components of the plane normal vector, and d is the constant term of the plane equation; Calculate the normal vector between the three points p1, p2, and p3 as: n1=p2-p1 n2=p3-p1 Cross product to get the normal vector: n = n1 × n2; Calculate the distance from each point in the point cloud to the plane, expressed as: Among them, distance represents the distance from each point in the point cloud to the plane.

6. The method for measuring the surface topography of a part in a laser directed energy process according to claim 1, characterized in that: Extract the first main direction of the point cloud, including: Perform the maximum principal component analysis on the point cloud and obtain the eigenvector corresponding to the maximum eigenvalue as the first principal direction, that is, the direction of the point cloud of the current layer.

7. The method for measuring the surface topography of a part in a laser directed energy process according to claim 6, characterized in that: Perform the maximum principal component analysis on the point cloud and obtain the eigenvector corresponding to the maximum eigenvalue as the first principal direction, which includes: Centralized processing of point cloud: p i =(P i -μ) Among them, μ represents the centroid of the point cloud, P i represents the original coordinates of the i-th point in the point cloud, p i Represents the coordinates of the i-th point after centralization; For the centered point cloud, find the covariance matrix: Where N represents the number of points in the point cloud; Compute the eigenvalues ​​and eigenvectors of the covariance matrix: Cv=λv Where C represents the covariance matrix, v represents the eigenvector of the covariance matrix, and λ represents the eigenvalue of the covariance matrix; The eigenvector corresponding to the maximum eigenvalue is the first principal direction of the point cloud.

8. The method for measuring the surface topography of a part in a laser directed energy process according to claim 1, characterized in that: The geometric features of thin walls are extracted by random sampling statistics, including: The number of sampling times and the distance threshold are set, and two planes are randomly selected within the 1 / 5 to 4 / 5 height of the point cloud. The distance from the point cloud to the two planes is calculated. The points whose distance is less than the distance threshold are the inner points of the plane. The inner points in the upper and lower half planes are fitted with the least squares method to obtain the upper and lower boundary lines. The wall thickness and center of the thin wall in the current plane are calculated to obtain a set of wall thickness change rate and center change rate. Multiple sets of wall thickness change rate and center change rate are obtained through multiple sampling experiments. Then, through histogram statistics, the average value of the wall thickness change rate and center change rate in the maximum distribution interval is taken as the final wall thickness change rate and center change rate.

9. The method for measuring the surface topography of a part in a laser directed energy process according to claim 8, characterized in that: The formulas for wall thickness change rate and center change rate are as follows: Width1=mean(upperBoundary1-lowerBoundary1) center1=mean((upperBoundary1+lowerBoundary1) / 2) Width2=mean(upperBoundary2-lowerBoundary2) center2=mean((upperBoundary2+lowerBoundary2) / 2 Among them, upperBoundary1, upperboundary2, lowerBoundary1, lowerBoundary2 represent the upper and lower boundary lines of the thin wall in random planes 1 and 2, respectively; Width1 and Width2 represent the average distance between the upper and lower boundary lines of the thin wall in random planes 1 and 2, respectively, as the thickness of the thin wall in the current plane; center1 and center2 represent the average values ​​of the center lines of the thin walls in random planes 1 and 2, respectively, as the center position of the thin wall in the current plane; Width variation rate represents the rate of change of the thin wall thickness; center variation rate represents the rate of change of the thin wall center; h1 and h2 represent the heights of random planes 1 and 2, respectively.

Images