Fastener concave-convex amount measuring method based on monocular surface structured light

Abstract

The invention provides a fastener concave-convex amount measuring method based on monocular surface structured light. The fastener concave-convex amount measuring method comprises the steps that internal parameters of a monocular camera and a projector are determined; calculating the relative position of the monocular camera and the projector according to the internal parameters of the monocular camera and the projector; in the relative position, the projector projects a grating pattern to the fastener surface, while the monocular camera captures a second image of the grating pattern; detecting a fastener area from the second image; calculating an absolute phase value of the third image of the fastener region; according to the internal reference of the monocular camera and the projector, the relative position of the monocular camera and the projector, and the absolute phase value of a third image of the fastener area, converting the absolute phase value of each pixel in the third image into coordinates of the object surface in a three-dimensional space by using a phase-height mapping relation, and obtaining a three-dimensional point cloud; and calculating the concave-convex amount of the fastener based on the three-dimensional point cloud.

Description

Technical Field

[0001] The present application belongs to the technical field of structured light three-dimensional reconstruction, and specifically relates to a method for measuring the concave-convex quantity of fasteners based on monocular structured light. Background Art

[0002] In the industrial manufacturing process, the convexity and concavity of fasteners is an important indicator to ensure assembly quality, which directly affects the appearance, function and service life of the product.

[0003] Traditional methods of measuring bumps and depressions mainly rely on contact tools, such as micrometers and calipers, which are not only inefficient but also prone to physical damage to the measured surface, affecting product quality. In addition, complex 3D profile measurements usually require expensive equipment and time-consuming operations, making it difficult to meet the needs of efficient and accurate measurement on modern production lines. Summary of the invention

[0004] Purpose of the invention: To provide a method for measuring the concave-convex amount of fasteners based on monocular structured light, which can solve the problems of insufficient detection accuracy, low efficiency and high detection cost in the prior art.

[0005] The present application provides a method for measuring the concave-convex amount of a fastener based on monocular structured light, the method comprising:

[0006] Determine the intrinsic parameters of the monocular camera and projector;

[0007] Calculating the relative position of the monocular camera and the projector according to the internal parameters of the monocular camera and the projector;

[0008] In the relative position, the projector projects a grating pattern onto a surface of the fastener, and at the same time, the monocular camera captures a second image of the grating pattern;

[0009] detecting a fastener region from the second image;

[0010] calculating an absolute phase value of the third image of the fastener region by a phase unwrapping algorithm;

[0011] According to the internal parameters of the monocular camera and the projector, the relative positions of the monocular camera and the projector, and the absolute phase value of the third image of the fastener area, the absolute phase value of each pixel in the third image is converted into the coordinates of the object surface in the three-dimensional space by using the phase-height mapping relationship to obtain a three-dimensional point cloud;

[0012] The three-dimensional point cloud is preprocessed for denoising, and a plane is fitted by a least square method to calculate the concave and convex amount of the fastener.

[0013] Preferably, the determining of the internal parameters of the monocular camera and the projector includes:

[0014] The projector projects a grating pattern onto a calibration plate; wherein the calibration plate comprises black and white dots;

[0015] The monocular camera captures a first image of the grating pattern;

[0016] According to the first image, the monocular camera and the projector are calibrated by using the Zhang Zhengyou calibration method to determine the intrinsic parameters of the monocular camera and the projector.

[0017] Preferably, detecting the fastener area from the second image comprises:

[0018] Processing the second image by using a Canny edge detection operator to enhance the clarity of the edge of the second image;

[0019] Automatically segmenting a first region according to a grayscale gradient change in the second image, and further filtering out noise interference in the first region;

[0020] The position of the fastener in the first area is located by using a geometric feature extraction algorithm, and the fastener area is extracted from the first area.

[0021] Preferably, calculating the absolute phase value of the third image of the fastener region by a phase unwrapping algorithm comprises:

[0022] Assuming there are M different phase shifts, for each phase φm, the intensity Im at a point on the surface of the object is expressed as:

[0023]

[0024] Where I0(x,y) is the background intensity, I1(x,y) is the modulation, and φ(x,y) is the phase of a point on the surface of the object;

[0025] By solving the brightness of multiple images, the wrapped phase of a pixel point (x, y) on the surface of the fastener is obtained as:

[0026]

[0027] After calculating the wrapped phase, the period G(x,y) of the stripes is obtained by decoding the Gray code image, and then combined with the phase image to solve the absolute phase value of each point on the surface of the object.

[0028]

[0029] Preferably, the denoising preprocessing of the three-dimensional point cloud, fitting a plane by the least square method, and calculating the concave-convex amount of the fastener include:

[0030] A filtering method based on statistical analysis is used to remove outliers or points with large errors in the three-dimensional point cloud, and Gaussian filtering technology is used to smooth the point cloud data and reduce noise;

[0031] Select the effective point cloud area around the fastener on the denoised three-dimensional point cloud, use the least squares method to fit the reference plane of the area, and restore the surface morphology of the object;

[0032] According to the distance between the fitted plane and the vertex of the fastener, the average value of the distance is calculated to obtain the concave-convex amount of the fastener.

[0033] Preferably, the filtering method based on statistical analysis, for removing outliers or points with large errors in the three-dimensional point cloud, comprises:

[0034] For each point in the three-dimensional point cloud, calculate the Euclidean distance between it and a specified number of nearest neighboring points, and calculate the average of these distances;

[0035] A threshold is set based on global statistical data, and points whose distance deviates from the mean by more than a certain standard deviation are marked as outliers. All point clouds marked as outliers are removed, leaving only credible three-dimensional point clouds.

[0036] Preferably, the applying of Gaussian filtering technology to smooth the point cloud data and reduce noise includes:

[0037] For each point p in the 3D point cloud, select its neighborhood point set N(p), use the Gaussian kernel function to perform weighted averaging on the points in the neighborhood, and calculate the position of the new smoothed point p':

[0038]

[0039] Where G(d(p,q)) is the Gaussian weight, calculated based on the distance d(p,q) between point p and its neighbor point q, where q is a point in the neighborhood N(p);

[0040] Update the position of each point to the smoothed new position p' to obtain the filtered point cloud.

[0041] Preferably, the method of selecting a valid point cloud area around the fastener on the denoised three-dimensional point cloud, fitting a reference plane of the area using the least squares method, and restoring the surface morphology of the object; and calculating the average value of the distance between the fitted plane and the vertex of the fastener to obtain the concave-convex amount of the fastener includes:

[0042] For the filtered point p', the plane equation of the skin surface α can be obtained by fitting the point set on the skin surface into a plane using the least squares method:

[0043] Aα x+B α y+C α z+D α =0

[0044] Among them, A α , B α , C α and D α is the coefficient of the skin plane equation, and A α , B α and C α Different from 0 at the same time;

[0045] After the skin α plane equation is obtained, traverse the point set P of the fastener convex surface β β , calculate P point by point i β (x i ,y i ,z i ), (i=1,2,3.....j) the distance r from the skin plane i :

[0046]

[0047] Finally, calculate the point set P β The average value of each point to the skin plane is the convexity r of the fastener:

[0048]

[0049] Where j is the number of point clouds on the surface of the fastener head.

[0050] This application has the following technical effects:

[0051] The present invention proposes a fastener concave-convex quantity measurement method based on monocular structured light, aiming to solve the problems of insufficient detection accuracy, complex operation, and low efficiency in the prior art. By calculating the absolute phase value through the phase unwrapping algorithm and fitting the three-dimensional point cloud plane with the least squares method, this method can achieve high-precision detection of the fastener concave-convex quantity. The system is not only easy to operate and has high measurement efficiency, but also can reduce hardware costs and improve the automation of the detection process. It is suitable for the rapid measurement of large-scale fasteners. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart of a method for measuring the concave-convex quantity of a fastener based on monocular structured light in an embodiment of the present application;

[0053] Figure 2 This is an example diagram of fastener area extraction in an embodiment of the present application;

[0054] Figure 3Schematic diagram of calculation of the concave-convex amount of a fastener in an embodiment of the present application. DETAILED DESCRIPTION

[0055] It should be noted that in the field of modern industrial inspection, non-contact measurement technology has gradually replaced traditional contact measurement methods due to its advantages of non-destructive, high-precision and high-speed measurement of the surface of the object being measured. Non-contact measurement technology obtains three-dimensional information of the target surface through optics, sound waves or other physical principles, avoiding surface damage and measurement errors that may be caused by contact measurement. On the production line, especially in the measurement needs of fine surfaces or tiny features, non-contact measurement technology shows great application potential and can achieve large-scale, high-efficiency automated inspection.

[0056] The method for measuring the concave-convex quantity based on monocular structured light is a typical application of non-contact measurement technology. A grating pattern is projected onto the surface to be measured, and a monocular camera is used to capture the deformation of the grating pattern on the surface to be measured. According to the degree of deformation of the grating pattern, combined with the principle of triangulation, the three-dimensional morphology of the fastener surface can be accurately reconstructed to achieve high-precision non-contact measurement of the concave-convex quantity of the fastener. This technology not only improves the measurement efficiency, but also avoids the damage to the surface caused by traditional contact measurement, and is suitable for real-time online detection in large-scale production environments. Therefore, this application proposes a method for measuring the concave-convex quantity of fasteners based on monocular structured light.

[0057] See also Figure 1-Figure 3 The present application embodiment provides a fastener concave-convex measurement method based on monocular structured light, such as Figure 1 As shown, Figure 1 The flowchart of the method for measuring the concave-convex quantity of fasteners based on monocular structured light in the embodiment of the present application includes:

[0058] S1: Use Zhang Zhengyou calibration method to determine the intrinsic parameters and relative positions of the monocular camera and projector by projecting a specific grating pattern;

[0059] Wherein, step S1 comprises:

[0060] S11: Project a specific grating pattern onto a black and white dot calibration plate through a projector, and capture the image of the projected grating pattern with a monocular camera;

[0061] S12: extract the corner points on the calibration plate, calibrate the monocular camera and the projector using the Zhang Zhengyou calibration method, and determine the internal parameters of the monocular camera and the projector;

[0062] S13: Calculate the relative position of the monocular camera and the projector through the feature point information on the calibration plate.

[0063] Specifically, to calibrate the monocular camera and projector, a solid dot calibration plate with black circles on a white background is needed. The posture of the calibration plate is adjusted, and at least 10 groups of images are taken with the monocular camera. The Zhang Zhengyou calibration method is used to obtain the rotation matrices Rc, Rp and translation matrices Tc, Tp of the monocular camera and projector respectively.

[0064] According to the obtained monocular camera parameters Rc, Tc and projector parameters Rp, Tp, use the following formula to obtain R and T, and complete the calibration of the monocular camera and projector. After the calibration is completed, the positions of the monocular camera and projector are fixed, and the relative positions cannot be changed, otherwise recalibration is required.

[0065]

[0066] S2: The monocular camera synchronously collects the grating image projected on the surface of the fastener, detects and extracts the fastener area in the image;

[0067] Wherein, step S2 comprises:

[0068] S21: Use a projector to project a specific grating pattern onto the surface of a target fastener, and use a monocular camera to capture a grating image of the surface of the target fastener;

[0069] S22: Processing the acquired raster image by using a Canny edge detection operator to extract a skeleton line of the edge of the fastener;

[0070] S23: Automatically segment the fastener area according to the grayscale gradient change in the image, and further filter out the noise interference in the image;

[0071] S24: Through the geometric feature extraction algorithm, the position of the fastener in the image is accurately located, and the fastener area is extracted to prepare for the subsequent concave and convex measurement.

[0072] Specifically, the projector sequentially projects multiple specially made grating patterns onto the surface of the target fastener, and the monocular camera captures the grating image of the surface of the target fastener, such as Figure 2(b) shows the edge contour information initially extracted by the Canny operator. Three indicators are selected to filter the edge skeleton line of the fastener area: the first is the perimeter, that is, the number of pixels occupied by the edge skeleton line, denoted as Length. Using the perimeter as a threshold can filter out contours that are too small, and avoid mistaking noise, defects or other irrelevant small edges for the contours of the target object; the second is the area, that is, the area of ​​the edge skeleton line, denoted as Area. When the skeleton line is not closed, its area is the number of pixels, and when the skeleton line is closed, its area is the total number of pixels occupied by the closed area; the third is the ratio of the area of ​​the edge skeleton line to the perimeter, denoted as Ratio. The ratio of area to perimeter is an indicator that reflects the compactness of the contour shape. For a given perimeter, contours with larger areas are usually more regular shapes, while contours with smaller areas may be slender or irregular shapes. Based on this, the edge skeleton line of the fastener area should satisfy the following formula:

[0073]

[0074] Among them, T0 is set to 5, the lower limit Lengthmin is set to 150, the upper limit Lengthmax is set to 300, and the lower limit Areamin is set to 1500, and the upper limit Areamax is set to 50000. Figure 2 (c) shows the edge contour after culling.

[0075] Ellipse fitting is performed on each contour using the RANSAC algorithm, such as Figure 2 (d) shows the fitted ellipse, and the circular contours that meet the conditions are screened out according to the set error threshold and eccentricity threshold. The detected contours are divided into two groups, a low threshold group and a high threshold group, and similar contours in the two groups are matched according to the shape characteristics of the contours. Finally, the matched contours are merged and a mask image containing all contours is generated, as shown in Figure 2 The circular contour fitted in (e) needs to have its radius enlarged by 1.8 times to prepare for the subsequent skin plane fitting. Figure 2 (f) is the final generated Mask.

[0076] S3: Using the grating image obtained after preprocessing, the absolute phase value of the grating image is calculated by the phase unwrapping algorithm. According to the system parameters and the phase-height mapping relationship, the absolute phase value of each pixel is converted into the coordinates of the object surface in three-dimensional space to obtain a three-dimensional point cloud;

[0077] Wherein, step S3 comprises:

[0078] S31: Perform phase unwrapping processing on the grating image to eliminate phase jumps and fuzzy areas, and ensure the continuity and accuracy of the phase value;

[0079] S32: Based on the internal parameters of the monocular camera and the projector and their relative positions, combined with the absolute phase value obtained by phase unwrapping, a phase-height mapping model is established;

[0080] S33: Calculate the specific coordinates of the object surface in three-dimensional space through the mapping relationship between the absolute phase value and the height corresponding to the pixel point, and generate high-precision three-dimensional point cloud data.

[0081] Specifically, the phase information of each stripe on the grating image is obtained by the Gray code-phase shift method. Assuming that the mth grating pattern has M different phase shifts, for each phase φm, the intensity Im at a point on the object surface can be expressed as:

[0082]

[0083] Where I0(x,y) is the background intensity, I1(x,y) is the modulation degree, and φ(x,y) is the phase of the point on the surface of the object.

[0084] By solving the brightness of each grating image, the wrapped phase of a pixel point (x, y) on the surface of the fastener can be obtained as:

[0085]

[0086] After calculating the wrapped phase, the period G(x,y) of the stripes is obtained by decoding the Gray code image in the special grating image, and then combined with the wrapped phase to solve the continuous phase value of each point on the surface of the object.

[0087] According to the phase-height mapping relationship, combined with the internal parameters of the monocular camera and projector and their relative positions, the height information of the object surface is mapped to the camera coordinate system, and the X, Y, and Z coordinates of each pixel in the three-dimensional space are calculated to obtain three-dimensional point cloud data.

[0088] S4: perform denoising preprocessing on the 3D point cloud, fit the plane using the least squares method, and calculate the concave and convex amount of the fastener;

[0089] Wherein, step S4 comprises:

[0090] S41: Remove outliers or points with large errors in the point cloud through a filtering method based on statistical analysis, and apply Gaussian filtering technology to smooth the point cloud data and reduce noise.

[0091] S42: Select a valid point cloud area around the fastener on the denoised point cloud, and use the least squares method to fit the reference plane of the area to accurately restore the surface morphology of the object.

[0092] S43: Calculate an average value of the distances between the fitted plane and the vertices of the fastener, thereby obtaining the concave-convex amount of the fastener.

[0093] Specifically, for each point in the point cloud, the Euclidean distance between it and the specified number of nearest neighboring points is calculated, and the average of these distances is calculated. A threshold is set based on global statistical data, and points whose distances deviate from the average by more than a certain standard deviation are marked as outliers. All point cloud data marked as outliers are removed, leaving only credible point cloud data.

[0094] For each point p in the point cloud, select its neighborhood point set N(p), use the Gaussian kernel function to perform weighted averaging on the points in the neighborhood, and calculate the position of the new smoothed point p':

[0095]

[0096] Where G(d(p,q)) is the Gaussian weight, which is calculated based on the distance d(p,q) between point p and its neighboring point q, where q is a point in the neighborhood N(p). The position of each point is updated to the smoothed new position p' to obtain the filtered point cloud.

[0097] For the filtered point p', Figure 3 The figure shows the calculation diagram of the concave-convex amount of the fastener. The plane equation of the skin surface α can be obtained by fitting the point set on the skin surface into a plane using the least squares method:

[0098] A α x+B α y+C α z+D α =0

[0099] Among them A α , B α , C α and D α is the coefficient of the skin plane equation, and A α , B α and C α Not all are 0.

[0100] After the skin α plane equation is obtained, traverse the point set P of the fastener convex surface β β , calculate P point by point i β (x i ,y i ,z i ), (i=1,2,3.....j) the distance r from the skin plane i :

[0101]

[0102] Finally, calculate the point set P β The average value of each point to the skin plane is the convexity r of the fastener:

[0103]

[0104] Where j is the number of point clouds on the surface of the fastener head.

[0105] The present invention discloses a method for measuring the concave-convex quantity of fasteners based on monocular surface structured light, which belongs to the field of three-dimensional reconstruction and is implemented based on a monocular camera and a projector, wherein the projector is used to project a specific grating pattern, and the monocular camera is used to capture the grating image of the surface of the target fastener. The method comprises: using the Zhang Zhengyou calibration method to determine the internal parameters and relative positions of the monocular camera and the projector; synchronously collecting the stripe image projected on the surface of the fastener by the monocular camera, detecting and extracting the fastener area in the image; using the grating image obtained after preprocessing, calculating the absolute phase value of the grating image by a phase unwrapping algorithm, combining the internal parameters and relative positions of the monocular camera and the projector, and using the phase-height mapping relationship, converting the absolute phase value of each pixel into the coordinates of the object surface in three-dimensional space, and obtaining a three-dimensional point cloud; performing denoising preprocessing on the three-dimensional point cloud, fitting the fastener skin plane by the least square method, and calculating the concave-convex quantity of the fastener. The present invention realizes high-precision measurement of the concave-convex quantity of the fastener, and is suitable for the concave-convex quantity measurement under various complex working conditions.

[0106] The present application can solve the problems of insufficient detection accuracy, low efficiency and high detection cost in the prior art.

[0107] The Zhang Zhengyou calibration method is used to determine the intrinsic parameters of the projector and camera and their relative positions by projecting a specific grating pattern.

[0108] The monocular camera synchronously collects the fringe image projected on the surface of the fastener, detects the fastener area in the image, and extracts the area of ​​interest in the image.

[0109] The absolute phase value of the grating image is calculated by the phase unwrapping algorithm using the grating image obtained after preprocessing. According to the system parameters and the phase-height mapping relationship, the absolute phase value of each pixel is converted into the coordinates of the object surface in three-dimensional space to obtain a three-dimensional point cloud.

[0110] The 3D point cloud is preprocessed for denoising, and the plane is fitted using the least squares method to calculate the concave and convex amount of the fastener.

[0111] The grating pattern is projected onto a specially made black and white dot calibration plate by a projector, and the image of the projected grating pattern is captured by a camera.

[0112] Extract the corner points on the calibration plate, use Zhang Zhengyou calibration method to calibrate the camera and projector, and determine the intrinsic parameters of the camera and projector.

[0113] The relative positions of the camera and projector are calculated through the feature point information on the calibration plate.

[0114] The collected fringe images are processed by the Canny edge detection operator to enhance the clarity of the fastener edges.

[0115] According to the grayscale gradient changes in the image, the fastener area is automatically segmented to further filter out the noise interference in the image.

[0116] Through the geometric feature extraction algorithm, the position of the fastener in the image is accurately located and the region of interest of the fastener is extracted.

[0117] A projector is used to project a specific grating pattern onto the surface of the target fastener, and an industrial camera captures the modulated grating image of the surface of the target fastener.

[0118] Phase unwrapping is performed on the grating image to eliminate phase jumps and fuzzy areas and ensure the continuity and accuracy of the phase value.

[0119] Based on the system calibration parameters (including internal and external parameters of the camera and geometric parameters of the projector), a phase-height mapping model is established to convert the phase value into depth information.

[0120] The specific coordinates of the object surface in three-dimensional space are calculated through the absolute phase values ​​corresponding to the pixel points, generating high-precision three-dimensional point cloud data.

[0121] Through the filtering method based on statistical analysis, outliers or points with large errors in the point cloud are removed, and the median filtering technology is applied to smooth the point cloud data and reduce noise.

[0122] The effective point cloud area around the fastener is selected on the denoised point cloud, and the reference plane of the area is fitted using the least squares method to accurately restore the surface morphology of the object.

[0123] According to the distance between the fitted plane and the fastener vertex, the concave and convex amount of the fastener is calculated, and the fitting result is optimized. The error is reduced through an iterative algorithm to ensure the accuracy and stability of the concave and convex amount calculation.

Claims

1. A method for measuring the concave-convex amount of fasteners based on monocular structured light, characterized in that: The method comprises: Determine the intrinsic parameters of the monocular camera and projector; Calculating the relative position of the monocular camera and the projector according to the internal parameters of the monocular camera and the projector; In the relative position, the projector projects a grating pattern onto a surface of the fastener, and at the same time, the monocular camera captures a second image of the grating pattern; detecting a fastener region from the second image; calculating an absolute phase value of the third image of the fastener region by a phase unwrapping algorithm; According to the internal parameters of the monocular camera and the projector, the relative positions of the monocular camera and the projector, and the absolute phase value of the third image of the fastener area, the absolute phase value of each pixel in the third image is converted into the coordinates of the object surface in the three-dimensional space by using the phase-height mapping relationship to obtain a three-dimensional point cloud; The three-dimensional point cloud is preprocessed for denoising, and a plane is fitted by a least square method to calculate the concave and convex amount of the fastener.

2. The method according to claim 1, characterized in that The step of determining the internal parameters of the monocular camera and the projector includes: The projector projects a grating pattern onto a calibration plate; wherein the calibration plate comprises black and white dots; The monocular camera captures a first image of the grating pattern; According to the first image, the monocular camera and the projector are calibrated by using the Zhang Zhengyou calibration method to determine the intrinsic parameters of the monocular camera and the projector.

3. The method according to claim 2, characterized in that The detecting the fastener area from the second image comprises: Processing the second image by using a Canny edge detection operator to enhance the clarity of the edge of the second image; Automatically segmenting a first region according to a grayscale gradient change in the second image, and further filtering out noise interference in the first region; The position of the fastener in the first area is located by using a geometric feature extraction algorithm, and the fastener area is extracted from the first area.

4. The method according to claim 3, characterized in that Calculating the absolute phase value of the third image of the fastener region by a phase unwrapping algorithm includes: Assuming there are M different phase shifts, for each phase φm, the intensity Im at a point on the surface of the object is expressed as: Where I0(x,y) is the background intensity, I1(x,y) is the modulation, and φ(x,y) is the phase of a point on the surface of the object; By solving the brightness of multiple images, the wrapped phase of a pixel point (x, y) on the surface of the fastener is obtained as: After calculating the wrapped phase, the period G(x,y) of the stripes is obtained by decoding the Gray code image, and then combined with the phase image to solve the absolute phase value of each point on the surface of the object.

5. The method according to claim 4, characterized in that The denoising preprocessing is performed on the three-dimensional point cloud, and the plane is fitted by the least square method to calculate the concave and convex amount of the fastener, including: A filtering method based on statistical analysis is used to remove outliers or points with large errors in the three-dimensional point cloud, and Gaussian filtering technology is used to smooth the point cloud data and reduce noise; Select the effective point cloud area around the fastener on the denoised three-dimensional point cloud, use the least squares method to fit the reference plane of the area, and restore the surface morphology of the object; According to the distance between the fitted plane and the vertex of the fastener, the average value of the distance is calculated to obtain the concave-convex amount of the fastener.

6. The method according to claim 5, characterized in that The filtering method based on statistical analysis, which removes outliers or points with large errors in the three-dimensional point cloud, includes: For each point in the three-dimensional point cloud, calculate the Euclidean distance between it and a specified number of nearest neighboring points, and calculate the average of these distances; A threshold is set based on global statistical data, and points whose distance deviates from the mean by more than a certain standard deviation are marked as outliers. All point clouds marked as outliers are removed, leaving only credible three-dimensional point clouds.

7. The method according to claim 6, characterized in that The Gaussian filtering technology is applied to smooth the point cloud data and reduce noise, including: For each point p in the 3D point cloud, select its neighborhood point set N(p), use the Gaussian kernel function to perform weighted averaging on the points in the neighborhood, and calculate the position of the new smoothed point p': Where G(d(p,q)) is the Gaussian weight, calculated based on the distance d(p,q) between point p and its neighbor point q, where q is a point in the neighborhood N(p); Update the position of each point to the smoothed new position p' to obtain the filtered point cloud.

8. The method according to claim 7, characterized in that The method further comprises selecting a valid point cloud region around the fastener on the denoised three-dimensional point cloud, fitting a reference plane of the region using the least squares method, and restoring the surface morphology of the object; According to the distance between the fitted plane and the vertex of the fastener, the average value of the distance is calculated to obtain the concave-convex amount of the fastener, including: For the filtered point p', the plane equation of the skin surface α can be obtained by fitting the point set on the skin surface into a plane using the least squares method: A α x+B α y+C α z+D α =0 Among them, A α , B α , C α and D α is the coefficient of the skin plane equation, and A α , B α and C α Different from 0 at the same time; After the skin α plane equation is obtained, traverse the point set P of the fastener convex surface β β , calculated point by point (i=1,2,3.....j) The distance r to the skin plane i : Finally, calculate the point set P β The average value of each point to the skin plane is the convexity r of the fastener: Where j is the number of point clouds on the surface of the fastener head.

Images