Method, device and storage medium for visual measurement of contact marks of aviation spiral bevel gears

Abstract

The present invention discloses a method, device and storage medium for visual measurement of contact marks of aviation spiral bevel gears. The method can use a mark segmentation model to quickly complete the extraction of two-dimensional mark color image data, and then a pre-constructed two-dimensional and three-dimensional conversion matrix can be used to map the color information of the two-dimensional mark color image to a three-dimensional point cloud to obtain color three-dimensional mark point cloud data. Finally, the color three-dimensional mark point cloud data is subjected to tooth surface fitting, tooth surface contour point cloud segmentation, expansion and other operations to complete the automatic measurement of the mark parameters, thereby eliminating the problems of unstable measurement results and low measurement efficiency caused by manual detection. At the same time, when collecting the first laser point cloud data and the first color image data, the working positions of the laser point cloud acquisition device and the color image acquisition device are consistent with the positions when the two-dimensional and three-dimensional conversion matrices are constructed, thereby ensuring the accuracy of the color three-dimensional mark point cloud data converted by the two-dimensional and three-dimensional conversion matrices.

Description

Technical Field

[0001] The invention relates to the field of aviation spiral bevel gears, and in particular to a method, a device and a storage medium for visually measuring contact marks of aviation spiral bevel gears. Background Art

[0002] With the rapid development of China's aviation industry, the aero-engine as its core power system, its key transmission component spiral bevel gear is subject to increasingly higher performance requirements and use loads. Among them, contact mark measurement is a direct and effective means to determine the gear contact state and evaluate the transmission performance. The traditional contact mark measurement is through manual measurement. The specific method is: after completing the color rolling test of the gear transmission, the mark is obtained on the tooth surface by manual stickers; then the quality inspector uses calipers to measure the parameters of the gear mark, judge whether the gear after grinding meets the design requirements, and record the relevant parameters and images. However, this manual measurement method is not only inefficient, but also difficult to ensure measurement accuracy, there are large subjective errors, and it cannot meet the requirements of aero-engines for improving the performance of spiral bevel gears. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a visual measurement method for the contact mark of an aviation spiral bevel gear, which can solve the problem of difficult to ensure measurement accuracy and low measurement efficiency caused by manual measurement.

[0004] The invention also proposes a visual measurement device for the contact mark of an aviation spiral bevel gear and a computer-readable storage medium.

[0005] The visual measurement method for the contact mark of an aviation spiral bevel gear according to the first embodiment of the present invention is applied to a spiral bevel gear contact mark detection platform, the spiral bevel gear contact mark detection platform comprises a motion platform, a laser point cloud acquisition device, and a color image acquisition device, and a placement plate is arranged on the motion platform;

[0006] The contact mark visual measurement method of aviation spiral bevel gears comprises:

[0007] Controlling the placement plate to move, and acquiring first laser point cloud data of the spiral bevel gear to be inspected placed on the placement plate collected by the laser point cloud acquisition device;

[0008] Acquire first color image data of the spiral bevel gear to be detected photographed by the color image acquisition device, and when acquiring the first color image data and when acquiring the first laser point cloud data, the relative position of the spiral bevel gear to be detected and the placement plate remains the same;

[0009] Inputting the first color image data into a pre-constructed print segmentation model to obtain two-dimensional print color image data;

[0010] According to the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix, the two-dimensional print color image is mapped onto the first laser point cloud data to obtain color three-dimensional print point cloud data;

[0011] Performing surface fitting on the color three-dimensional impression point cloud data, and combining it with the first laser point cloud data, determining that the point cloud on the fitting surface is the segmented tooth surface point cloud;

[0012] Expanding the segmented tooth surface point cloud into a plane, completing the imprint measurement, and obtaining the imprint detection result;

[0013] The two-dimensional and three-dimensional transformation matrices are obtained by the following steps:

[0014] Acquire conversion basic data when the calibration block is in a plurality of placement states, each set of the basic conversion data includes the second color image data of the calibration block acquired by the color image acquisition device and the second laser point cloud data of the calibration block acquired by the laser point cloud acquisition device; and when acquiring each set of the conversion basic data, the working position of the laser point cloud acquisition device is the same as the working position of the laser point cloud acquisition device when acquiring the first laser point cloud data, and the working position of the color image acquisition device is the same as the working position of the color image acquisition device when acquiring the first color image data;

[0015] Determine the three-dimensional corner point coordinates of each corner point of the calibration block in the second laser point cloud data in each group of the conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block in the second color image data, and obtain the coordinate correspondence between the multiple three-dimensional corner points and the multiple two-dimensional corner points;

[0016] The two-dimensional and three-dimensional transformation matrices are obtained according to the multiple sets of coordinate correspondences.

[0017] The visual measurement method for contact marks of aviation spiral bevel gears according to the embodiment of the present invention has at least the following beneficial effects:

[0018] The spiral bevel gear contact mark detection platform can realize the automatic collection of the first laser point cloud data and the first color image data of the spiral bevel gear to be detected, and the mark segmentation model can be used to quickly complete the extraction of the two-dimensional mark color image data. After that, the pre-constructed two-dimensional and three-dimensional conversion matrix can be used to map the color information of the two-dimensional mark color image to the three-dimensional point cloud to obtain the color three-dimensional mark point cloud data. Finally, the color three-dimensional mark point cloud data is subjected to the tooth surface fitting, tooth surface contour point cloud segmentation, expansion and other operations to complete the automatic measurement of the mark parameters, thereby eliminating the problems of unstable measurement results and low measurement efficiency caused by manual detection; at the same time, when collecting the first laser point cloud data and the first color image data, the working positions of the laser point cloud acquisition device and the color image acquisition device are consistent with the positions when constructing the two-dimensional and three-dimensional conversion matrices, thereby effectively ensuring the validity of the constructed two-dimensional and three-dimensional conversion matrices, and finally ensuring the accuracy of the color three-dimensional mark point cloud data converted by the two-dimensional and three-dimensional conversion matrices.

[0019] According to some embodiments of the present invention, the step of obtaining conversion basic data when the calibration block is in multiple placement states includes:

[0020] Placing the calibration block on a pure color background structure, and coating the surrounding side of the calibration block with the same color as the pure color background structure;

[0021] Each time the height of the pure color background structure and / or the placement angle of the calibration block is adjusted, a set of the conversion basic data is acquired through the laser point cloud acquisition device and the color image acquisition device until multiple sets of the conversion basic data are obtained.

[0022] According to some embodiments of the present invention, the calibration block is a rectangular parallelepiped calibration block.

[0023] According to some embodiments of the present invention, the three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each group of the conversion basic data are obtained by the following steps:

[0024] Performing preliminary filtering on the second laser point cloud data;

[0025] Performing plane fitting on the second laser point cloud data after preliminary filtering, and projecting the second laser point cloud data after preliminary filtering onto a fitting plane formed after fitting, to obtain projected two-dimensional point cloud data;

[0026] Performing threshold processing, morphological filtering processing, and canny edge detection processing on the projected two-dimensional point cloud data, and determining the two-dimensional projection coordinates of each corner point of the calibration block by using a straight line iterative fitting method to find the intersection point;

[0027] The three-dimensional corner point coordinates corresponding to the two-dimensional projection coordinates of each corner point of the calibration block are obtained by inverse transformation.

[0028] According to some embodiments of the present invention, the step of performing threshold processing, morphological filtering processing, and canny edge detection processing on the projected two-dimensional point cloud data, and determining the two-dimensional projection coordinates of each corner point of the calibration block by using a straight line iterative fitting method to find intersection points, includes:

[0029] Encoding the projected two-dimensional point cloud data according to a preset sampling interval, and obtaining the encoded origin after encoding;

[0030] Determine the encoding plane coordinates of each projection point in the projected two-dimensional point cloud data based on the encoding origin;

[0031] Determine an all-zero list based on the projection point in the projected two-dimensional point cloud data and the sampling interval, and set the position of the projection point in the all-zero list to 1;

[0032] Performing morphological filtering on the area corresponding to the all-zero list, and performing edge detection after filtering to obtain a first edge contour of the calibration block;

[0033] Determine four first preliminary corner points on the first edge contour, wherein the area of ​​a quadrilateral formed by the four first preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the first edge contour;

[0034] Iteratively fitting straight lines is performed on the four sides of the quadrilateral formed by the four first preliminary corner points, and in each iterative operation, the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded until the four sides of the quadrilateral formed by the four first preliminary corner points meet the calibration edge line requirements;

[0035] The intersection points of the four edges that meet the requirements of the calibration edge line are obtained to obtain the two-dimensional projection coordinates of each corner point of the calibration block.

[0036] According to some embodiments of the present invention, the two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data in each set of the conversion basic data are obtained by the following steps:

[0037] Performing threshold processing, morphological filtering processing, maximum connected domain extraction processing, and Canny edge detection processing on the second color image data to obtain a second edge contour of the calibration block;

[0038] Determine four second preliminary corner points on the second edge contour, wherein the area of ​​a quadrilateral formed by the four second preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the second edge contour;

[0039] Iteratively fitting straight lines is performed on the four sides of the quadrilateral formed by the four second preliminary corner points, and in each iterative operation, the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded until the four sides of the quadrilateral formed by the four second preliminary corner points meet the calibration edge line requirements;

[0040] The intersection points of the four edges that meet the requirements of the calibration edge line are obtained to obtain the two-dimensional corner point coordinates of each corner point of the calibration block.

[0041] According to some embodiments of the present invention, the imprint segmentation model is obtained by the following steps:

[0042] Acquire a plurality of annotated color images for training, annotate the plurality of annotated color images for training at the pixel level, and the plurality of annotated color images for training include contact marks of various types of spiral bevel gears, wherein pixels of the annotated color images for training are greater than or equal to preset separation measurement pixels, and the separation measurement pixels are determined according to the detection accuracy requirements of the spiral bevel gears to be detected;

[0043] Based on the plurality of labeled color images for training, the pre-trained weights are loaded and YOLOV8 model training based on transfer learning is performed to obtain the imprint segmentation model.

[0044] According to some embodiments of the present invention, the separation measurement pixels are greater than or equal to 2448×2048 pixels.

[0045] According to some embodiments of the present invention, mapping the two-dimensional print color image onto the first laser point cloud data and a pre-constructed two-dimensional and three-dimensional conversion matrix to obtain color three-dimensional print point cloud data comprises the following steps:

[0046] The two-dimensional print color image data is converted into a three-dimensional space by using the two-dimensional and three-dimensional conversion matrix to obtain a corresponding relationship between a single pixel in the two-dimensional print color image data and the first laser point cloud data in the three-dimensional space;

[0047] If a single pixel in the two-dimensional print color image data corresponds to multiple point clouds in the first laser point cloud data, the point cloud with the largest point in the Z direction among the multiple point clouds corresponding to the single pixel is retained to obtain the color three-dimensional print point cloud data.

[0048] According to some embodiments of the present invention, mapping the two-dimensional print color image onto the first laser point cloud data and a pre-constructed two-dimensional and three-dimensional conversion matrix to obtain color three-dimensional print point cloud data comprises the following steps:

[0049] The two-dimensional print color image data is converted into a three-dimensional space by using the two-dimensional and three-dimensional conversion matrix to obtain a corresponding relationship between a single pixel in the two-dimensional print color image data and the first laser point cloud data in the three-dimensional space;

[0050] If a single pixel in the two-dimensional print color image data corresponds to multiple point clouds in the first laser point cloud data, retain the point cloud with the largest Z direction among the multiple point clouds corresponding to the single pixel, and retain the point clouds whose average height exceeds the preset pixel allocation requirement based on the clustering method to obtain the colored three-dimensional print point cloud data.

[0051] According to some embodiments of the present invention, the step of unfolding the segmented tooth surface point cloud into a plane and completing the imprint measurement to obtain the imprint detection result comprises the following steps:

[0052] Generating a solid surface mesh according to the segmented tooth surface point cloud;

[0053] Executing a UV unfolding algorithm on the solid surface mesh image to unfold the solid surface mesh into a detection plane;

[0054] The mark detection result is obtained by measuring the minimum distance between the contact mark and the small end of the tooth surface, the minimum distance between the contact mark and the large end of the tooth surface, the minimum distance between the contact mark and the tooth top, and the contact mark width based on the detection plane.

[0055] According to some embodiments of the present invention, the laser point cloud acquisition device is disposed on the motion platform via a multi-directional moving device.

[0056] According to some embodiments of the present invention, the multi-directional moving device includes a first moving mechanism, a lifting structure, a second moving mechanism and a rotating and fixing mechanism arranged on the second moving mechanism, the first moving mechanism is slidably arranged on the motion platform, and the sliding direction of the first moving mechanism is consistent with the moving direction of the storage plate; the lifting structure is arranged on the first moving mechanism and can be lifted and lowered along the height direction of the first moving mechanism; the second moving mechanism is arranged on the lifting structure and can move perpendicular to the sliding direction of the first moving mechanism and the lifting direction of the lifting structure; the rotating and fixing mechanism is arranged on the second moving mechanism, the laser point cloud acquisition device is arranged on the rotating and fixing mechanism, and the laser point cloud acquisition device is rotated by the rotating and fixing mechanism to adjust the angle between the line of sight of the laser point cloud acquisition device and the storage plate.

[0057] According to some embodiments of the present invention, the color image acquisition device is disposed on an auxiliary working platform via a multi-directional rotating bracket, the auxiliary working platform is located on one side of the motion platform, and the color image acquisition device adjusts the angle between itself and the storage plate via the multi-directional rotating bracket.

[0058] According to some embodiments of the present invention, the spiral bevel gear contact mark detection platform further includes a lighting device.

[0059] According to some embodiments of the present invention, the lighting device uses a ring-shaped light source.

[0060] The aviation spiral bevel gear contact mark visual measurement device according to the second aspect of the present invention is applied to a spiral bevel gear contact mark detection platform, the spiral bevel gear contact mark detection platform includes a motion platform, a laser point cloud acquisition device, and a color image acquisition device, and a storage plate is provided on the motion platform;

[0061] The aviation spiral bevel gear contact mark visual measurement device comprises:

[0062] A point cloud data acquisition unit, used for controlling the movement of the placement plate and acquiring first laser point cloud data of the spiral bevel gear to be inspected placed on the placement plate acquired by the laser point cloud acquisition device;

[0063] an image data acquisition unit, configured to acquire first color image data of the spiral bevel gear to be detected taken by the color image acquisition device, wherein the relative position of the spiral bevel gear to be detected and the placement plate remains the same when acquiring the first color image data and when acquiring the first laser point cloud data;

[0064] An impression image acquisition unit, used for inputting the first color image data into a pre-built impression segmentation model to obtain two-dimensional impression color image data;

[0065] A three-dimensional conversion unit, used for mapping the two-dimensional print color image onto the first laser point cloud data according to the first laser point cloud data and a pre-constructed two-dimensional and three-dimensional conversion matrix, to obtain color three-dimensional print point cloud data;

[0066] A tooth surface point cloud determination unit, used for performing surface fitting on the color three-dimensional impression point cloud data, and combining the first laser point cloud data to determine that the point cloud on the fitting surface is the segmented tooth surface point cloud;

[0067] A measuring unit, used for unfolding the segmented tooth surface point cloud into a plane, completing the imprint measurement, and obtaining the imprint detection result;

[0068] The two-dimensional and three-dimensional transformation matrices are obtained by the following steps:

[0069] Acquire conversion basic data when the calibration block is in a plurality of placement states, each set of the basic conversion data includes the second color image data of the calibration block acquired by the color image acquisition device and the second laser point cloud data acquired by the laser point cloud acquisition device; and when acquiring each set of the conversion basic data, the working position of the laser point cloud acquisition device is the same as the working position of the laser point cloud acquisition device when acquiring the first laser point cloud data, and the working position of the color image acquisition device is the same as the working position of the color image acquisition device when acquiring the first color image data;

[0070] Determine the three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each group of the conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data, and obtain the coordinate correspondence between the plurality of the three-dimensional corner points and the plurality of the two-dimensional corner points;

[0071] Determine a plurality of sets of coordinate correspondences to obtain the two-dimensional and three-dimensional transformation matrices.

[0072] The contact mark visual measuring device for aviation spiral bevel gears according to the embodiment of the present invention has at least the following beneficial effects:

[0073] The spiral bevel gear contact mark detection platform can realize the automatic collection of the first laser point cloud data and the first color image data of the spiral bevel gear to be detected, and the mark segmentation model can be used to quickly complete the extraction of the two-dimensional mark color image data. After that, the pre-constructed two-dimensional and three-dimensional conversion matrix can be used to map the color information of the two-dimensional mark color image to the three-dimensional point cloud to obtain the color three-dimensional mark point cloud data. Finally, the color three-dimensional mark point cloud data is subjected to the tooth surface fitting, tooth surface contour point cloud segmentation, expansion and other operations to complete the automatic measurement of the mark parameters, thereby eliminating the problems of unstable measurement results and low measurement efficiency caused by manual detection; at the same time, when collecting the first laser point cloud data and the first color image data, the working positions of the laser point cloud acquisition device and the color image acquisition device are consistent with the positions when constructing the two-dimensional and three-dimensional conversion matrices, thereby effectively ensuring the validity of the constructed two-dimensional and three-dimensional conversion matrices, and finally ensuring the accuracy of the color three-dimensional mark point cloud data converted by the two-dimensional and three-dimensional conversion matrices.

[0074] The computer-readable storage medium according to the fourth aspect of the present invention stores computer-executable instructions, and the computer-executable instructions are used to execute the visual measurement method for the contact mark of the aviation spiral bevel gear as described in the first aspect of the present invention. Since the computer-readable storage medium adopts all the technical solutions of the visual measurement method for the contact mark of the aviation spiral bevel gear of the above embodiment, it at least has all the beneficial effects brought by the technical solutions of the above embodiment.

[0075] Other features and advantages of the present invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0077] Figure 1 A schematic structural diagram of a spiral bevel gear contact mark detection platform according to an embodiment of the present invention;

[0078] Figure 2 A flow chart of a method for visually measuring contact marks of aviation spiral bevel gears according to an embodiment of the present invention;

[0079] Figure 3 A flowchart of obtaining two-dimensional and three-dimensional conversion matrices according to an embodiment of the present invention;

[0080] Figure 4 A flowchart of obtaining conversion basic data according to an embodiment of the present invention;

[0081] Figure 5 A flowchart of obtaining the three-dimensional corner point coordinates of each corner point of a calibration block according to an embodiment of the present invention;

[0082] Figure 6 A flowchart of obtaining the two-dimensional projection coordinates of each corner point of the calibration block according to an embodiment of the present invention;

[0083] Figure 7 A flowchart of obtaining the two-dimensional corner point coordinates of each corner point of a calibration block according to an embodiment of the present invention;

[0084] Figure 8 A flowchart of obtaining a print segmentation model according to an embodiment of the present invention;

[0085] Fig. 9 A schematic diagram of segmenting a tooth surface point cloud according to an embodiment of the present invention;

[0086] Fig.10 A flowchart of obtaining color three-dimensional print point cloud data according to an embodiment of the present invention;

[0087] Fig.11 A flowchart of obtaining color three-dimensional print point cloud data according to another embodiment of the present invention;

[0088] Fig.12 A flowchart of obtaining an imprint detection result according to segmented tooth surface point cloud according to an embodiment of the present invention;

[0089] Fig.13 A schematic diagram of a solid surface mesh according to an embodiment of the present invention;

[0090] Fig.14 A schematic diagram of a detection plane according to an embodiment of the present invention;

[0091] Fig.15 FIG. 1 is a schematic diagram of ABCH measurement according to an embodiment of the present invention.

[0092] Figure Number:

[0093] Sports platform 100, storage flat plate 110,

[0094] Laser point cloud acquisition device 200, laser controller 210,

[0095] Color image acquisition device 300,

[0096] Spiral bevel gear 400 to be tested,

[0097] The first moving mechanism 510, the lifting structure 520, the second moving mechanism 530, and the rotating fixing mechanism 540

[0098] Auxiliary working platform 600, multi-directional rotating bracket 610,

[0099] Lighting device 700, lighting power supply 710,

[0100] Data processing unit 800. DETAILED DESCRIPTION

[0101] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0102] In the description of the present invention, if there is a description of first, second, etc., it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0103] In the description of the present invention, it should be understood that descriptions involving orientation, such as orientation or positional relationship indicated as up, down, etc., are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0104] In the description of the present invention, it should be noted that, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific content of the technical solution.

[0105] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described below are only some embodiments of the present invention, not all embodiments.

[0106] In order to better describe the visual measurement method for the contact mark of the aviation spiral bevel gear according to the embodiment of the present invention, the contact mark detection platform of the spiral bevel gear applied is briefly described here. Figure 1 The spiral bevel gear contact mark detection platform includes a motion platform 100, a laser point cloud acquisition device 200, a color image acquisition device 300, and a data processing unit 800. A placement plate 110 is provided on the motion platform 100, and the placement plate 110 can move along the Y direction.

[0107] The laser point cloud acquisition device 200 uses a 2D line laser sensor head. The trigger mode and exposure time can be adjusted through the laser controller 210 to realize the operation control of the 2D line laser sensor head. The laser point cloud acquisition device 200 can be set on the motion platform 100 through a multi-directional moving device. The multi-directional moving device includes a first moving mechanism 510, a lifting structure 520, a second moving mechanism 530 and a rotating and fixing mechanism 540 arranged on the second moving mechanism 530. The first moving mechanism 510 is slidably arranged on the motion platform 100, and the sliding direction of the first moving mechanism 510 is consistent with the moving direction of the storage plate 110; the lifting structure 520 is arranged on the first moving mechanism 510, and can be lifted and lowered along the height direction of the first moving mechanism 510; the second moving mechanism 530 is arranged on the lifting structure 520, and can move perpendicularly to the sliding direction of the first moving mechanism 510 and the lifting direction of the lifting structure 520; the rotating and fixing mechanism 540 is arranged on the second moving mechanism 530, and the laser point cloud acquisition device 200 is arranged on the rotating and fixing mechanism 540. The laser point cloud acquisition device 200 is rotated by the rotating and fixing mechanism 540 to adjust The angle between the line of sight of the entire laser point cloud acquisition device 200 and the storage plate 110 can ultimately be adjusted over a wide range of the shooting position of the laser point cloud acquisition device 200 through the first moving mechanism 510, the lifting structure 520, the second moving mechanism 530 and the rotating and fixing mechanism 540; wherein, the first moving mechanism 510 can directly adopt a gantry; the lifting structure 520 includes a cross bar that can be lifted and set on the gantry, and sliding sleeves are provided at both ends of the cross bar, so that Z-direction adjustment can be achieved, and a rack is provided on the cross bar along the X direction. The second moving mechanism 530 realizes X-direction movement by cooperating with the rack, and finally, the rotation angle of the laser point cloud acquisition device 200 can be adjusted by adjusting the rotating and fixing mechanism 540, wherein the rotating and fixing mechanism 540 can directly adopt a clamp to achieve the adjustment of the fixing and rotation angle of the laser point cloud acquisition device 200.

[0108] The color image acquisition device 300 is set on the auxiliary work platform 600 through a multi-directional rotating bracket 610. The auxiliary work platform 600 is located on one side of the motion platform 100. The color image acquisition device 300 adjusts the angle between it and the storage plate 110 through the multi-directional rotating bracket 610. The color image acquisition device 300 can use an RGB industrial camera.

[0109] Meanwhile, another multi-directional rotating bracket is also provided on the auxiliary work platform 600, and the multi-directional rotating bracket is used to fix the lighting device 700, and the lighting device 700 can be used to fill light in the shooting area to ensure the shooting effect, wherein the lighting device 700 is preferably a ring light source, and the lighting power supply 710 of the lighting device 700 is provided on the auxiliary work platform 600. The data processing unit 800 can be directly a personal computer.

[0110] The following describes a visual measurement method for the contact mark of an aviation spiral bevel gear according to an embodiment of the present invention based on the above-mentioned spiral bevel gear contact mark detection platform.

[0111] See also Figure 2 As shown, Figure 2 The flowchart of a method for visually measuring the contact marks of aviation spiral bevel gears provided by an embodiment of the present invention includes but is not limited to the following steps:

[0112] Control the placement plate 110 to move, and obtain the first laser point cloud data of the spiral bevel gear 400 to be inspected placed on the placement plate 110 collected by the laser point cloud collection device 200;

[0113] Acquire the first color image data of the spiral bevel gear 400 to be detected taken by the color image acquisition device 300, and when acquiring the first color image data and when acquiring the first laser point cloud data, the relative position of the spiral bevel gear 400 to be detected and the placement plate 110 remains the same;

[0114] Inputting the first color image data into a pre-constructed print segmentation model to obtain two-dimensional print color image data;

[0115] According to the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix, the two-dimensional print color image is mapped onto the first laser point cloud data to obtain the color three-dimensional print point cloud data;

[0116] Performing surface fitting on the color three-dimensional impression point cloud data, and combining it with the first laser point cloud data, determining that the point cloud on the fitting surface is the segmented tooth surface point cloud;

[0117] Expand the segmented tooth surface point cloud into a plane, complete the imprint measurement, and obtain the imprint detection result;

[0118] Among them, the two-dimensional and three-dimensional transformation matrices are obtained by the following steps:

[0119] Acquire conversion basic data when the calibration block is in various placement states, each set of basic conversion data includes the second color image data of the calibration block acquired by the color image acquisition device 300 and the second laser point cloud data of the calibration block acquired by the laser point cloud acquisition device 200; and when acquiring each set of conversion basic data, the working position of the laser point cloud acquisition device 200 is the same as the working position of the laser point cloud acquisition device 200 when acquiring the first laser point cloud data, and the working position of the color image acquisition device 300 is the same as the working position of the color image acquisition device 300 when acquiring the first color image data;

[0120] Determine the three-dimensional corner point coordinates of each corner point of the calibration block in the second laser point cloud data in each set of conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block in the second color image data, and obtain the coordinate correspondence between the multiple three-dimensional corner points and the multiple two-dimensional corner points;

[0121] The two-dimensional and three-dimensional transformation matrices are obtained according to the corresponding relationships between multiple sets of coordinates.

[0122] In order to better describe the visual measurement method for the contact mark of the aviation spiral bevel gear according to the embodiment of the present invention, the two-dimensional and three-dimensional transformation matrices to be applied in this embodiment are briefly described. Figure 3 , the construction of the two-dimensional and three-dimensional conversion matrix requires the use of a calibration block, and the calibration block can be selected as a rectangular block structure, which can effectively simplify the calculation process; when constructing the two-dimensional and three-dimensional conversion matrix, it is first necessary to collect the second color image data and the second laser point cloud data of the calibration block, and the two types of data are collected at the same time, which can better avoid the relative position change of the calibration block and the placement plate 110 when collecting the two types of data, and ensure the accuracy of the subsequent two-dimensional and three-dimensional coordinate correspondence relationship. It should be noted that the second color image data and the second laser point cloud data need to be collected in multiple groups, and when collecting each group of conversion basic data, the calibration block will be in a different position. In some embodiments, two heights are selected and three orientations are selected at each height, so that the calibration block can be in 6 position states, so that six groups of conversion basic data can be collected, which is sufficient to ensure the conversion accuracy requirements of the obtained two-dimensional and three-dimensional conversion matrix. After that, the three-dimensional corner point coordinates of the calibration block in the second laser point cloud data and the two-dimensional corner point coordinates of the calibration block in the second color image data can be obtained in each set of conversion basic data, thereby obtaining a set of coordinate correspondences between three-dimensional corner points and two-dimensional corner points, and multiple sets of conversion basic data can obtain multiple sets of coordinate correspondences, thereby solving the two-dimensional and three-dimensional conversion matrices. It should also be noted that after the two-dimensional and three-dimensional conversion matrices are obtained, if it is necessary to detect the spiral bevel gear 400 to be detected, when obtaining the first color image data and the first laser point cloud data of the spiral bevel gear 400 to be detected, it is necessary to keep the working positions of the laser point cloud acquisition device 200 and the color image acquisition device 300 consistent with those when obtaining the two-dimensional and three-dimensional conversion matrices, so as to avoid the two-dimensional and three-dimensional conversion matrices being unable to adaptively complete the conversion from two-dimensional to three-dimensional due to position changes.

[0123] Based on the foregoing, the following describes the automatic detection process of the contact mark of the non-contact aviation spiral bevel gear according to the embodiment of the present invention.

[0124] First, place the spiral bevel gear 400 to be detected on the storage plate 110. Different types of spiral bevel gears 400 to be detected can be placed at different angles to ensure that the laser point cloud acquisition device 200 and the color image acquisition device 300 can collect the completed image and point cloud data of the tooth surface. Adjust the triggering mode and working parameters of the laser point cloud acquisition device 200, and then control the movement of the storage plate 110 to complete the collection of the first laser point cloud data of the tooth surface of the spiral bevel gear 400 to be detected, and the color image acquisition device 300 can be started synchronously to complete the collection of the first color image data of the spiral bevel gear 400 to be detected. It should be noted that no matter whether it is collecting data of the spiral bevel gear 400 to be detected or collecting data of the calibration block, it is necessary to keep the field of view of the laser point cloud acquisition device 200 and the color image acquisition device 300 as close to overlap as possible.

[0125] Afterwards, the collected first color image data is input into the imprint segmentation model to obtain two-dimensional imprint color image data containing the imprint, and then the two-dimensional and three-dimensional conversion matrix is ​​used to map the two-dimensional imprint color image data to the three-dimensional space. It should be noted that because the two-dimensional imprint color image data is essentially composed of pixels, the conversion process is essentially to map pixels to the three-dimensional space, and the first laser point cloud data itself corresponds to the three-dimensional space, so that the color information in the two-dimensional imprint color image data can be assigned to the first laser point cloud data to obtain color three-dimensional imprint point cloud data, and then the spiral bevel gear tooth surface fitting can be performed based on the color three-dimensional imprint point cloud data to obtain the segmented tooth surface point cloud (such as Fig. 9 As shown); the segmented tooth surface point cloud itself is still in a three-dimensional state. In order to facilitate measurement, the segmented tooth surface point cloud is unfolded into a plane, and the measurement can be completed in the plane state according to the ABCH measurement standard (A: the minimum distance between the contact mark and the small end of the tooth surface, B: the minimum spatial distance between the contact mark and the large end of the tooth surface, C: the minimum distance between the contact mark and the tooth top, H: the width of the contact mark, and the mark measurement is completed to obtain the mark detection result).

[0126] The visual measurement method for contact marks of aviation spiral bevel gears in an embodiment of the present invention can realize automatic collection of the first laser point cloud data and the first color image data of the spiral bevel gear 400 to be detected through the spiral bevel gear contact mark detection platform, and can use the mark segmentation model to quickly complete the extraction of the two-dimensional mark color image data, and then the pre-constructed two-dimensional and three-dimensional conversion matrix can be used to map the color information of the two-dimensional mark color image to the three-dimensional point cloud to obtain the color three-dimensional mark point cloud data, and finally the automatic measurement of the mark parameters is completed by performing tooth surface fitting, tooth surface contour point cloud segmentation, expansion and other operations on the color three-dimensional mark point cloud data, thereby eliminating the problems of unstable measurement results and low measurement efficiency caused by manual detection; at the same time, when collecting the first laser point cloud data and the first color image data, the working positions of the laser point cloud acquisition device 200 and the color image acquisition device 300 are consistent with the positions when constructing the two-dimensional and three-dimensional conversion matrices, thereby effectively ensuring the validity of the constructed two-dimensional and three-dimensional conversion matrices, and finally ensuring the accuracy of the color three-dimensional mark point cloud data converted by the two-dimensional and three-dimensional conversion matrices.

[0127] like Figure 4 As shown, in some embodiments, obtaining conversion basic data when the calibration block is in multiple placement states includes:

[0128] The calibration block is placed on a solid color background structure, and the surrounding side of the calibration block is painted with the same color as the solid color background structure;

[0129] Each time the height of the pure color background structure and / or the placement angle of the calibration block is adjusted, a set of conversion basic data is acquired through the laser point cloud acquisition device 200 and the color image acquisition device 300 until multiple sets of conversion basic data are obtained.

[0130] A solid color background can better highlight the calibration block and reduce interference information. In this embodiment, the axial side of the calibration block is painted the same color as the solid color background structure, so that the image data and point cloud data of the top surface of the calibration block can be directly and effectively extracted, further reducing the presence of interference information and effectively reducing the amount of calculation. Subsequently, it is only necessary to collect a set of basic conversion data each time the position of the calibration block is adjusted. It should be noted that in order to ensure the validity of the Z direction data, there is a displacement in the Z direction when at least two sets of basic conversion data are collected.

[0131] In addition, when selecting the colors of the calibration block and the solid color background structure, the color difference will be highlighted as much as possible. For example, when the calibration block adopts a high-flatness, silver-white rectangular structure, the solid color background can be black. It should be noted that the solid color background structure can directly use solid color flannel, such as black flannel. When the height of the calibration block needs to be changed, it is only necessary to place a padding block on the placement plate 110, and then place a solid color flannel on the padding block, and then place the calibration block.

[0132] In some embodiments, the calibration block is a rectangular parallelepiped calibration block. The standard shape of the rectangular parallelepiped calibration block can facilitate the subsequent better construction of two-dimensional and three-dimensional conversion matrices.

[0133] like Figure 5 As shown, in some embodiments, the three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each set of conversion basic data are obtained by the following steps:

[0134] Performing preliminary filtering on the second laser point cloud data;

[0135] Performing plane fitting on the second laser point cloud data after preliminary filtering, and projecting the second laser point cloud data after preliminary filtering onto a fitting plane formed after fitting, to obtain projected two-dimensional point cloud data;

[0136] The projected two-dimensional point cloud data is processed by threshold, morphological filtering, and canny edge detection, and the two-dimensional projection coordinates of each corner point of the calibration block are determined by the method of iterative straight line fitting and intersection point calculation.

[0137] The inverse transformation method is used to obtain the three-dimensional corner point coordinates corresponding to the two-dimensional projection coordinates of each corner point of the calibration block.

[0138] The scanned second laser point cloud data will have a large number of adhesion and non-adhesion interference points. Therefore, it is necessary to first eliminate some interference points through preliminary filtering, and then perform plane fitting on the second laser point cloud data after the preliminary filtering, and project the second laser point cloud data after the preliminary filtering to the fitting plane formed after fitting to obtain the projected two-dimensional point cloud data, and encode the projected two-dimensional point cloud data into a processable two-dimensional image, so that the two-dimensional image processing algorithm can be used for further processing to further improve the accuracy. Specifically, image processing can be performed through threshold processing, morphological filtering processing, canny edge detection processing, etc., and then the two-dimensional projection coordinates of the corner points can be obtained based on the method of iterative fitting of straight lines to obtain the intersection points, and finally the three-dimensional intersection coordinates can be obtained through inverse transformation.

[0139] like Figure 6As shown, in some embodiments, threshold processing, morphological filtering processing, and canny edge detection processing are performed on the projected two-dimensional point cloud data, and the two-dimensional projection coordinates of each corner point of the calibration block are determined by using a straight line iterative fitting method to find the intersection point, including:

[0140] Encode the projected two-dimensional point cloud data according to a preset sampling interval, and obtain the encoded origin after encoding;

[0141] Determine the encoding plane coordinates of each projection point in the projected two-dimensional point cloud data based on the encoding origin;

[0142] Determine an all-zero list based on the projection points and sampling intervals in the projected two-dimensional point cloud data, and set the positions of the projection points in the all-zero list to 1;

[0143] Performing morphological filtering on the area corresponding to the all-zero list, and performing edge detection after filtering to obtain a first edge contour of the calibration block;

[0144] Determine four first preliminary corner points on the first edge contour, wherein the area of ​​a quadrilateral formed by the four first preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the first edge contour;

[0145] An iterative straight line fitting operation is performed on the four sides of the quadrilateral formed by the four first preliminary corner points, and the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded in each iterative operation until the four sides of the quadrilateral formed by the four first preliminary corner points meet the calibration edge line requirements;

[0146] Find the intersection points of the four edges that meet the calibration edge line requirements and obtain the two-dimensional projection coordinates of each corner point of the calibration block.

[0147] After obtaining the projected 2D point cloud data, a sampling interval will be set so that the projected 2D point cloud data can be re-encoded according to the sampling interval and the encoding origin can be determined. The sampling interval can be selected as 0.02mm to ensure the final accuracy requirement. After that, the coordinates of each point after the new projection are obtained by subtracting the encoding origin from the projection point coordinates and then dividing by the sampling interval. An all-zero list is defined with the minimum circumscribed rectangle size after the projection point is divided by the sampling interval, and the positions of the projection points in the all-zero list are set to 1. The rest are set to 0. On the basis of the foregoing, the updated all-zero list can be subjected to morphological filtering and edge detection to obtain the first edge contour of the calibration block, and four first preliminary corner points are determined on the first edge contour to obtain a quadrilateral, and then the four sides of the quadrilateral are iteratively fitted with a straight line, and the projection points that exceed the fitting accuracy requirement from the fitting straight line are discarded each time (the fitting accuracy requirement can be selected as 5%, or it can be appropriately adjusted according to actual needs, in essence to eliminate the influence of interference points), until a quadrilateral that meets the requirements is obtained, and then only the coordinates of the intersection of the quadrilateral need to be taken to determine the two-dimensional projection coordinates; it should be noted that in actual operation, whether the calibration edge line requirements are met can be determined directly by adjusting the number of iterations in the iterative fitting line operation, that is, it is only necessary to determine whether a specific number of iterations have been performed, without the need to make a separate judgment on the calibration edge line requirements.

[0148] like Figure 7 As shown, in some embodiments, the two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data in each set of conversion basic data are obtained by the following steps:

[0149] Performing threshold processing, morphological filtering processing, maximum connected domain extraction processing, and Canny edge detection processing on the second color image data to obtain a second edge contour of the calibration block;

[0150] Determine four second preliminary corner points on the second edge contour, wherein the area of ​​a quadrilateral formed by the four second preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the second edge contour;

[0151] An iterative straight line fitting operation is performed on the four sides of the quadrilateral formed by the four second preliminary corner points, and the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded in each iterative operation, until the four sides of the quadrilateral formed by the four second preliminary corner points meet the calibration edge line requirements;

[0152] Find the intersection points of the four edges that meet the calibration edge line requirements and obtain the two-dimensional corner point coordinates of each corner point of the calibration block.

[0153] By performing threshold processing, morphological filtering processing, maximum connected domain extraction processing, and Canny edge detection processing on the second color image data, the second edge contour of the calibration block can be obtained, and then four second preliminary corner points are determined on the second edge contour to obtain a quadrilateral. Then, an iterative straight line fitting operation is performed on the four sides of the quadrilateral, and the projection points that exceed the fitting accuracy requirement from the fitting straight line in each operation are discarded (the fitting accuracy requirement can be selected as 5%, and can also be adjusted appropriately according to actual needs) until a quadrilateral that meets the requirements is obtained, and then only the coordinates of the intersection of the quadrilateral need to be taken to determine the coordinates of the two-dimensional corner points; it should be noted that in actual operation, whether the calibration edge line requirements are met can be determined directly by adjusting the number of iterations in the iterative straight line fitting operation, that is, it is only necessary to determine whether a specific number of iterations have been performed, without the need to make a separate judgment on the calibration edge line requirements.

[0154] like Figure 8 As shown, in some embodiments, the imprint segmentation model is obtained by the following steps:

[0155] Acquire multiple training annotated color images, annotate the multiple training annotated color images at the pixel level, and the multiple training annotated color images include multiple types of spiral bevel gear contact marks, wherein the pixels of the training annotated color images are greater than or equal to preset separation measurement pixels, and the separation measurement pixels are determined according to the detection accuracy requirements of the spiral bevel gear 400 to be detected;

[0156] Based on multiple labeled color images for training, the pre-trained weights are loaded and the YOLOV8 model training based on transfer learning is performed to obtain the imprint segmentation model.

[0157] The footprint segmentation model is built based on the YOLOV8 model training of transfer learning. In order to ensure the detection accuracy requirements, the training labeled color images will use high-pixel images and will be labeled at the pixel level. In addition, a large number of selected training labeled color images will contain various types and various perspectives of spiral bevel gear contact footprints, so that the footprint segmentation model finally trained can have extremely high-precision separation capabilities.

[0158] In some embodiments, the separation measurement pixel is greater than or equal to 2448×2048 pixels. When the pixel is higher than this, the requirements for precision measurement in the current market environment can be met to a higher degree. It is understandable that as the precision requirements increase, the separation measurement pixel can be further increased.

[0159] like Fig.10 As shown, in some embodiments, according to the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix, mapping the two-dimensional print color image onto the first laser point cloud data to obtain the color three-dimensional print point cloud data includes the following steps:

[0160] The two-dimensional imprint color image data is converted into a three-dimensional space by using a two-dimensional and three-dimensional conversion matrix to obtain a corresponding relationship between a single pixel in the two-dimensional imprint color image data and the first laser point cloud data in the three-dimensional space;

[0161] If a single pixel in the two-dimensional print color image data corresponds to multiple point clouds in the first laser point cloud data, the point cloud with the largest point in the Z direction among the multiple point clouds corresponding to the single pixel is retained to obtain color three-dimensional print point cloud data.

[0162] It should be noted that when mapping the two-dimensional imprint color image data to the three-dimensional space, a single pixel may correspond to multiple point cloud data. At this time, in order to ensure the accuracy of detection, only the largest point cloud in the Z direction will be retained. That is, from a bird's-eye view, the color image acquisition device 300 usually captures the color of the outermost surface of the spiral bevel gear 400 to be inspected. Therefore, using the outermost point cloud data can make the conversion correspondence more accurate.

[0163] like Fig.11 As shown, in some embodiments, according to the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix, mapping the two-dimensional print color image onto the first laser point cloud data to obtain the color three-dimensional print point cloud data includes the following steps:

[0164] The two-dimensional imprint color image data is converted into a three-dimensional space by using a two-dimensional and three-dimensional conversion matrix to obtain a corresponding relationship between a single pixel in the two-dimensional imprint color image data and the first laser point cloud data in the three-dimensional space;

[0165] If a single pixel in the two-dimensional print color image data corresponds to multiple point clouds in the first laser point cloud data, retain the point cloud with the largest Z direction among the multiple point clouds corresponding to the single pixel, and retain the point clouds whose average height exceeds the preset pixel allocation requirements based on the clustering method to obtain color three-dimensional print point cloud data.

[0166] On the basis of retaining the largest point cloud in the Z direction, we can further consider that due to the large density of the point cloud, some points will be assigned to adjacent pixels, so we can use the clustering method to save the point cloud clusters with a larger average height.

[0167] like Fig.12 As shown, in some embodiments, the segmented tooth surface point cloud is unfolded into a plane, and the imprint measurement is completed to obtain the imprint detection result, including the following steps:

[0168] Generate a solid surface mesh based on the segmented tooth surface point cloud;

[0169] A UV unfolding algorithm is performed on the solid surface mesh map to unfold the solid surface mesh into a detection plane;

[0170] Based on the detection plane, the minimum distance between the contact mark and the small end of the tooth surface, the minimum distance between the contact mark and the large end of the tooth surface, the minimum distance between the contact mark and the tooth top, and the contact mark width are measured to obtain the mark detection result.

[0171] Point cloud data cannot be measured directly, so the tooth surface point cloud (such as Fig. 9 ) to fit the surface and generate a solid surface mesh (as shown Fig.13 As shown), and then unfold it through the UV unfolding algorithm, so that the detection plane can be obtained (as shown Fig.14 As shown in the figure, the contour lines of the tooth surface and the imprint are included on the inspection plane image. The subsequent inspection only needs to be completed based on the ABCH measurement standard (as shown in the figure). Fig.15 As shown in the figure, the grayscale part in the middle of the figure is the imprint), and the imprint detection result is obtained.

[0172] In some embodiments, the laser point cloud acquisition device 200 is disposed on the motion platform 100 through a multi-directional moving device. This part of the content has been explained above and will not be repeated here.

[0173] In some embodiments, the multi-directional moving device includes a first moving mechanism 510, a lifting structure 520, a second moving mechanism 530, and a rotating fixing mechanism 540 disposed on the second moving mechanism 530. The first moving mechanism 510 is slidably disposed on the motion platform 100, and the sliding direction of the first moving mechanism 510 is consistent with the moving direction of the storage plate 110; the lifting structure 520 is disposed on the first moving mechanism 510, and can be lifted and lowered along the height direction of the first moving mechanism 510; the second moving mechanism 530 is disposed on the lifting structure 520, and can move perpendicularly to the sliding direction of the first moving mechanism 510 and the lifting direction of the lifting structure 520; the rotating fixing mechanism 540 is disposed on the second moving mechanism 530, and the laser point cloud acquisition device 200 is disposed on the rotating fixing mechanism 540. The laser point cloud acquisition device 200 is rotated by the rotating fixing mechanism 540 to adjust the angle between the line of sight of the laser point cloud acquisition device 200 and the storage plate 110. This part of the content has been explained above and will not be repeated here.

[0174] In some embodiments, the color image acquisition device 300 is disposed on the auxiliary work platform 600 via a multi-directional rotating bracket 610, the auxiliary work platform 600 is located on one side of the motion platform 100, and the color image acquisition device 300 adjusts the angle between the color image acquisition device 300 and the placement plate 110 via the multi-directional rotating bracket 610. This part of the content has been explained above and will not be repeated here.

[0175] In some embodiments, the spiral bevel gear contact mark detection platform further includes an illumination device 700. This part of the content has been described above and will not be repeated here.

[0176] In some embodiments, the lighting device 700 uses a ring light source, which has been described above and will not be repeated here.

[0177] The embodiment of the present invention further provides a contact mark visual measuring device for aviation spiral bevel gears, the contact mark visual measuring device for aviation spiral bevel gears comprises a point cloud data acquisition unit, an image data acquisition unit, an impression image acquisition unit, a three-dimensional conversion unit, a tooth surface point cloud determination unit and a measurement unit;

[0178] A point cloud data acquisition unit, used to control the movement of the placement plate 110 and acquire the first laser point cloud data of the spiral bevel gear 400 to be inspected placed on the placement plate 110 acquired by the laser point cloud acquisition device 200;

[0179] An image data acquisition unit, used to acquire first color image data of the spiral bevel gear 400 to be detected photographed by the color image acquisition device 300, and when acquiring the first color image data and when acquiring the first laser point cloud data, the relative position of the spiral bevel gear 400 to be detected and the placement plate 110 remains the same;

[0180] The print image acquisition unit is used to input the first color image data into the pre-built print segmentation model to obtain two-dimensional print color image data;

[0181] A three-dimensional conversion unit, for mapping the two-dimensional impression color image onto the first laser point cloud data and a pre-constructed two-dimensional and three-dimensional conversion matrix to obtain color three-dimensional impression point cloud data;

[0182] A tooth surface point cloud determination unit is used to perform surface fitting on the color three-dimensional impression point cloud data, and to combine the first laser point cloud data to determine that the point cloud on the fitting surface is the segmented tooth surface point cloud;

[0183] The measuring unit is used to unfold the segmented tooth surface point cloud into a plane, complete the imprint measurement, and obtain the imprint detection result;

[0184] Among them, the two-dimensional and three-dimensional transformation matrices are obtained by the following steps:

[0185] Acquire conversion basic data when the calibration block is in various placement states, each set of basic conversion data includes the second color image data of the calibration block acquired by the color image acquisition device 300 and the second laser point cloud data acquired by the laser point cloud acquisition device 200; and when acquiring each set of conversion basic data, the working position of the laser point cloud acquisition device 200 is the same as the working position of the laser point cloud acquisition device 200 when acquiring the first laser point cloud data, and the working position of the color image acquisition device 300 is the same as the working position of the color image acquisition device 300 when acquiring the first color image data;

[0186] Determine the three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each set of conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data, and obtain the coordinate correspondence between the multiple three-dimensional corner points and the multiple two-dimensional corner points;

[0187] Determine the correspondence between multiple sets of coordinates to obtain two-dimensional and three-dimensional transformation matrices.

[0188] In order to better describe the visual measurement method of the contact mark of the aviation spiral bevel gear of the embodiment of the present invention, the two-dimensional and three-dimensional conversion matrix to be applied in this embodiment is briefly described first. The construction of the two-dimensional and three-dimensional conversion matrix requires the use of a calibration block, and the calibration block can select a rectangular block structure, which can effectively simplify the calculation process; when constructing the two-dimensional and three-dimensional conversion matrix, it is first necessary to collect the second color image data and the second laser point cloud data of the calibration block, and the two data are collected at the same time, so that the relative position change of the calibration block and the placement plate 110 when collecting the two data can be better avoided, and the accuracy of the corresponding relationship of the two-dimensional and three-dimensional coordinates constructed subsequently can be ensured. It should be noted that the second color image data and the second laser point cloud data need to be collected in multiple groups, and when collecting each group of conversion basic data, the calibration block will be in a different position. In some embodiments, two heights will be selected and three orientations will be selected at each height, so that the calibration block can be in 6 positions, so that six groups of conversion basic data can be collected, which is sufficient to ensure the conversion accuracy requirements of the obtained two-dimensional and three-dimensional conversion matrix. After that, the three-dimensional corner point coordinates of the calibration block in the second laser point cloud data and the two-dimensional corner point coordinates of the calibration block in the second color image data can be obtained in each set of conversion basic data, thereby obtaining a set of coordinate correspondences between three-dimensional corner points and two-dimensional corner points, and multiple sets of conversion basic data can obtain multiple sets of coordinate correspondences, thereby solving the two-dimensional and three-dimensional conversion matrices. It should also be noted that after the two-dimensional and three-dimensional conversion matrices are obtained, if it is necessary to detect the spiral bevel gear 400 to be detected, when obtaining the first color image data and the first laser point cloud data of the spiral bevel gear 400 to be detected, it is necessary to keep the working positions of the laser point cloud acquisition device 200 and the color image acquisition device 300 consistent with those when obtaining the two-dimensional and three-dimensional conversion matrices, so as to avoid the two-dimensional and three-dimensional conversion matrices being unable to adaptively complete the conversion from two-dimensional to three-dimensional due to position changes.

[0189] Based on the foregoing, the following describes the automatic detection process of the contact mark of the non-contact aviation spiral bevel gear according to the embodiment of the present invention.

[0190] First, place the spiral bevel gear 400 to be detected on the storage plate 110. Different types of spiral bevel gears 400 to be detected can be placed at different angles to ensure that the laser point cloud acquisition device 200 and the color image acquisition device 300 can collect the completed image and point cloud data of the tooth surface. Adjust the triggering mode and working parameters of the laser point cloud acquisition device 200, and then control the movement of the storage plate 110 to complete the collection of the first laser point cloud data of the tooth surface of the spiral bevel gear 400 to be detected, and the color image acquisition device 300 can be started synchronously to complete the collection of the first color image data of the spiral bevel gear 400 to be detected. It should be noted that no matter whether it is collecting data of the spiral bevel gear 400 to be detected or collecting data of the calibration block, it is necessary to keep the field of view of the laser point cloud acquisition device 200 and the color image acquisition device 300 as close to overlap as possible.

[0191] Afterwards, the collected first color image data is input into the print segmentation model to obtain two-dimensional print color image data containing the print, and then the two-dimensional print color image data is mapped to the three-dimensional space using the two-dimensional and three-dimensional conversion matrix. It should be noted that because the two-dimensional print color image data is essentially composed of pixels, the conversion process is essentially to map pixels to the three-dimensional space, and the first laser point cloud data itself corresponds to the three-dimensional space, so that the color information in the two-dimensional print color image data can be assigned to the first laser point cloud data to obtain Color three-dimensional imprint point cloud data, and then the spiral bevel gear tooth surface fitting can be performed based on the color three-dimensional imprint point cloud data to obtain the segmented tooth surface point cloud; the segmented tooth surface point cloud itself is still in a three-dimensional state. In order to facilitate measurement, the segmented tooth surface point cloud is unfolded into a plane, and the measurement can be completed in the plane state according to the ABCH measurement standard (A: the minimum distance between the contact mark and the small end of the tooth surface, B: the minimum spatial distance between the contact mark and the large end of the tooth surface, C: the minimum distance between the contact mark and the tooth top, H: the contact mark width, and the mark measurement is completed to obtain the mark detection result).

[0192] The visual measurement of the contact mark of the aviation spiral bevel gear in the embodiment of the present invention can realize the automatic collection of the first laser point cloud data and the first color image data of the spiral bevel gear to be detected through the spiral bevel gear contact mark detection platform, and can use the mark segmentation model to quickly complete the extraction of the two-dimensional mark color image data, and then the pre-constructed two-dimensional and three-dimensional conversion matrix can be used to map the color information of the two-dimensional mark color image to the three-dimensional point cloud to obtain the color three-dimensional mark point cloud data, and finally the automatic measurement of the mark parameters is completed by performing tooth surface fitting, tooth surface contour point cloud segmentation, expansion and other operations on the color three-dimensional mark point cloud data, thereby eliminating the problems of unstable measurement results and low measurement efficiency caused by manual detection; at the same time, when collecting the first laser point cloud data and the first color image data, the working positions of the laser point cloud acquisition device and the color image acquisition device are consistent with the positions when constructing the two-dimensional and three-dimensional conversion matrices, thereby effectively ensuring the validity of the constructed two-dimensional and three-dimensional conversion matrices, and finally ensuring the accuracy of the color three-dimensional mark point cloud data converted by the two-dimensional and three-dimensional conversion matrices.

[0193] In addition, an embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are executed by a processor or a control module, so that the above-mentioned processor can execute the visual measurement method of the contact mark of the aviation spiral bevel gear in the above-mentioned embodiment, for example, execute the method described above.

[0194] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed methods above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium or a non-transitory medium and a communication medium or a temporary medium. As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disk DVD or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0195] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A method for visually measuring the contact mark of an aviation spiral bevel gear, characterized in that: Applied to the spiral bevel gear contact mark detection platform, the spiral bevel gear contact mark detection platform includes a motion platform, a laser point cloud acquisition device, and a color image acquisition device, and the motion platform is provided with a storage plate; The contact mark visual measurement method of aviation spiral bevel gears comprises: Controlling the placement plate to move, and acquiring first laser point cloud data of the spiral bevel gear to be inspected placed on the placement plate collected by the laser point cloud acquisition device; Acquire first color image data of the spiral bevel gear to be detected photographed by the color image acquisition device, and when acquiring the first color image data and when acquiring the first laser point cloud data, the relative position of the spiral bevel gear to be detected and the placement plate remains the same; Inputting the first color image data into a pre-constructed print segmentation model to obtain two-dimensional print color image data; According to the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix, the two-dimensional print color image is mapped onto the first laser point cloud data to obtain color three-dimensional print point cloud data; Performing surface fitting on the color three-dimensional impression point cloud data, and combining it with the first laser point cloud data, determining that the point cloud on the fitting surface is the segmented tooth surface point cloud corresponding to the impression of the spiral bevel gear to be detected; Expanding the segmented tooth surface point cloud into a plane, completing the imprint measurement, and obtaining the imprint detection result; The two-dimensional and three-dimensional transformation matrices are obtained by the following steps: Acquire conversion basic data when the calibration block is in a plurality of placement states, each set of the conversion basic data includes the second color image data of the calibration block acquired by the color image acquisition device and the second laser point cloud data of the calibration block acquired by the laser point cloud acquisition device; and when acquiring each set of the conversion basic data, the working position of the laser point cloud acquisition device is the same as the working position of the laser point cloud acquisition device when acquiring the first laser point cloud data, and the working position of the color image acquisition device is the same as the working position of the color image acquisition device when acquiring the first color image data; Determine the three-dimensional corner point coordinates of each corner point of the calibration block in the second laser point cloud data in each group of the conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block in the second color image data, and obtain the coordinate correspondence between the multiple three-dimensional corner points and the multiple two-dimensional corner points; The two-dimensional and three-dimensional transformation matrices are obtained according to the multiple sets of coordinate correspondences.

2. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The step of obtaining conversion basic data when the calibration block is in various placement states includes: Placing the calibration block on a pure color background structure, and coating the surrounding side of the calibration block with the same color as the pure color background structure; Each time the height of the pure color background structure and / or the placement angle of the calibration block is adjusted, a set of the conversion basic data is acquired through the laser point cloud acquisition device and the color image acquisition device until multiple sets of the conversion basic data are obtained.

3. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each group of the conversion basic data are obtained by the following steps: Performing preliminary filtering on the second laser point cloud data; Performing plane fitting on the second laser point cloud data after preliminary filtering, and projecting the second laser point cloud data after preliminary filtering onto a fitting plane formed after fitting, to obtain projected two-dimensional point cloud data; Performing threshold processing, morphological filtering processing, and canny edge detection processing on the projected two-dimensional point cloud data, and determining the two-dimensional projection coordinates of each corner point of the calibration block by using a straight line iterative fitting method to find the intersection point; The three-dimensional corner point coordinates corresponding to the two-dimensional projection coordinates of each corner point of the calibration block are obtained by inverse transformation.

4. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 3, characterized in that: The method of performing threshold processing, morphological filtering processing, and canny edge detection processing on the projected two-dimensional point cloud data, and determining the two-dimensional projection coordinates of each corner point of the calibration block by using a straight line iterative fitting method to find the intersection point, includes: Encoding the projected two-dimensional point cloud data according to a preset sampling interval, and obtaining the encoded origin after encoding; Determine the encoding plane coordinates of each projection point in the projected two-dimensional point cloud data based on the encoding origin; Determine an all-zero list based on the projection point in the projected two-dimensional point cloud data and the sampling interval, and set the position of the projection point in the all-zero list to 1; Performing morphological filtering on the area corresponding to the all-zero list, and performing edge detection after filtering to obtain a first edge contour of the calibration block; Determine four first preliminary corner points on the first edge contour, wherein the area of ​​a quadrilateral formed by the four first preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the first edge contour; Iteratively fitting straight lines is performed on the four sides of the quadrilateral formed by the four first preliminary corner points, and in each iterative operation, the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded until the four sides of the quadrilateral formed by the four first preliminary corner points meet the calibration edge line requirements; The intersection points of the four edges that meet the requirements of the calibration edge line are obtained to obtain the two-dimensional projection coordinates of each corner point of the calibration block.

5. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data in each group of the conversion basic data are obtained by the following steps: Performing threshold processing, morphological filtering processing, maximum connected domain extraction processing, and Canny edge detection processing on the second color image data to obtain a second edge contour of the calibration block; Determine four second preliminary corner points on the second edge contour, wherein the area of ​​a quadrilateral formed by the four second preliminary corner points is the largest among the areas of quadrilaterals formed by any four corner points on the second edge contour; Iteratively fitting straight lines is performed on the four sides of the quadrilateral formed by the four second preliminary corner points, and in each iterative operation, the projection points whose distance from the fitting straight line exceeds the fitting accuracy requirement are discarded until the four sides of the quadrilateral formed by the four second preliminary corner points meet the calibration edge line requirements; The intersection points of the four edges that meet the requirements of the calibration edge line are obtained to obtain the two-dimensional corner point coordinates of each corner point of the calibration block.

6. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The imprint segmentation model is obtained by the following steps: Acquire a plurality of annotated color images for training, annotate the plurality of annotated color images for training at the pixel level, and the plurality of annotated color images for training include contact marks of various types of spiral bevel gears, wherein pixels of the annotated color images for training are greater than or equal to preset separation measurement pixels, and the separation measurement pixels are determined according to the detection accuracy requirements of the spiral bevel gears to be detected; Based on the plurality of labeled color images for training, the pre-trained weights are loaded and YOLOV8 model training based on transfer learning is performed to obtain the imprint segmentation model.

7. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The method of mapping the two-dimensional print color image onto the first laser point cloud data and the pre-constructed two-dimensional and three-dimensional conversion matrix to obtain the color three-dimensional print point cloud data comprises the following steps: The two-dimensional print color image data is converted into a three-dimensional space by using the two-dimensional and three-dimensional conversion matrix to obtain a corresponding relationship between a single pixel in the two-dimensional print color image data and the first laser point cloud data in the three-dimensional space; If a single pixel in the two-dimensional print color image data corresponds to multiple point clouds in the first laser point cloud data, retain the point cloud with the largest Z direction among the multiple point clouds corresponding to the single pixel, and retain the point clouds whose average height exceeds the preset pixel allocation requirement based on the clustering method to obtain the colored three-dimensional print point cloud data.

8. The method for visually measuring the contact mark of an aviation spiral bevel gear according to claim 1, characterized in that: The step of unfolding the segmented tooth surface point cloud into a plane and completing the imprint measurement to obtain the imprint detection result comprises the following steps: Generating a solid surface mesh according to the segmented tooth surface point cloud; Executing a UV unfolding algorithm on the solid surface mesh image to unfold the solid surface mesh into a detection plane; The mark detection result is obtained by measuring the minimum distance between the contact mark and the small end of the tooth surface, the minimum distance between the contact mark and the large end of the tooth surface, the minimum distance between the contact mark and the tooth top, and the contact mark width based on the detection plane.

9. A visual measurement device for contact marks of aviation spiral bevel gears, characterized in that: Applied to the spiral bevel gear contact mark detection platform, the spiral bevel gear contact mark detection platform includes a motion platform, a laser point cloud acquisition device, and a color image acquisition device, and the motion platform is provided with a storage plate; The aviation spiral bevel gear contact mark visual measurement device comprises: A point cloud data acquisition unit, used for controlling the movement of the placement plate and acquiring first laser point cloud data of the spiral bevel gear to be inspected placed on the placement plate acquired by the laser point cloud acquisition device; an image data acquisition unit, configured to acquire first color image data of the spiral bevel gear to be detected taken by the color image acquisition device, wherein the relative position of the spiral bevel gear to be detected and the placement plate remains the same when acquiring the first color image data and when acquiring the first laser point cloud data; An impression image acquisition unit, used for inputting the first color image data into a pre-built impression segmentation model to obtain two-dimensional impression color image data; A three-dimensional conversion unit, used for mapping the two-dimensional print color image onto the first laser point cloud data according to the first laser point cloud data and a pre-constructed two-dimensional and three-dimensional conversion matrix, to obtain color three-dimensional print point cloud data; A tooth surface point cloud determination unit, used for performing surface fitting on the color three-dimensional impression point cloud data, and combining the first laser point cloud data, determining that the point cloud on the fitting surface is the segmented tooth surface point cloud corresponding to the impression of the spiral bevel gear to be detected; A measuring unit, used for unfolding the segmented tooth surface point cloud into a plane, completing the imprint measurement, and obtaining the imprint detection result; The two-dimensional and three-dimensional transformation matrices are obtained by the following steps: Acquire conversion basic data when the calibration block is in a plurality of placement states, each set of the conversion basic data includes the second color image data of the calibration block acquired by the color image acquisition device and the second laser point cloud data acquired by the laser point cloud acquisition device; and when acquiring each set of the conversion basic data, the working position of the laser point cloud acquisition device is the same as the working position of the laser point cloud acquisition device when acquiring the first laser point cloud data, and the working position of the color image acquisition device is the same as the working position of the color image acquisition device when acquiring the first color image data; Determine the three-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second laser point cloud data in each group of the conversion basic data, and the two-dimensional corner point coordinates of each corner point of the calibration block corresponding to the second color image data, and obtain the coordinate correspondence between the plurality of the three-dimensional corner points and the plurality of the two-dimensional corner points; Determine a plurality of sets of coordinate correspondences to obtain the two-dimensional and three-dimensional transformation matrices.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the contact mark visual measurement method for aviation spiral bevel gears according to any one of claims 1 to 8.

Images