WO2023142608A1 - 获得飞机面型的系统和方法 - Google Patents

获得飞机面型的系统和方法 Download PDF

Info

Publication number
WO2023142608A1
WO2023142608A1 PCT/CN2022/131707 CN2022131707W WO2023142608A1 WO 2023142608 A1 WO2023142608 A1 WO 2023142608A1 CN 2022131707 W CN2022131707 W CN 2022131707W WO 2023142608 A1 WO2023142608 A1 WO 2023142608A1
Authority
WO
WIPO (PCT)
Prior art keywords
aircraft
point cloud
coordinate system
laser radar
point
Prior art date
Application number
PCT/CN2022/131707
Other languages
English (en)
French (fr)
Inventor
陈洁
宋袁曾
卢鹄
冯源
俞威
赵云龙
褚玉平
刘院君
Original Assignee
上海飞机制造有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 上海飞机制造有限公司 filed Critical 上海飞机制造有限公司
Publication of WO2023142608A1 publication Critical patent/WO2023142608A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
    • G01S17/88—Lidar systems specially adapted for specific applications
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Definitions

  • the present application relates to the field of digital measurement, for example, to a system and method for obtaining aircraft surface shape.
  • Lidar is a radar system that emits laser beams to detect characteristic quantities such as the position and speed of targets.
  • LiDAR has already played an indispensable role in ranging, scanning, autonomous driving, robotics and other fields.
  • spraying and maintenance of the outer surface of large aircraft it is necessary to perform high-precision surface measurement on the outer contour of the aircraft to lay the foundation for subsequent trajectory planning. Therefore, for this problem, a high-precision non-contact measurement tool is required, and lidar just meets this requirement.
  • the structure of the measured object is relatively simple, and the lidar is only used to scan a one-dimensional line or a two-dimensional plane; in addition, the magnitude of the measured object is small, That is, small size, small area, small volume, small footprint, and small scanning range of lidar.
  • the measured object such as an airplane is a large three-dimensional object, its structure and the shape of the outer contour are complex, and the outer surface is a complex and precise three-dimensional curved surface, which has a large size, large surface area, large volume, and takes up a lot of space. It needs to scan a large range of three-dimensional space. , so a technical solution to this problem is needed.
  • the present application proposes a method for obtaining the surface shape of an aircraft, which is simple, has few steps, and is easy to implement quickly, improves the measurement efficiency and ensures the detection accuracy at the same time.
  • the present application provides a method for obtaining the surface shape of an aircraft.
  • the method includes: establishing a three-dimensional mathematical model of the aircraft, and obtaining the three-dimensional coordinates of the three-dimensional mathematical model of the aircraft in the model coordinate system; Around the aircraft, and control each laser radar to scan the predetermined area of the aircraft to measure the point cloud of each predetermined area; splicing the point clouds of multiple predetermined areas measured by the multiple laser radars ; performing filtering processing on the spliced point cloud to remove noise points in the spliced point cloud; and comparing the filtered point cloud of the aircraft with the three-dimensional mathematical model of the aircraft, In the case that the error between the two exceeds a threshold, the three-dimensional mathematical model of the aircraft is reconstructed.
  • the process of arranging the plurality of laser radars around the aircraft includes: using a target to mark the position of each laser radar, and placing each laser radar The coordinates in the model coordinate system of the aircraft are obtained by converting the position relative to the target.
  • the process of arranging the multiple laser radars around the aircraft further includes: rotating the multiple laser radars multiple times based on the predetermined rotation step angle ⁇ and the rotation direction , so that each laser radar scans the aircraft once per rotation, and records the rotation angle ⁇ , the distance r and angle of each scanning point in the polar coordinate system of each laser radar itself And the reflection intensity p of each scanning point, so as to obtain the spherical coordinate A of each scanning point
  • the process of arranging the multiple laser radars around the aircraft further includes: establishing a Cartesian coordinate system at each laser radar, and placing each laser radar The measured spherical coordinates of each scan point are converted to Cartesian coordinates.
  • the process of stitching the point clouds of multiple predetermined areas measured by the multiple laser radars includes: every two adjacent points in the multiple predetermined areas Select at least three common points with high reflection intensity in the overlapping point cloud of the predetermined area; convert the Cartesian coordinates of the at least three common points with high reflection intensity under one laser radar to the Cartesian coordinates of another laser radar coordinate system, and obtain the coordinate transformation matrix between the Cartesian coordinate system of the one laser radar and the Cartesian coordinate system of the other laser radar; and obtain the Cartesian coordinate system and A coordinate transformation matrix between the model coordinate systems of the aircraft, so as to finally unify the coordinates of all points in the point cloud of the plurality of predetermined areas into the model coordinate system of the aircraft.
  • the process of performing filter processing on the spliced point cloud includes performing filter processing on the spliced point cloud by using straight-through filtering and statistical filtering.
  • the process of comparing the filtered point cloud of the aircraft with the three-dimensional mathematical model of the aircraft includes: calculating each The distance between a point and the corresponding point in the three-dimensional mathematical model of the aircraft, in the case that the distance is within the threshold, the model is constructed using the points in the three-dimensional mathematical model of the aircraft; If it is outside the threshold, use the points in the spliced point cloud after filtering to reconstruct the three-dimensional mathematical model of the aircraft.
  • the plurality of lidars are all single-line lidars.
  • the multiple laser radars include 4 laser radars respectively arranged at the nose of the aircraft, the wings on both sides, and the tail of the aircraft.
  • the predetermined rotation step angle ⁇ is 5°.
  • the present application also proposes an aircraft surface measurement system for performing the above method, the aircraft surface measurement system comprising: a plurality of laser radars arranged around the aircraft; for each laser radar, the aircraft surface measurement system It also includes: a support plate, the support plate is configured to be able to fix each laser radar; a rotating platform, the rotating platform is connected with the support plate and can drive the support plate to rotate; and, the suspension platform, Rails are installed on the suspension platform, so that the rotating platform can move up and down along the rails along the height direction of the suspension platform.
  • the present application also proposes a computer-readable storage medium, including a computer program for executing the above method.
  • Fig. 1 is a schematic diagram of a measurement system according to an alternative embodiment of the present application.
  • Fig. 2 is a schematic diagram of the layout of measuring the aircraft surface shape through the measuring system.
  • Fig. 3 is a schematic diagram of point cloud data obtained by laser radar through scanning measurement.
  • This application proposes a technical solution for large-scale three-dimensional measured objects, which can use laser radar to sample the surface of a large area of the measured object and quickly obtain three-dimensional space data of sampling points for high-precision reconstruction of surface contours And matching, so as to provide a good solution for large area spraying and maintenance of the surface.
  • the lidar When scanning the measured object, due to the outline and surrounding environment of the measured object and the technical limitations of the laser radar itself, it is difficult for a single laser radar to obtain all the point cloud data information of the measured object in one scan. Therefore, for objects with complex shapes such as airplanes, the lidar must scan and measure around the object to obtain all the point cloud data of the object. Multiple laser radars can be arranged around the measured object, so that these laser radars can scan all the point cloud data of the measured object, and then splicing the point cloud data obtained by different laser radars to obtain a complete image of the measured object surface point cloud.
  • the terms "point cloud” and "point cloud data” may mean the same meaning, which means a collection of data points generated when scanning an object under test, and may be used interchangeably.
  • the algorithms are relatively mature, including straight-through filtering, statistical filtering, conditional filtering, and radius filtering. Each algorithm has its own characteristics, and the processing effect is also better.
  • One is the three-dimensional reconstruction based on vision.
  • the relevant data images of the space measured objects are obtained through the depth camera, and then analyzed and processed.
  • the other is to scan and measure the surface of the measured object through the laser radar sensor to obtain a large amount of distance information, and then reconstruct the three-dimensional data of the measured object surface, Then the surface shape (or appearance) of the measured object is obtained.
  • contour refers to the overall shape of the measured object, while the surface shape is more detailed than the contour and reflects the measured object.
  • surface type can also be referred to as “topography”, and the two can be used interchangeably.
  • the object to be measured is described as an aircraft as an example to describe the method of this application, but it is not limited thereto.
  • the object to be measured can be other mechanical and electrical products or structures, as long as it needs to measure its surface shape to achieve various purposes, this method can be used.
  • the method of application and fall within the scope of protection of the present application.
  • the surface shape of its outer surface is obtained through actual scanning and measurement, its outer surface is reconstructed and reconstructed into a new mathematical model, and subsequent spraying and surface maintenance are carried out based on the reconstructed aircraft outer surface ( For example, the repair of the paint layer on the surface of the aircraft), through this digital reconstruction, robots can also be used for automatic spraying to achieve automation in this field.
  • the measurement system of the present application includes a plurality of measurement devices, and each measurement device includes a laser radar 1 , a support plate 2 , a rotating platform 3 and a suspension platform 4 .
  • Laser radar can be divided into single-line laser radar and multi-line laser radar according to the emitted wire harness.
  • Multi-line laser radar refers to laser rotating ranging radar that transmits and receives multiple laser beams at the same time.
  • Single-line lidar refers to the radar in which the line beam emitted by the laser source is a single beam of light. It has fast scanning speed, strong resolution, and high reliability. Compared with multi-line lidar, single-line lidar responds faster in terms of angular frequency and sensitivity. Therefore, it is more accurate in testing the distance and accuracy of surrounding objects.
  • the laser radar 1 is a single-line laser radar. As shown in Figure 3, when the laser radar 1 performs a scan, its transmitter will rotate at a uniform speed in a plane, and a laser will be emitted every time a small angle is rotated. The laser line beam is emitted to the surrounding 360-degree range in a circle. This plane is called the rotation plane, which is represented by the symbol 6 in FIG. 3 . It can be seen from Fig. 3 that in the Cartesian coordinate system xyz, the rotation plane 6 is a vertical plane along the z-axis.
  • the emitted laser beam intersects the measured object in the rotation plane 6, so that The laser beam forms a line on the surface of the measured object, and the coordinate position information of all points on this line (ie point cloud) is obtained. Therefore, one scan can only obtain the coordinate information of one line, which means that single-line lidar can only recognize one row or one column of point clouds, and can only describe two-dimensional line information, but cannot describe three-dimensional surfaces.
  • the laser radar 1 itself can only scan in the z-axis vertical rotation plane 6, so it can only scan the points on the vertical intersection line with the rotation plane on the aircraft surface cloud data.
  • the laser radar 1 needs to have degrees of freedom on the x-axis and the y-axis.
  • a rotating platform 3 is provided.
  • the rotating platform 3 can rotate in the xy plane, and the laser radar 1 is arranged On the rotating platform 3.
  • the rotating platform 3 can drive the lidar 1 to rotate around the z-axis.
  • the laser radar 1 when performing a scan, the laser radar 1 itself uses the polar coordinate system to record the information of each point in the point cloud data Among them, the pole or origin of the polar coordinate system coincides with the origin of the lidar, the polar axis coincides with the vertical z-axis, and r is the radius coordinate (also known as the polar radius), indicating the distance between the point scanned in the rotation plane and the pole , is the angular coordinate (also known as the polar angle), indicating the angle between the line connecting the point and the pole and the polar axis. Note that for each rotation angle ⁇ , LiDAR 1 itself only records the polar radius r and polar angle of the scanning point The information collection of the rotation angle ⁇ is carried out by the rotating platform 3 .
  • the laser radar 1 is connected to the support plate 2 by, for example, bolts, and the support plate 2 is connected to the rotating platform 3 by, for example, bolts.
  • the lidar 1 can be rotated under the drive of the rotating platform 3 to scan the surface of the aircraft at a fixed height.
  • a suspension platform 4 is provided, to which the rotating platform 3 is connected via a pivot shaft.
  • the hanging platform 4 is a telescopic structure, including an outer sleeve support and an inner sleeve support movably arranged in the outer sleeve support.
  • the rotating platform 3 is connected to the bottom of the inner sleeve support, and the inner sleeve support can move up and down according to It needs to protrude out of the outer sleeve support or retract into the outer sleeve support, so that the height of the inner sleeve support can be adjusted, so that the laser radar 1 can move up and down along the vertical height direction, and can rotate left and right.
  • a track is installed on the outer sleeve support, and the inner sleeve support can move up and down along the track in the outer sleeve support along the height direction of the hanging platform.
  • FIG 2 shows the arrangement of four measuring devices, that is, four laser radars.
  • one laser radar is arranged on the nose, the wings on both sides, and the tail of the aircraft.
  • These laser radars can be raised and lowered. and rotate to scan these areas.
  • Each lidar corresponds to a certain scanning area, and the combination of these areas covers the predetermined outer surface of the aircraft to be scanned. Therefore, the point cloud of the predetermined surface can be obtained by stitching and combining the point clouds obtained by scanning these areas.
  • the predetermined outer surface may refer to the entire outer surface of the aircraft, or to a part of the outer surface (such as the outer surface of the upper or lower half of the aircraft) or the outer surface of an aircraft section (such as the fuselage, tail or wing, etc.). surface.
  • the layout and quantity of lidars are optimized, and a good balance is achieved between cost and accuracy.
  • connection seam between two adjacent corresponding areas may be repeatedly scanned by two lidars, that is, the overlapping scanning area described below.
  • the figure shows the arrangement of 4 laser radars, but this is only an example, and more or fewer laser radars can be arranged at corresponding positions according to the area to be scanned, such as 2, 3, 5, 6 or More lidars.
  • the position calibration of the laser radars is performed through the target.
  • the world coordinate system is composed of three mutually perpendicular and intersecting coordinate axes x, y, and z. It is a fixed coordinate system and independent For all components (including lidar, aircraft, etc.). In actual measurement, the coordinates of the aircraft itself, the location where it is parked, and the location of the lidar arrangement will eventually be transformed into the world coordinate system to facilitate the unification of coordinates.
  • the position coordinates of each laser radar are determined through calibration, and these laser radars are unified into the same coordinate system, which facilitates the rapid realization of the subsequent whole process.
  • a square calibration plate ie, target
  • the coordinates of the center of the hollow circle on the calibration plate are the origin (0,0,0)
  • the edge detection of the calibration plate is performed by the lidar
  • the coordinates of the center of the circle are obtained by point cloud edge detection.
  • This coordinate is the coordinate of the lidar relative to the origin of the world coordinate system, so the position of the lidar in the world coordinate system can be obtained. Knowing the position coordinates of the lidar, the point cloud of the aircraft surface obtained by the subsequent measurement of the lidar itself can be converted into the world coordinate system.
  • the position coordinate z of the laser radar in the world coordinate system will change, and at this time, the target can be used every time it moves up and down. Recalibrate the position of the lidar, or measure and record the z coordinate increment of each movement, and add the increment to the z coordinate before the movement to obtain the new position coordinate z.
  • the three-dimensional mathematical model of the aircraft to be measured has been modeled in the computer and transformed into the world coordinate system, thereby knowing the position coordinates of multiple points on the outer surface of the aircraft.
  • the model coordinate system of the three-dimensional mathematical model of the aircraft can be directly replaced with the world coordinate system, so the unification of point cloud data into the world coordinate system described below can be understood as the mathematics of unifying these point cloud data into the aircraft In the model coordinate system, it is convenient for comparison and matching between the two.
  • the aircraft can be scanned.
  • the spherical coordinate system and the Cartesian three-dimensional coordinate system xyz are established at the origin of each lidar itself, that is, the polar coordinate system of the lidar itself, the spherical coordinate system and the three-dimensional Cartesian coordinate system are three These three coordinate systems are used to characterize the resulting point cloud as described below.
  • the lidar rotates in a predetermined rotation direction (for example, counterclockwise or clockwise around the z-axis) according to the set rotation step angle ⁇ , and performs a laser scan every time it rotates, and records relevant data information.
  • the rotation step angle ⁇ may be 5°, and the rotation is performed at a rotation speed of 5°/second, that is, the rotation is 5° per second, and then one scan is performed.
  • the rotation step angle can be other larger or smaller values, such as 1°, 2°, 3°, 4°, 6°, 7°, 8°, etc.
  • the rotation speed can be Larger or smaller values are also possible, such as 1°/sec, 2°/sec, 3°/sec, 4°/sec, 6°/sec, 7°/sec, 8°/sec, etc.
  • 5 degrees is used as the step angle, which not only realizes the measurement speed of fast scanning, but also considers the measurement accuracy.
  • the rotating platform 3 drives the laser radar 1 to achieve step-by-step rotation, and records the rotation angle ⁇ at the scanning position after the rotation is in place.
  • the lidar obtains a point cloud of a line at each scan, and uses its own polar coordinate system to record the distance r and angle of each point of the line i.e. get the polar coordinates The polar coordinates of all points on the scan line in the plane of rotation are thus determined.
  • the lidar Since the lidar has a rotation plane at each rotation angle ⁇ , and emits laser light in each rotation plane, scanning the outer surface of the aircraft, the intersection of the laser and the outer surface of the aircraft is a scanning line, and the information of this line is obtained by measurement , and so on, through multiple successive rotations, the data information of corresponding multiple scanning lines is obtained, and these scanning lines and data are superimposed and combined to reflect the surface information of the outer surface of the aircraft as a whole.
  • the lidar also collects and records the reflection intensity p of each point, and the intensity p reflects the lightness and darkness of the collected points. For example, if the object scanned by the laser is similar to a mirror, the reflection intensity of the laser is relatively large. Part of it will be absorbed, the reflected light will be weaker, the reflection intensity will be lower, and the value of reflection intensity will be smaller.
  • the reflection intensity is recorded for use in point cloud stitching described below. In the following point cloud stitching, the target point with high reflection breadth is the point with high reflection intensity.
  • each point can be represented by the established spherical coordinate system, namely As shown in Figure 3, a Cartesian three-dimensional coordinate system is also established on each lidar.
  • a Cartesian three-dimensional coordinate system is also established on each lidar.
  • each lidar-acquired point cloud is represented as a corresponding Cartesian coordinate.
  • the initial coordinate transformation of point cloud data is realized.
  • these Cartesian coordinates are relative to each laser radar's own Cartesian coordinate system.
  • Multiple laser radars have their own coordinate origins, and there are still differences in coordinate information between the data they collect. Since the point cloud data measured by multiple lidars are not expressed in the same coordinate system, these point clouds cannot be directly stitched together, and the coordinates of all points need to be unified in the same coordinate system.
  • all point cloud data measured by all lidars can be unified into the world coordinate system.
  • it can be realized by public point splicing through high inverse breadth target points.
  • Each laser radar scans 360 degrees in the scanning plane, and there will be an overlapping area of scanning between every two adjacent laser radars. If it is only scanned once, the overlapping area is an overlapping line, but after turning the platform When the rotation scans multiple times, there will be overlapping of multiple scan lines. Look for or identify 3 and/or more than 3 high reflection breadth target points whose reflection intensity is significantly stronger than other measurement points in the scanning overlap area of each two adjacent laser radars.
  • the high reflection breadth target point is the above-mentioned
  • the above-mentioned points with high reflection intensity these points have large reflection intensity values and obvious characteristics, and are easy to pick out from the point cloud, otherwise it will be difficult to identify those points that overlap between the two point clouds, and then the point cloud cannot be spliced .
  • These high inverse breadth target points are the common points in the overlapping area scanned by two adjacent lidars, and these common points have two sets of spatial position coordinates in the coordinate system of the two lidars and the measured point clouds respectively. Transforming one set of spatial position coordinates into the coordinate system of another set of spatial position coordinates through coordinate transformation realizes the unification of coordinate systems.
  • Identify 3 or more high inverse breadth target points (that is, common points) in the point cloud data in the overlapping area because for a single common point, the two point clouds can rotate arbitrarily around the single point in space, and cannot be uniquely determined The actual relative position between the two point clouds.
  • the two point clouds can be rotated arbitrarily around the straight line formed by the connection of these two points, but the positional relationship between the two cannot be uniquely determined.
  • these points can uniquely determine a plane, and the relative positions of the two point clouds connected by the common plane can be uniquely determined.
  • ⁇ is the scale parameter
  • R( ⁇ , ⁇ , ⁇ ) is the rotation matrix
  • T is the translation matrix
  • ⁇ , ⁇ , ⁇ are the rotation parameters of the three coordinate axes
  • x 0 , y 0 , z 0 are the translation parameters.
  • every two adjacent laser radars perform a coordinate transformation between each other, and finally unify into the Cartesian coordinate system of the same laser radar, for example, the second laser radar, the third laser radar and the fourth laser radar
  • the coordinates of the point cloud measured by the lidar are unified to the Cartesian coordinates of the first lidar, and the first lidar itself has been calibrated relative to the world coordinate system and has a definite alignment relationship (that is, the coordinate transformation matrix ), and finally all lidar and measured point cloud data can be unified into the world coordinate system.
  • the coordinate transformation matrix between these lidars and the world coordinate system has been calculated, so for the points measured by each lidar
  • the cloud can also directly use these coordinate transformation matrices to directly transform the point cloud from the Cartesian coordinate system of each lidar to the world coordinate system.
  • the secondary coordinate transformation of the point cloud data is realized, and finally unified into the model coordinate system of the aircraft.
  • the function of splicing is to combine the limited point clouds measured by each lidar to obtain the point cloud of the predetermined outer surface of the aircraft.
  • the spliced point cloud needs to be filtered, and the filtering is to remove points that are obviously not point clouds on the surface of the aircraft.
  • a lidar will record 360-degree point cloud information during a scan, but some points in these point clouds are not points on the surface of the aircraft, but may be points on other objects in the environment.
  • the role of filtering is to remove points that are not on the surface of the aircraft (these points can be called noise points or noise points in engineering).
  • Straight-through filtering and statistical filtering are two different filtering methods.
  • Straight-through filtering is to remove point clouds outside the range of a specific channel.
  • “Through-through” means setting a channel based on the point cloud space coordinate system, and the Points outside the channel are filtered out, thereby retaining the point cloud inside the channel.
  • This "channel” is represented in the Point Cloud Library (PCL) as a limited range set for one coordinate axis. For example, the point cloud on the surface of an aircraft has ranges on multiple coordinate axes. Therefore, outside the range The points obviously do not belong to the plane surface and can be directly removed by through filtering.
  • PCL Point Cloud Library
  • outliers are usually far away from other point cloud data. They will make the distribution of the point cloud uneven, affect the smoothness of the point cloud, and reduce the point cloud data. quality, which in turn affects the accuracy of the subsequent 3D reconstruction model. However, these outliers may appear within the channel range of the straight-through filter, so the straight-through filter has no effect on these points, so statistical filtering is also required.
  • noise points in the point cloud are removed by using a commonly used specific filtering method, which has strong pertinence and good filtering effect.
  • these point cloud data can be used to remodel the surface of the aircraft to obtain the surface shape of the aircraft.
  • these point cloud data is combined with the existing 3D model to reconstruct the aircraft surface, which can save time, reduce workload and improve efficiency.
  • the 3D mathematical model of the previously modeled aircraft already exists in the computer, and the mathematical model is located in the same world coordinate system as the scanned aircraft surface, so the two can interact directly, for example, for comparison and matching.
  • the filtered point cloud of the aircraft surface is matched with the aforementioned three-dimensional model of the aircraft, so as to reconstruct the aircraft surface, obtain the actual coordinate data and information of the aircraft surface, and obtain the real surface shape of the aircraft surface.
  • Matching refers to, for example, analyzing, calculating, and comparing the degree of identity or equivalence and difference between the two, and using the two as equal or replacing them according to the degree of similarity and difference. For example, when the difference between the two is small, the coordinate data of the existing 3D model can be used directly; and when the difference is large, the theoretical 3D model is replaced with the actually measured point cloud data.
  • the error between the corresponding points is calculated, that is, the model reconstruction error is calculated.
  • the distance between corresponding points can be calculated, that is, the coordinate difference.
  • the distance between points is used as the error measure, and the calculation is simple and easy to implement.
  • the reconstruction of the model will be explained by taking the distance between the calculation points as the error as an example.
  • the distance between them is calculated as described above, and the calculation result is compared with a predetermined error tolerance threshold (ie, distance threshold).
  • distance threshold ie, distance threshold
  • the distance between them is within the distance threshold, that is, when the error is small, the points of the 3D model are still used to build the model; and when the calculated distance is outside the threshold, that is, when the error is relatively large, the points of the measured point cloud are used instead of 3D model points to reconstruct the model.
  • the two extreme cases are: (1) When the error between each point in the measured point cloud and the corresponding point in the 3D model is within the threshold, the 3D model can be used directly without rebuilding the model ; (2) When the error between each point in the measured point cloud and the corresponding point in the 3D model is not within the threshold, the mathematical model needs to be discarded, and all points in the point cloud are used to reconstruct the model.
  • the measured point cloud By combining the measured point cloud with the original 3D model, the measured data is used to ensure the accuracy and precision of the obtained aircraft surface shape, and the data information of the original theoretical 3D model is selectively used to improve efficiency.
  • the Poisson algorithm is used to reconstruct the surface of the point cloud data above the error threshold. Input the point cloud data beyond the error threshold into the Poisson algorithm; set the appropriate fineness parameter, the parameter is degree[1,5], the larger the value, the finer it is, and the longer it takes; the setting falls into an octree node
  • the method of the present application is simple, has few steps, and is easy to realize quickly.
  • the existing theoretical model and actual measurement data are used for error analysis, which improves the measurement efficiency and ensures the detection accuracy.
  • This application is aimed at surface reconstruction, spraying and maintenance of large aircraft, and proposes a system and method for obtaining aircraft surface shape, which can achieve high-precision aircraft surface measurement; this application splices, Filtering and reconstruction processing can be integrated into a set of software, which can realize the quick conversion from point cloud data acquisition to model acquisition, and at the same time, model error analysis is carried out at the end, which improves the measurement efficiency and ensures the detection accuracy.

Abstract

本申请公开了一种获得飞机面型的方法，所述方法包括：建立飞机的三维数学模型，并获取所述飞机的三维数学模型在模型坐标系中的三维坐标；将多台激光雷达布置在所述飞机的四周，使并控制每台激光雷达扫描所述飞机的预定区域以测量得到每个预定区域的点云；将通过所述多台激光雷达测量得到的多个预定区域的点云进行拼接；对拼接好的点云进行滤波处理，以除去所述拼接好的点云中的噪点；以及将所述飞机的经过所述滤波处理后的的点云和所述飞机的三维数学模型进行对比，在二者之间的误差超过阈值的情况下，重建所述飞机三维数学模型。

Description

获得飞机面型的系统和方法
本申请要求在2022年01月26日提交中国专利局、申请号为202210091050.X的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本申请涉及数字化测量领域，例如涉及一种获得飞机面型的系统和方法。
背景技术
激光雷达是发射激光束探测目标的位置、速度等特征量的雷达系统。激光雷达已经在测距、扫描、自动驾驶、机器人等领域都发挥着不可或缺的作用。在大型飞机外表面重建、喷涂以及维修中，需要对飞机的外轮廓进行高精度的曲面测量，为后续轨迹规划奠定基础。因此，针对此问题，需要高精度的非接触测量工具，而激光雷达正好满足这一要求。
然而，在应用中，例如自动驾驶、机器人等领域，被测对象结构比较简单，激光雷达仅用于对一维的线或二维的平面进行扫描；此外，被测对象的量级较小，即尺寸小、面积小、体积小、占用空间小，激光雷达扫描的范围小。而飞机等被测对象为大型三维物体，其结构以及外轮廓的形状复杂，外表面为复杂精密的三维曲面，其尺寸大、表面积大、体积大、占用空间大，需要扫描大范围的三维空间，因此需要解决此问题的技术方案。
发明内容
本申请提出了一种获得飞机面型的方法，方法简单、步骤少、易于快速实现，提高了测量效率的同时也保证了检测精度。
本申请提供了一种获得飞机面型的方法，所述方法包括：建立飞机的三维数学模型，并获取所述飞机的三维数学模型在模型坐标系中的三维坐标；将多台激光雷达布置在所述飞机的四周，并控制每台激光雷达扫描所述飞机的预定区域以测量得到每个预定区域的点云；将通过所述多台激光雷达测量得到的多个预定区域的点云进行拼接；对拼接好的点云进行滤波处理，以除去所述拼接好的点云中的噪点；以及将所述飞机的经过所述滤波处理后的点云和所述飞机的三维数学模型进行对比，在二者之间的误差超过阈值的情况下，重建所述飞机的三维数学模型。
根据本申请的一种实施方式，所述将所述多台激光雷达布置在所述飞机的 四周的过程包括：利用标靶对每台激光雷达的位置进行标定，并将所述每台激光雷达相对于所述标靶的位置换算得到在所述飞机的模型坐标系下的坐标。
根据本申请的一种实施方式，所述将所述多台激光雷达布置在所述飞机的四周的过程还包括：基于预定的旋转步进角度Δθ和旋转方向多次旋转所述多台激光雷达，使得每旋转一次，每台激光雷达对所述飞机执行一次扫描，并记录旋转角度θ、每个扫描点在所述每台激光雷达自身的极坐标系下的距离r和角度
Figure PCTCN2022131707-appb-000001
以及所述每个扫描点的反射强度p，从而得到所述每个扫描点的球坐标A
Figure PCTCN2022131707-appb-000002
根据本申请的一种实施方式，所述将所述多台激光雷达布置在所述飞机的四周的过程还包括：在每台激光雷达处建立笛卡尔坐标系，并将所述每台激光雷达测得的每个扫描点的球坐标转换为笛卡尔坐标。
根据本申请的一种实施方式，所述将通过所述多台激光雷达测量得到的多个预定区域的点云进行拼接的过程包括：在所述多个预定区域中的相邻的每两个预定区域的重叠点云中选取至少三个反射强度高的公共点；将所述至少三个反射强度高的公共点在一台激光雷达下的笛卡尔坐标转换到另一台激光雷达的笛卡尔坐标系下，并获得所述一台激光雷达的笛卡尔坐标系和所述另一台激光雷达的笛卡尔坐标系之间的坐标转换矩阵；以及求取每台激光雷达的笛卡尔坐标系与所述飞机的模型坐标系之间的坐标转换矩阵，从而最终将所述多个预定区域的点云中的所有点的坐标统一到所述飞机的模型坐标系下。
根据本申请的一种实施方式，所述拼接好的点云进行滤波处理的过程包括利用直通滤波与统计滤波对拼接的点云进行滤波处理。
根据本申请的一种实施方式，所述将所述飞机的经过所述滤波处理后的点云和所述飞机的三维数学模型进行对比的过程包括：计算所述滤波处理后的点云中每个点与所述飞机的三维数学模型中对应点之间的距离，在所述距离在阈值以内的情况下，利用所述飞机的三维数学模型中的点构建模型；在所述距离在所述阈值以外的情况下，利用所述滤波处理后的拼接好的点云中的点重建飞机的三维数学模型。
根据本申请的一种实施方式，所述多台激光雷达均为单线激光雷达。
根据本申请的一种实施方式，所述多台激光雷达包括分别布置在所述飞机的机头、两侧机翼以及机尾处的4台激光雷达。
根据本申请的一种实施方式，所述预定的旋转步进角度Δθ为5°。本申请还提出一种用于执行上述方法的飞机面型测量系统，所述飞机面型测量系统包括：布置在飞机四周的多台激光雷达；对于每台激光雷达，所述飞机面型测量 系统还包括：支撑板，所述支撑板被构造成能够固定所述每台激光雷达；转动平台，所述转动平台与所述支撑板连接并能够带动所述支撑板旋转；以及，吊挂平台，所述吊挂平台安装有轨道，使得所述转动平台能够沿着所述轨道沿所述吊挂平台的高度方向上下移动。
本申请还提出一种计算机可读存储介质，包括用于执行上述方法的计算机程序。
附图说明
图1为根据本申请的可选实施方式的测量系统的示意图。
图2为通过测量系统测量飞机面型的布局示意图。
图3为激光雷达通过扫描测量得到点云数据的示意图。
附图标记：1-激光雷达；2-支撑板；3-转动平台；4-吊挂平台；5-原点；6-旋转平面；7-点云。
具体实施方式
下面结合说明书附图对本申请的可选实施例进行描述，以下的描述为示例性的，并非对本申请的限制，任何的其他类似情形也都落入本申请的保护范围之中。
本申请提出一种技术方案，针对大型三维被测对象，能够利用激光雷达对被测对象的大范围区域的表面进行采样并快速获取采样点的三维空间数据，以进行表面轮廓的高精度重构和匹配，从而为表面的大面积喷涂以及维修提供良好的解决方案。
在扫描被测对象时，由于被测对象的轮廓和周边环境以及激光雷达本身的技术限制等问题，单个激光雷达难以通过一次扫描就取得被测对象的全部点云数据信息。所以对于飞机等形状复杂的被测对象来说，激光雷达必须环绕被测对象的一周扫描测量，才能获得被测对象的全部点云数据。可以采用多台激光雷达环绕被测对象一周布置，如此这些激光雷达能够扫描得到被测对象的全部点云数据，然后通过对不同激光雷达获得的点云数据进行拼接处理以获得被测对象的完整表面点云。在本文中，术语“点云”和“点云数据”可以表示相同的含义，意指在扫描被测对象时所生成的数据点的集合，并且可以互换使用。
针对点云的滤波处理，其算法已较为成熟，包括直通滤波、统计滤波、条件滤波、半径滤波等。每种算法都有其有特点，处理效果也都较好。
针对于点云的重建处理，其研究主要分为两大方向，一类是基于视觉的三 维重建，通过深度照相机获取空间被测对象的相关数据图像，然后对其进行分析处理，根据计算机视觉理论及图像处理技术解算出现实环境中被测对象的三维信息；另一类是通过激光雷达传感器对被测对象表面进行扫描、测量得到大量距离信息，再对被测对象表面的三维数据进行重建，进而得到被测对象的面型(或形貌)。
术语“面型”的含义类似于被测对象的轮廓或外轮廓，但是又有所区别，区别之处在于轮廓是指被测对象的整体外形，而面型比轮廓更加精细、反映被测对象的区段或表面的实际的精确尺寸(包括局部的形状细节)。此外，在本文中，术语“面型”也可以称为“形貌”，二者可以互换使用。
在本文中，被测对象以飞机为例描述本申请的方法，但不限于此，被测对象可以是其他机电产品或结构，只要需要测量其面型以实现多种目的的物体均可以使用本申请的方法，并且落入本申请的保护范围内。对于飞机这种大型机械产品，通过实际扫描测量得到其外表面的面型，将其外表面重建和重构成新的数学模型，并基于重构后的飞机外表面进行后续的喷涂和表面维修(例如进行飞机表面漆层的修复)，通过这种数字化重构，也可以利用机器人进行自动化喷涂，实现该领域的自动化。
如图1和图2所示，本申请的测量系统包括多个测量装置，每个测量装置包括激光雷达1、支撑板2、转动平台3以及吊挂平台4。激光雷达按照发射的线束可分为单线激光雷达以及多线激光雷达，多线激光雷达是指同时发射及接收多束激光的激光旋转测距雷达。单线激光雷达是指激光源发出的线束是单束光线的雷达，其扫描速度快、分辨率强、可靠性高，相比多线激光雷达，单线激光雷达在角频率及灵敏度上反应更快捷，所以，在测试周围物体的距离和精度上都更加精准。
激光雷达1为单线激光雷达，如图3所示，激光雷达1在进行一次扫描时，其发射器会在一平面内进行匀速的旋转，每旋转一个小角度即发射一次激光，轮巡旋转一圈而向周围360度的范围发射激光线束，这个平面叫旋转平面，图3中由标号6表示。从图3中可以看出，在笛卡尔坐标系xyz中，旋转平面6为沿z轴的竖向平面，在进行一次扫描时，发射的激光束在旋转平面6内与被测对象相交，使激光束在被测对象表面形成一条线，获取这条线上所有点的坐标位置信息(即点云)。因此，一次扫描只能获得一条线的坐标信息，这意味着单线激光雷达只能识别一排或一列点云，只能描述二维线状信息，无法描述三维表面。
为此，结合图1至图3，可以看出，激光雷达1本身只能在z轴竖向旋转平面6内扫描，因此只能扫描得到飞机表面上与该旋转平面的竖向相交线的点云 数据。为了得到飞机的三维表面的点云，激光雷达1需要在在x轴和y轴上具有自由度，为此，设置了转动平台3，转动平台3可以在xy平面内旋转，将激光雷达1布置在转动平台3上。转动平台3可以带动激光雷达1绕z轴旋转，每旋转到一个角度θ，执行一次扫描，得到飞机表面上的一条线的点云数据，多个旋转角度产生多条线的点云，这些线叠加形成飞机表面的点云，实现从二维线到三维面的过渡。
如图3所示，在执行一次扫描时，激光雷达1自身利用极坐标系记录点云数据中每个点的信息
Figure PCTCN2022131707-appb-000003
其中，极坐标系的极点或原点与激光雷达的原点重合，极轴与竖向z轴重合，r为半径坐标(又称为极径)，表示在旋转平面内扫描得到的点距极点的距离，
Figure PCTCN2022131707-appb-000004
为角坐标(又称为极角)，表示该点与极点之间的连线和极轴之间的角度。注意的是，对于每个旋转角度θ，激光雷达1本身仅记录扫描点的极径r和极角
Figure PCTCN2022131707-appb-000005
而旋转角度θ的信息采集是由转动平台3进行的。
如图1所示，激光雷达1通过例如螺栓连接在支撑板2上，支撑板2通过例如螺栓连接在转动平台3上。通过这种连接，激光雷达1可以在转动平台3的驱动下旋转，从而在固定高度处扫描飞机表面。考虑到飞机的较大体积和占用空间，仅在一个固定高度处执行扫描可能无法在高度方向上得到飞机的预定外表面的点云数据。为此，设置了吊挂平台4，转动平台3通过枢转轴连接到吊挂平台4。吊挂平台4为伸缩结构，包括外套筒支架和可移动地设置在外套筒支架内的内套筒支架，转动平台3连接在内套筒支架的下部，内套筒支架可以上下移动、根据需要伸出到外套筒支架外部或缩回到外套筒支架内，使得内套筒支架的高度可以调节，从而激光雷达1可以沿竖直高度方向上下升降移动，并且可以左右旋转。或者，外套筒支架上安装有轨道，内套筒支架可以沿着轨道在外套筒支架内沿吊挂平台的高度方向上下移动。
参考图2，由于飞机为大型部件，单个激光雷达无法扫描得到飞机的所有部段的点云，因此设置多个激光雷达。这些激光雷达布置在飞机的四周，图2中示出了布置4个测量装置，即4个激光雷达，其中，机头、两侧机翼以及机尾分别布置一个激光雷达，这些激光雷达可以升降并旋转而对这些区域进行扫描。每个激光雷达对应一定扫描区域，这些区域的组合涵盖了飞机的待扫描的预定外表面，因此对这些区域扫描而获得的点云拼接组合在一起即可得到预定表面的点云。预定外表面可以指飞机的整个外表面，也可以指部分外表面(例如飞机上半部分或下半部分的外表面)或一个飞机部段(例如机身、机尾或机翼等)的外表面。根据本实施方式的方法，优化了激光雷达的布局和数量，在成本和精度之间实现了良好的平衡。
相邻的两个对应区域之间的连接接缝可能被两个激光雷达重复扫描到，即 下文所述的重叠扫描区域。图中示出了布置4个激光雷达，但这仅是示例，可以根据所要扫描的区域在相应位置布置更多或更少数量的激光雷达，例如2个、3个、5个、6个或更多个激光雷达。
如上所述安装并布置好多个激光雷达之后，通过标靶进行激光雷达的位置标定。在空间中布置激光雷达后，需要知道激光雷达在世界坐标系中的位置，世界坐标系由三个互相垂直并相交的坐标轴x、y、z组成、是固定不变的坐标系，并且独立于所有部件(包括激光雷达、飞机等)。在实际测量中，飞机自身的坐标以及所停放的位置、激光雷达布置的位置等最终都会转化到世界坐标系中，方便坐标的统一。根据本实施方式的方法，通过标定确定每台激光雷达的位置坐标，将这些激光雷达统一到同一坐标系下，便于后续整个过程的快速实现。
因此，在对飞机进行扫描前，会在世界坐标系的原点处布置一个内部被圆形镂空的正方形标定板(即标靶)，标定板上镂空圆心的坐标就是原点(0,0,0)，通过激光雷达对标定板进行边缘检测，通过点云边缘检测获得圆心坐标，这个坐标就是激光雷达相对于世界坐标系原点的坐标，因此可以获得激光雷达在世界坐标系中的位置。知晓了激光雷达的位置坐标，激光雷达自身后续测量得到的飞机表面点云可以转换到世界坐标系中。值得注意的是，由于激光雷达1在扫描时可能需要通过吊挂平台4上下移动，因此激光雷达在世界坐标系中的位置坐标z会发生变化，此时可以在每次上下移动时利用标靶对激光雷达的位置重新标定，或者也可以测量并记录每次移动的z坐标增量，并将该增量与移动前的z坐标相加，得到新的位置坐标z。
此外，待测量的飞机的三维数学模型已经在计算机中建模，并将其转换到世界坐标系下，由此知道飞机的外表面的多个点的位置坐标。为了便于使用，可以将飞机的三维数学模型的模型坐标系直接替换为世界坐标系，因此下文所述的将点云数据统一到世界坐标系下可以理解为将这些点云数据统一到飞机的数学模型坐标系下，便于二者的对比和匹配。
在多个激光雷达的位置标定好之后，可以对飞机进行扫描。如上所述以及图3所示，在每个激光雷达自身的原点处建立球坐标系和笛卡尔三维坐标系xyz，即激光雷达自身的极坐标系、球坐标系与该笛卡尔三维坐标系三者的原点重合，从而如下所述利用这三个坐标系来表征得到的点云。坐标系建立好以后，激光雷达根据设置好的旋转步进角度Δθ沿预定旋转方向(例如围绕z轴逆时针或顺时针)旋转，每转动一次，执行一次激光扫描，并记录相关数据信息。旋转步进角度Δθ可以为5°，并且以5°/秒的旋转速度进行旋转，即每秒转动5°，然后执行一次扫描。但这仅是示例，旋转步进角度可以为其他更大或更小的数值，例如1°、2°、3°、4°、6°、7°、8°等，相应地，旋转速度可以也可以为 更大或更小的值，例如1°/秒、2°/秒、3°/秒、4°/秒、6°/秒、7°/秒、8°/秒等。根据本实施方式的方法，采用5度作为步进角，即实现了快速扫描的测量速度，又考虑了测量精度。
转动平台3带动激光雷达1实现步进旋转，并在转动到位以后记录该扫描位置处的旋转角度θ。如上所述，激光雷达在每次扫描时得到一条线的点云，并利用自身的极坐标系记录这条线的每个点的距离r、角度
Figure PCTCN2022131707-appb-000006
即得到极坐标
Figure PCTCN2022131707-appb-000007
从而确定旋转平面内的扫描线上所有点的极坐标。由于激光雷达在每个旋转角度θ下具有一个旋转平面，并在每个旋转平面内发射激光，扫描飞机的外表面，激光与飞机的外表面相交为一条扫描线，测量获取这条线的信息，以此类推，通过彼此相继的多次旋转，得到相应多条扫描线的数据信息，这些扫描线和数据进行叠加组合，整体反映飞机外表面的面型信息。
此外，激光雷达还采集并记录每个点的反射强度p，强度p反映的是所采集到的点的明暗程度。比如激光所扫描的物体类似于镜子，则激光的反射强度比较大，当采集镜子上的点时，该点的反射强度的数值会比较大；而如果扫描的物体是黑色的物体，则激光就会被吸收一部分，反射的光会弱一点，反射强度较低，反射强度的数值较小。记录反射强度是为了供下文所述的点云拼接使用。在下述的点云拼接中，高反广度靶点就是反射强度大的点。总结来说，在扫描时，记录并存储激光雷达的旋转角度θ、以及该旋转角度下测得的相应点云的距离r、角度
Figure PCTCN2022131707-appb-000008
强度p。根据本实施方式的方法，通过多次扫描并记录相关数据实现了测量过程的分解以及信息的完整采集。
如上所述，转动平台3记录每个旋转角度θ，同时在每个旋转角度下，激光雷达利用自身的极坐标系得到点云的坐标
Figure PCTCN2022131707-appb-000009
将这些信息组合，每个点可以利用所建立的球坐标系表示，即
Figure PCTCN2022131707-appb-000010
如图3所示，在每个激光雷达上也建立了一个笛卡尔三维坐标系，如下文所述，为了方便将点云统一到世界坐标系下，对于每个激光雷达，首先将其采集得到的点云的球坐标
Figure PCTCN2022131707-appb-000011
转换到每个激光雷达的笛卡尔三维坐标系xyz下的坐标P(x,y,z)。即，利用以下公式(1)将每个雷达测量的点云的球坐标
Figure PCTCN2022131707-appb-000012
转换为笛卡尔坐标P(x,y,z)：
Figure PCTCN2022131707-appb-000013
从而，每个激光雷达获得的点云都表示为对应的笛卡尔坐标。根据本实施方式的方法，实现了点云数据的初次坐标转换。但是这些笛卡尔坐标是相对于每个激光雷达自身的笛卡尔坐标系的，多个激光雷达具有自己的坐标原点，它们所采集的数据彼此之间仍然存在坐标信息的差异。由于多个激光雷达测量的 点云数据未在同一个坐标系下表示，因此不能将这些点云直接拼接在一起，需要将所有点的坐标统一到同一个坐标系下。
如上所述，可以将所有激光雷达测得的所有点云数据统一到世界坐标系下。对于这些点云数据，可以通过高反广度靶点进行公共点拼接来实现。每个激光雷达在扫描平面内都是360度扫描，每两个相邻的激光雷达之间会存在扫描的重叠区域，如果只是扫描一次的话，重叠区域是一条重叠的线，但是在通过转动平台的转动扫描多次时，会存在多条扫描线的重叠。在每两个相邻激光雷达的扫描重叠区域中寻找或识别反射强度明显强于其他测量点的3个和/或3个以上的高反广度靶点，高反广度靶点即是上文所述的反射强度高的点，这些点的反射强度数值大，特征比较明显，容易从点云中挑选出来，否则将难以识别两片点云之间重叠的那些点，进而无法进行点云的拼接。
这些高反广度靶点是相邻两个激光雷达扫描的重叠区域中的公共点，这些公共点分别在两个激光雷达自身的坐标系和所测量的各自点云中具有两组空间位置坐标。将其中一组空间位置坐标通过坐标转换而变换到另一组空间位置坐标的坐标系中，即实现了坐标系的统一。
在重叠区域的点云数据中识别3个及以上的高反广度靶点(即公共点)，是因为对于单个公共点，两片点云可以围绕该单个点在空间中任意旋转，不能唯一确定两片点云之间的实际相对位置。同样的道理，对于两个公共点，两片点云可以围绕这两个点连接形成的直线为轴进行任意旋转，依然不能唯一确定二者之间的位置关系。而对于3个及以上的公共点，这些点可以唯一确定一个平面，该公共平面所连接的两片点云的相对位置可以唯一确定。以共面的3个公共点A、B、C为例，它们在两片点云P和Q(即两个激光雷达的笛卡尔坐标系)中的空间位置坐标分别为A1(x1,y1,z1)、B1(x2,y2,z2)、C1(x3,y3,z3)以及A2(X1,Y1,Z1)、B2(X2,Y2,Z2)、C2(X3,Y3,Z3)，利用以下公式(2)求取坐标A1、B1、C1到A2、B2、C2的变换矩阵(即旋转平移矩阵)，即将两个激光雷达的坐标系统一到其中一个激光雷达的坐标系中：
Figure PCTCN2022131707-appb-000014
式中，
Figure PCTCN2022131707-appb-000015
T＝[x 0,y 0,z 0] T
λ为尺度参数，R(α,β,γ)为旋转矩阵，T为平移矩阵，α、β、γ为三个坐标 轴的旋转参数，x 0、y 0、z 0为平移参数。旋转矩阵和平移矩阵的计算方法是本领域已知的，在此不再赘述。
如上所述，每两个相邻激光雷达彼此之间执行一次坐标变换，最终统一到同一个激光雷达的笛卡尔坐标系下，例如将第二台激光雷达、第三台激光雷达以及第四台激光雷达测得的点云的坐标统一到第一台激光雷达的笛卡尔坐标下，而第一台激光雷达自身已经相对于世界坐标系进行了位置标定、具有确定的对齐关系(即坐标转换矩阵)，最终可以将所有激光雷达以及所测得的点云数据统一到世界坐标系下。注意的是，由于多台激光雷达在扫描之前已经相对于世界坐标系进行了位置标定，这些激光雷达与世界坐标系之间的坐标转换矩阵已经计算好，所以对于每台激光雷达测得的点云，也可以直接利用这些坐标转换矩阵将点云从每台激光雷达的笛卡尔坐标系直接转换到世界坐标系下。根据本实施方式的方法，实现点云数据的二次坐标转换，最终统一到飞机的模型坐标系下。
拼接的作用是把每个激光雷达测到的有限的点云组合起来，得到飞机的预定外表面的点云。对拼接好的点云需要进行滤波处理，滤波是为了除去明显不是飞机表面点云的点。比如一个激光雷达在一次扫描时会记录360度的点云信息，但是这些点云中有一些点并不是飞机表面的点，可能是环境中其他物体的点。滤波的作用是去除不是飞机表面的点(这些点在工程中可以称之为噪声点或噪点)。
由于噪声点具有不同类型，在本公开中，分别针对两种不同类型的噪声点使用两种方法，通过直通滤波与统计滤波对点云进行处理。直通滤波与统计滤波是两种不同的滤波方式，直通滤波是将特定通道范围之外的点云去除，“直通”即基于点云空间坐标系设定一个通道，将点云中的位于通道范围之外的点剔除滤掉，从而保留通道里边的点云。这个“通道”在点云库(Point Cloud Library，PCL)里面表现为针对一个坐标轴设定的限定范围，比如飞机表面的点云在多个坐标轴上是有范围的，因此，范围之外的点明显不属于飞机表面，可以直接通过直通滤波去除。
统计滤波是针对于点云中的离群点进行的，所谓的离群点通常距离其他点云数据较远，它们会使点云的分布不均匀，影响点云的平滑性，降低点云数据的质量，进而影响后续的三维重建模型的精度。但这些离群点是可能在直通滤波的通道范围内出现的，因此直通滤波对这些点是没有作用的，因此，还需要进行统计滤波。
首先进行直通滤波：将拼接好的点云数据输入到直通滤波算法中；设置直通滤波的轴及取值范围；执行直通滤波算法进行滤波。然后对直通滤波后的点 云数据进行统计滤波：设置统计滤波的近邻点个数及标准差乘数；执行统计滤波算法进行滤波；对所有点进行上述循环，直至去除所有的离群点。
根据本实施方式的方法，通过常用的特定滤波方法除去点云中的噪点，针对性强，滤波效果好。
经过上述滤波之后，这些点云数据可以用来对飞机表面重新建模，获得飞机的面型。在实践中，并不是完全利用这些点云数据直接建模获得飞机的面型，而是与已有的三维模型相结合来重建飞机面型，如此可以省时，减少工作量，提高效率。如上所述，计算机中已经存在之前建模好的飞机的三维数学模型，该数学模型与扫描得到的飞机表面一样位于世界坐标系下，因此二者可以直接交互，例如进行比较、匹配。
将滤波后的飞机表面的点云与前述的飞机三维模型进行匹配，从而对飞机表面重构，获得飞机表面的实际坐标数据和信息，获得飞机表面的真实面型。匹配是指例如分析、计算、对比二者之间的相同或等同程度以及差异，并根据异同程度将二者等同或替换使用。例如，当二者之间的差异较小时，可以直接使用已有的三维模型的坐标数据；而当差异较大时，则用实际测得的点云数据替换理论的三维模型。
在将实测点云数据与三维模型匹配时，计算对应的点之间的误差，即计算模型重建误差。例如，对于这些点，可以计算对应点之间的距离，即坐标差值。例如，设三维模型中的点的坐标为(x,y,z)，实测点云中对应的点的坐标为(x’,y’,z’)，这两个点之间的距离为
Figure PCTCN2022131707-appb-000016
替代地，也可以计算其他项之间的误差，例如仅计算点的x坐标、坐标y或坐标z之间的差值。根据本实施方式的方法，利用点之间的距离作为误差度量，计算简单，易于实现。下文以计算点之间的距离作为误差为例，对模型的重建进行说明。
对于实测点云和三维模型中的每对对应的点，如上所述计算它们之间的距离，并将计算结果与预定的误差允许阈值(即距离阈值)进行比较。当它们之间的距离在距离阈值以内时，即误差较小时，仍然利用三维模型的点构建模型；而当计算的距离在阈值以外时，即误差比较大时，利用实测点云的点代替三维模型的点来重建模型。两种极端的情况是，(1)实测的点云中的每个点与三维模型中的对应点之间的误差均在阈值以内时，则直接使用三维模型即可，不需要对模型进行重建；(2)实测的点云中的每个点与三维模型中的对应点之间的误差均不在阈值以内时，则需要弃用数学模型，而使用点云的所有点对模型进行重建。通过实测点云和原有三维模型相结合，既利用实测数据保证获得的飞机面型的准确性和精度，又择优利用原有理论三维模型的数据信息提高效率。
利用泊松(Poisson)算法对上述在误差阈值之外的点云数据进行曲面重建。 将在误差阈值以外的点云数据输入到Poisson算法中；设置合适的精细度参数，参数为degree[1,5]，值越大越精细，耗时越久；设置落入一个八叉树结点中的样本点的最小数量pn.setSamplesPerNode()，无噪声取为[1.0-5.0]，有噪声取为[15.0-20.0]，运行程序，重建完成。本申请的方法简单、步骤少、易于快速实现，利用已有理论模型和实测数据，进行误差分析，提高了测量效率的同时也保证了检测精度。
在符合本领域常识的基础上，上述多个可选条件，可任意组合，即得本申请多个可选实例。
本申请针对于大型飞机表面重构、喷涂以及维修，提出了一种获得飞机面型的系统和方法，能够实现高精度的飞机表面测量；本申请对激光雷达获取的点云数据进行了拼接、滤波、重建处理，可以集成到一套软件中，可以实现从点云数据的获取到模型获得的快捷转化，同时最后进行了模型误差分析，提高了测量效率的同时也保证了检测精度。

Claims (12)

  1. 一种获得飞机面型的方法，包括：
    建立飞机的三维数学模型，并获取所述飞机的三维数学模型在模型坐标系中的三维坐标；
    将多台激光雷达布置在所述飞机的四周，并控制每台激光雷达扫描所述飞机的预定区域以测量得到每个预定区域的点云；
    将通过所述多台激光雷达测量得到的多个预定区域的点云进行拼接；
    对拼接好的点云进行滤波处理，以除去所述拼接好的点云中的噪点；以及
    将所述飞机的经过所述滤波处理后的点云和所述飞机的三维数学模型进行对比，在二者之间的误差超过阈值的情况下，重建所述飞机的三维数学模型。
  2. 如权利要求1所述的方法，其中，所述将所述多台激光雷达布置在所述飞机的四周的过程包括：利用标靶对每台激光雷达的位置进行标定，并将所述每台激光雷达相对于所述标靶的位置换算得到在所述飞机的模型坐标系下的坐标。
  3. 如权利要求2所述的方法，其中，所述将所述多台激光雷达布置在所述飞机的四周的过程还包括：基于预定的旋转步进角度Δθ和旋转方向多次旋转所述多台激光雷达，使得每旋转一次，每台激光雷达对所述飞机执行一次扫描，并记录旋转角度θ、每个扫描点在所述每台激光雷达自身的极坐标系下的距离r和角度
    Figure PCTCN2022131707-appb-100001
    以及所述每个扫描点的反射强度p，从而得到所述每个扫描点的球坐标A
    Figure PCTCN2022131707-appb-100002
  4. 如权利要求3所述的方法，其中，所述将所述多台激光雷达布置在所述飞机的四周的过程还包括：在每台激光雷达处建立笛卡尔坐标系，并将所述每台激光雷达测得的每个扫描点的球坐标转换为所述笛卡尔坐标。
  5. 如权利要求4所述的方法，其中，所述将通过所述多台激光雷达测量得到的多个预定区域的点云进行拼接的过程包括：在所述多个预定区域中的每两个相邻的预定区域的重叠点云中选取至少三个反射强度高的公共点；将所述至少三个反射强度高的公共点在一台激光雷达下的笛卡尔坐标转换到另一台激光雷达的笛卡尔坐标系下，并获得所述一台激光雷达的笛卡尔坐标系和所述另一台激光雷达的笛卡尔坐标系之间的坐标转换矩阵；以及求取每台激光雷达的笛卡尔坐标系与所述飞机的模型坐标系之间的坐标转换矩阵，从而最终将所述多个预定区域的点云中的所有点的坐标统一到所述飞机的模型坐标系下。
  6. 如权利要求5所述的方法，其中，所述对拼接好的点云进行滤波处理的过程包括利用直通滤波与统计滤波对拼接的点云进行滤波处理。
  7. 如权利要求6所述的方法，其中，所述将所述飞机的经过所述滤波处理后的点云和所述飞机的三维数学模型进行对比的过程包括：计算所述滤波处理后的点云中每个点与所述飞机的三维数学模型中对应点之间的距离，在所述距离在阈值以内的情况下，利用所述飞机的三维数学模型中的点构建模型；在所述距离在所述阈值以外的情况下，利用所述滤波处理后的点云中的点重建所述飞机的三维数学模型。
  8. 如权利要求1所述的方法，其中，所述多台激光雷达均为单线激光雷达。
  9. 如权利要求1所述的方法，其中，所述多台激光雷达包括分别布置在所述飞机的机头、两侧机翼以及机尾处的4台激光雷达。
  10. 如权利要求3所述的方法，其中，所述预定的旋转步进角度Δθ为5°。
  11. 一种用于执行如权利要求1至10中任一项所述的方法的飞机面型测量系统，包括：布置在飞机四周的多台激光雷达；
    对于每台激光雷达，所述飞机面型测量系统还包括：支撑板，所述支撑板被构造成能够固定所述每台激光雷达；转动平台，所述转动平台与所述支撑板连接并能够带动所述支撑板旋转；以及，吊挂平台，所述吊挂平台安装有轨道，使得所述转动平台能够沿着所述轨道沿所述吊挂平台的高度方向上下移动。
  12. 一种计算机可读存储介质，包括用于执行如权利要求1至10中任一项所述的方法的计算机程序。
PCT/CN2022/131707 2022-01-26 2022-11-14 获得飞机面型的系统和方法 WO2023142608A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210091050.XA CN116540255A (zh) 2022-01-26 2022-01-26 利用多台激光雷达测量并获得飞机面型的系统和方法
CN202210091050.X 2022-01-26

Publications (1)

Publication Number Publication Date
WO2023142608A1 true WO2023142608A1 (zh) 2023-08-03

Family

ID=87452941

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/131707 WO2023142608A1 (zh) 2022-01-26 2022-11-14 获得飞机面型的系统和方法

Country Status (2)

Country Link
CN (1) CN116540255A (zh)
WO (1) WO2023142608A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116772729A (zh) * 2023-08-22 2023-09-19 中铁二十三局集团第一工程有限公司 基于激光雷达的桥梁预制构件外观尺寸检测方法
CN117440308A (zh) * 2023-12-21 2024-01-23 深圳市好兄弟电子有限公司 一种麦克风话筒生产产线智能控制方法及系统
CN117440308B (zh) * 2023-12-21 2024-04-26 深圳市好兄弟电子有限公司 一种麦克风话筒生产产线智能控制方法及系统

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104180791A (zh) * 2014-07-30 2014-12-03 中国航空工业集团公司沈阳飞机设计研究所 一种飞机水平测量方法
JP2015010911A (ja) * 2013-06-28 2015-01-19 公益財団法人鉄道総合技術研究所 空撮測量方法および装置
CN105651160A (zh) * 2014-11-04 2016-06-08 北京浩宇天地测绘科技发展有限公司 一种安装定位装置的三维激光扫描仪球形标靶及其使用方法
US20190033461A1 (en) * 2017-07-28 2019-01-31 The Boeing Company Live metrology of an object during manufacturing or other operations
CN109872397A (zh) * 2019-02-18 2019-06-11 北京工业大学 一种基于多目立体视觉的飞机零件的三维重建方法
CN110363861A (zh) * 2019-07-14 2019-10-22 南京林业大学 基于激光雷达点云的田地作物三维重构方法
CN111540063A (zh) * 2020-04-15 2020-08-14 富德康(北京)科技股份有限公司 一种基于多站激光点云数据全自动高精度拼接方法
CN112489212A (zh) * 2020-12-07 2021-03-12 武汉大学 一种基于多源遥感数据的建筑物智能化三维测图方法
CN113379844A (zh) * 2021-05-25 2021-09-10 成都飞机工业（集团）有限责任公司 一种飞机大范围表面质量检测方法
CN113390605A (zh) * 2021-07-20 2021-09-14 中国空气动力研究与发展中心设备设计与测试技术研究所 一种风洞试验飞机机翼变形全场测量方法

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015010911A (ja) * 2013-06-28 2015-01-19 公益財団法人鉄道総合技術研究所 空撮測量方法および装置
CN104180791A (zh) * 2014-07-30 2014-12-03 中国航空工业集团公司沈阳飞机设计研究所 一种飞机水平测量方法
CN105651160A (zh) * 2014-11-04 2016-06-08 北京浩宇天地测绘科技发展有限公司 一种安装定位装置的三维激光扫描仪球形标靶及其使用方法
US20190033461A1 (en) * 2017-07-28 2019-01-31 The Boeing Company Live metrology of an object during manufacturing or other operations
CN109872397A (zh) * 2019-02-18 2019-06-11 北京工业大学 一种基于多目立体视觉的飞机零件的三维重建方法
CN110363861A (zh) * 2019-07-14 2019-10-22 南京林业大学 基于激光雷达点云的田地作物三维重构方法
CN111540063A (zh) * 2020-04-15 2020-08-14 富德康(北京)科技股份有限公司 一种基于多站激光点云数据全自动高精度拼接方法
CN112489212A (zh) * 2020-12-07 2021-03-12 武汉大学 一种基于多源遥感数据的建筑物智能化三维测图方法
CN113379844A (zh) * 2021-05-25 2021-09-10 成都飞机工业（集团）有限责任公司 一种飞机大范围表面质量检测方法
CN113390605A (zh) * 2021-07-20 2021-09-14 中国空气动力研究与发展中心设备设计与测试技术研究所 一种风洞试验飞机机翼变形全场测量方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
FENG, SHANGZHAO; LIU, SHA: "Aircraft Point Cloud Preprocessing Technology Based on Laser Scanning", JISUANJI YU SHUZI GONGCHENG - COMPUTER AND DIGITAL ENGINEERING, ZHONGGUO CHUANBO ZHONGGONG JITUAN GONGSI. DI-709 YANJIUSUO, CN, vol. 46, no. 7, 20 July 2018 (2018-07-20), CN , pages 1279 - 1283, XP009548024, ISSN: 1672-9722, DOI: 10.3969/j.issn.1672-9722.2018.07.002 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116772729A (zh) * 2023-08-22 2023-09-19 中铁二十三局集团第一工程有限公司 基于激光雷达的桥梁预制构件外观尺寸检测方法
CN116772729B (zh) * 2023-08-22 2024-01-09 中铁二十三局集团第一工程有限公司 基于激光雷达的桥梁预制构件外观尺寸检测方法
CN117440308A (zh) * 2023-12-21 2024-01-23 深圳市好兄弟电子有限公司 一种麦克风话筒生产产线智能控制方法及系统
CN117440308B (zh) * 2023-12-21 2024-04-26 深圳市好兄弟电子有限公司 一种麦克风话筒生产产线智能控制方法及系统

Also Published As

Publication number Publication date
CN116540255A (zh) 2023-08-04

Similar Documents

Publication Publication Date Title
Fröhlich et al. Terrestrial laser scanning–new perspectives in 3D surveying
WO2022170878A1 (zh) 一种无人机测量输电线路影像距离的系统及方法
CN109900712B (zh) 基于摄像测量的风电叶片机械故障在线检测系统及其方法
Shang et al. Measurement methods of 3D shape of large-scale complex surfaces based on computer vision: A review
CN106932784B (zh) 基于二维激光雷达的车斗形容器三维扫描系统测量方法
CN109900713B (zh) 摄像引导的无人机风电叶片缺陷动态检测系统及其方法
Xia et al. Detection method of manufacturing defects on aircraft surface based on fringe projection
Acosta et al. Laser triangulation for shape acquisition in a 3D scanner plus scan
CN111737822B (zh) 一种基于点云数据的航空构件贴合间隙三维形貌评估方法
CN105806251A (zh) 基于线激光传感器的四轴测量系统及其测量方法
Wang et al. Density-invariant registration of multiple scans for aircraft measurement
Wu et al. A novel high precise laser 3D profile scanning method with flexible calibration
WO2023142608A1 (zh) 获得飞机面型的系统和方法
Aldao et al. Metrological comparison of LiDAR and photogrammetric systems for deformation monitoring of aerospace parts
Wieczorowski et al. The use of drones in modern length and angle metrology
US8467992B1 (en) Vision based location and measurement device and methods
Kim et al. As-is geometric data collection and 3D visualization through the collaboration between UAV and UGV
Zhang et al. Global measurement method for large-scale components based on a multiple field of view combination
CN113203385B (zh) 一种无干涉五轴扫描轨迹生成方法及系统
Kaushik et al. Accelerated patch-based planar clustering of noisy range images in indoor environments for robot mapping
Malek et al. Realtime conversion of cracks from pixel to engineering scale using Augmented Reality
CN113702384A (zh) 一种回转构件表面缺陷检测装置、检测方法及标定方法
Jaw et al. Feature-based registration of terrestrial lidar point clouds
Wang et al. Pose calibration of line structured light probe based on ball bar target in cylindrical coordinate measuring machines
Sánchez et al. Fast and accurate mesh registration applied to in-line dimensional inspection processes

Legal Events

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

Ref document number: 22923402

Country of ref document: EP

Kind code of ref document: A1