Object size high-precision measurement method and system based on multi-laser profilometer

Abstract

The application discloses a high-precision object size measurement method and system based on a multi-laser profiler, which comprises the following steps: a system calibration step, in which a standard calibration object is used to calculate the transformation matrix of each scanner coordinate system relative to the global reference coordinate system; a template establishment step, in which the complete point cloud of a standard template object is collected and the theoretical boundary position thereof is recorded; during actual measurement, multiple scanners are synchronously triggered to collect local profile point clouds, which are spliced into a complete profile by using the transformation matrix; the actual boundary is extracted by using the AlphaShape method, and the translation deviation and rotation deviation of the template boundary are calculated; dynamic compensation is performed based on the template size, and the actual width and length are calculated; the system comprises N ring-type laser profilers, a synchronous control module and a central processing unit. The application realizes full-view angle non-blind area measurement, sub-pixel level point cloud fusion and dynamic compensation for randomly placed objects, and is suitable for high-precision online detection of high-transmittance and high-reflectivity precision workpieces.

Description

Technical Field

[0001] This invention relates to the field of industrial automation precision measurement technology, and in particular to a high-precision measurement method and system for object size based on a multi-laser profile scanner. Background Technology

[0002] In the current field of industrial automation and precision manufacturing, especially for online dimensional inspection of mobile phone cover glass (CG), display panels (LCD / OLED), and various precision workpieces, the commonly used technologies include single-point laser ranging combined with displacement platform scanning, single 2D line laser profilometer scanning, and traditional 2D vision imaging measurement. However, these existing technologies have revealed significant limitations and technical bottlenecks when facing the high-precision and high-efficiency production demands of Industry 4.0.

[0003] First, although the "single-point laser ranging + displacement platform" solution has a simple structure, its measurement efficiency is extremely low. It must rely on the reciprocating motion of the mechanical platform to construct the contour, which not only increases the complexity of the equipment, but also makes the measurement results highly susceptible to interference from the accuracy of the mechanical guide rail, wear and tear and environmental vibration, making it difficult to meet the cycle time requirements of high-speed production lines.

[0004] Secondly, although some production lines have introduced 2D line laser profilometers, they typically only use a single device for cross-sectional scanning. The biggest drawback of this approach is the blind spot and occlusion problem: when measuring large objects or irregularly shaped workpieces with complex structures, the object's own geometry often obstructs the laser beam, and when the size is large, the laser field of view cannot cover the entire product, resulting in missing edge data and making it impossible to achieve complete capture of the entire perimeter.

[0005] Furthermore, traditional visual camera measurement technology is highly dependent on optical imaging quality. For objects with highly reflective surfaces or transparent / semi-transparent materials, due to the complexity of light refraction and reflection, it is difficult for the camera to accurately extract edge features. Moreover, the light source scheme is difficult to determine, and even slight shaking can cause imaging instability, resulting in large fluctuations in measurement results and a high signal-to-noise ratio.

[0006] Furthermore, existing multi-sensor measurement solutions often lack a rigorous coordinate unification mechanism. Each sensor typically operates independently, and the data is simply superimposed during post-processing. This coarse fusion method cannot eliminate splicing misalignments caused by sensor installation angle deviations and positional errors, significantly reducing the accuracy of overall dimension calculations. Simultaneously, most existing technologies assume the object is ideally centered on the conveyor belt, lacking dynamic compensation capabilities for the object's actual orientation (rotation, translation). Once the object shifts, the measurement error increases significantly. Summary of the Invention

[0007] The main objective of this invention is to provide a high-precision measurement method and system for object size based on a multi-laser contour scanner, aiming to solve existing technical problems.

[0008] To achieve the above objectives, this invention provides a high-precision object size measurement method based on a multi-laser contour scanner, comprising the following steps: Calibration steps: Using standard calibration objects, calibrate the spatial position transformation relationship between N laser profile scanners, where N≥2, and use the coordinate system of one of the scanners as the global reference coordinate system; Template creation steps: Using a standard template object, the complete point cloud data of the standard template object in the global reference coordinate system is collected by the N laser contour scanners, and the theoretical boundary position information of the standard template object is recorded; Acquisition and stitching steps: When the object to be measured enters the measurement area, the N laser contour scanners are simultaneously triggered to acquire local contour point clouds of different parts of the object to be measured, and the spatial position transformation relationship is used to transform all local contour point clouds to the global reference coordinate system and stitch them together to form the complete contour point cloud of the object to be measured. Deviation calculation steps: Extract the actual boundary position of the object to be measured from the complete contour point cloud, and calculate the pose deviation between the actual boundary position and the theoretical boundary position of the standard template object. The pose deviation includes translation deviation and / or rotation deviation. Size compensation step: Based on the extracted actual boundary position and combined with the pose deviation, the actual size of the object to be measured is calculated through dynamic compensation.

[0009] Furthermore, in the calibration step: The standard calibration object is a standard rectangular calibration plate, whose standard length and width dimensions are known. Each laser contour scanner extracts the edge contour coordinates of the standard rectangular calibration plate in its respective field of view, and fits the boundary line using the least squares method. Using the coordinate system of one of the laser profile scanners as the global reference coordinate system, and employing the fitted boundary lines, the transformation matrix T of the coordinate systems of the other laser profile scanners relative to the global reference coordinate system is calculated. ij The transformation matrix T ij It includes rotation component R and translation component t, used to describe the spatial transformation relationship between coordinate systems, satisfying: P global =T ij ·P local ; Among them, P local P represents the local point cloud coordinates. global These are the transformed global coordinates.

[0010] Furthermore, in the deviation calculation step, the actual boundary position of the object to be measured is found and extracted from the complete contour point cloud using the AlphaShape method, specifically including: Perform Delaunay triangulation on the input point cloud data S; Traverse all triangles obtained from triangulation, and filter the edges that form an AlphaShape according to a preset parameter α. The filtering criterion is: if the radius of the circumcircle of the triangle is R... circum If ≤α, then all three sides of the triangle are candidate sides; Collect all edges that meet the conditions and remove duplicate edges to form the boundary contour of the object under test. The parameter α is used to control the compactness of the boundary. The smaller the value of α, the closer the boundary is to the original point cloud. The larger the value of α, the closer the boundary is to the convex hull.

[0011] Furthermore, in the size compensation step, the actual width W of the object to be measured... actual and actual length L actual The calculation uses a fine-tuning compensation formula based on template size: The width compensation formula is: W actual =W template +(ΔX right -ΔX left ); The length compensation formula is: L actual =L template +(ΔY bottom -ΔY top ); Among them, W template L is the width of the standard template object. template The length of the standard template object; ΔX left Let ΔX be the deviation of the left boundary of the object to be measured from the left boundary of the template on the X-axis. right This is the deviation of the right boundary of the object to be measured from the right boundary of the template on the X-axis. ΔY top ΔY is the deviation between the upper boundary of the object to be measured and the upper boundary of the template on the Y-axis. bottom This represents the deviation between the lower boundary of the object under test and the lower boundary of the template on the Y-axis.

[0012] Furthermore, in the deviation calculation step, the rotational deviation θ is calculated in the following manner: Edge lines in the complete contour point cloud are detected by Hough transform; Calculate the angle between the detected edge line and the corresponding edge line of the standard template object, and use this angle as the rotation angle deviation θ of the object under test; Furthermore, by calculating the actual geometric center of the object under test and the geometric center C of the standard template object... template The offset between the two values ​​is used to obtain the translation deviation.

[0013] Furthermore, when the object under test is a workpiece made of a material with high light transmittance or high reflectivity, a point cloud preprocessing step based on material characteristics is included after the acquisition and stitching step and before the deviation calculation step: Set a filter window that matches the material properties to filter the original point cloud and suppress ambient light interference and high reflectivity noise; A threshold segmentation strategy that matches the material properties is used to separate the point cloud data of the target object from the background; The logic for removing chamfered interference points is executed. By utilizing the normal vector change features or curvature features of the point cloud at the chamfered edge, false contour points generated by the chamfered edge or curved surface of the object are identified and removed, while the true main body edge point cloud of the object is preserved.

[0014] A high-precision object size measurement system based on a multi-laser contour scanner, characterized in that, for performing the above-described method, the system comprises: N laser contour scanners, where N≥2, are fixedly installed around the measurement area to form a surrounding distribution layout of the object to be measured. Each laser contour scanner is used to collect local contour point cloud data of the corresponding side of the object to be measured. The data acquisition synchronization control module is connected to each of the N laser contour scanners and is used to send a synchronization trigger signal to all laser contour scanners to ensure that each scanner acquires data at the same time or according to the same spatial position reference. A central processing unit is connected to the N laser contour scanners and the data acquisition synchronization control module, respectively. The central processing unit includes: The calibration module is used to calculate the transformation matrix of each laser profile scanner coordinate system relative to the global reference coordinate system using standard calibration objects; The template storage module is used to store the theoretical boundary position information and geometric center data of standard template objects in the global reference coordinate system; The point cloud stitching module, connected to the calibration module, is used to transform the local contour point clouds collected by each laser contour scanner to the global reference coordinate system in real time using the transformation matrix, and stitch them together to form a complete object contour point cloud. The pose calculation module is connected to the point cloud stitching module and the template storage module. It is used to extract the actual edge coordinates of the object under test from the complete object contour point cloud and to calculate the translational deviation and rotational angle deviation between the object under test and the standard template object. The size compensation module, connected to the pose calculation module, is used to calculate and output the actual size of the object under test based on the actual edge coordinates and the translation and rotation angle deviations using a dynamic compensation formula.

[0015] Furthermore, the number of the N laser contour scanners is 4, which are respectively set above, below, left and right of the object to be measured. The included angle between adjacent scanners is 90 degrees, forming an independent full-coverage measurement architecture for the four edges of the object to be measured. The upper scanner is responsible for capturing the upper edge contour of the object under test, the lower scanner is responsible for capturing the lower edge contour, the left scanner is responsible for capturing the left edge contour, and the right scanner is responsible for capturing the right edge contour.

[0016] Furthermore, the system also includes: A conveyor belt motion control system is used to carry and transport the object to be measured through the measurement areas of the N laser contour scanners. An encoder, connected to the conveyor belt motion control system and the data acquisition synchronization control module, is used to generate pulse signals according to the movement distance of the conveyor belt. The data acquisition synchronization control module, based on the pulse signal from the encoder, synchronously triggers the N laser contour scanners to acquire data when the object under test moves to a preset trigger position, thereby achieving online dynamic measurement synchronized with the movement of the conveyor belt.

[0017] Furthermore, the central processing unit also includes a hand-eye calibration module for establishing a coordinate mapping relationship between the measurement system and the external robotic arm; The hand-eye calibration module is configured as follows: Obtain the laser sensor coordinates (x, y) of a standard sphere at multiple different spatial locations. si ,y si ,z si At the same time, the theoretical coordinates (x, y) of the end effector of the robot arm in the robot arm's base coordinate system are obtained. ri ,y ri ,z ri ); Using the collected coordinates of multiple corresponding points, the rotation matrix R and translation vector T from the laser sensor coordinate system to the robot's base coordinate system are solved using the singular value decomposition algorithm, satisfying the following relationship: ; The obtained rotation matrix R and translation vector T are output as calibration results, which are used to convert the point cloud coordinates of the measurement system into the motion coordinates of the robot, realizing robot guidance and positioning based on vision measurement.

[0018] The beneficial effects of this invention are reflected in: This invention employs N (N≥2, preferably 4) laser contour scanners, arranged in a ring or array around the object to be measured. Each scanner independently captures a local contour of one side of the object. Compared to traditional single 2D line laser contour scanning schemes, this invention completely eliminates measurement blind spots caused by the limited field of view of a single scanner and occlusion due to the object's own geometric structure. Through the collaborative acquisition of multi-view data, seamless and complete capture of the entire perimeter contour can be achieved regardless of the object's size or shape, significantly improving the feasibility and data integrity of online measurement of complex workpieces and large objects.

[0019] This invention introduces a rigorous global coordinate unification method based on a physical calibration board. During system initialization, the transformation matrix of each scanner's coordinate system relative to the global reference coordinate system is accurately calculated using a standard calibration board and either the least squares method or the iterative nearest point algorithm. This calibration method effectively compensates for coordinate system offsets caused by mechanical installation angle deviations and positional errors, enabling local point cloud data acquired by different scanners to be transformed and stitched into the same global coordinate system with sub-pixel precision. Compared to the simple data overlay in existing technologies, this invention significantly reduces stitching misalignment errors, laying a solid data foundation for subsequent high-precision dimensional calculations.

[0020] For special workpieces with high light transmittance and high reflectivity, such as mobile phone cover glass and display panels, existing vision camera technologies are easily affected by light refraction and reflection, making it difficult to reliably extract edge features. This invention leverages the inherent insensitivity of laser contour scanners to ambient light and surface reflection, further optimizing the point cloud preprocessing algorithm. By designing a dedicated filtering window, an adaptive threshold segmentation strategy, and chamfer interference point removal logic, it effectively suppresses false contour points generated by glass edge chamfers and curved surfaces, accurately extracting the true main body edge of the workpiece. This enables the invention to be stably applied to precision glass device inspection scenarios where traditional vision solutions are inadequate, significantly broadening the applicability of the technology.

[0021] This invention abandons the stringent prerequisite of existing technologies that "objects must be ideally centered or placed at a fixed angle on the conveyor belt." By pre-establishing a standard template database, during actual measurement, the system can automatically identify the actual boundaries of the object under test in real time using an edge detection algorithm, and accurately calculate its translational and rotational deviations relative to the standard templates. Based on these dynamic deviations, this invention directly corrects the final output dimension result through a compensation formula. This dynamic compensation mechanism ensures that even if the object is randomly placed on the conveyor belt, has a positional offset, or is tilted at an angle, the system can still accurately restore its true dimensions, thereby greatly simplifying the positioning requirements of the feeding mechanism, reducing the frequency of manual intervention, and significantly improving the efficiency of online automated inspection and the fault tolerance of the production line. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the high-precision object size measurement method based on a multi-laser contour scanner according to the present invention. Figure 2 This is a schematic diagram of the data flow of the device of the present invention; Figure 3 This is a schematic diagram of laser scanning according to the present invention; Figure 4 This invention relates to the AlphaShape point cloud edge finding method; Figure 5 This invention relates to the AlphaShape point cloud edge finding method; Figure 6 This is a schematic diagram of the high-precision object size measurement system based on a multi-laser contour scanner according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1-5 The present invention provides a high-precision measurement method for object size based on a multi-laser contour scanner, comprising the following steps; Calibration preparation: Use a standard rectangular calibration plate as the standard calibration object. The standard length and width dimensions Dstd of the calibration plate are known. Place the standard rectangular calibration plate at the center of the measurement area, making it approximately parallel to the scanning plane of each scanner.

[0025] Data Acquisition: Each laser contour scanner S1 to S4 independently acquires the edge contour data of the calibration plate in its respective field of view. Specifically, S1 acquires the upper edge contour data of the calibration plate, S2 acquires the lower edge contour data, S3 acquires the left edge contour data, and S4 acquires the right edge contour data.

[0026] Boundary line fitting: For the edge contour point cloud data acquired by each scanner, the least squares method is used to fit the boundary line. Taking S1 as an example, let its upper edge point cloud coordinates be (x... i ,y i For each i = 1, 2, ..., m, the fitted straight line equation is y = k1x + b1. The objective function is minimized as follows: E= 2 ; Solving for the slope k1 and intercept b1 allows us to determine the equation of the straight line at the edge of the calibration plate in the field of view S1. Similarly, the equations of the straight lines at the corresponding edges in the fields of view S2, S3, and S4 can be fitted respectively.

[0027] Transformation Matrix Solution: Using the coordinate system of one scanner (e.g., S1) as the global reference coordinate system, and employing the fitted calibration plate edge lines from each scanner, calculate the transformation matrix Tij for the other scanners (S2, S3, S4) relative to the reference coordinate system. This transformation matrix Tij contains rotation components R and translation components t, describing the spatial transformation relationship between the coordinate systems. The calibration mathematical model is as follows: P global =T ij ·P local ; Among them, P local P represents the local point cloud coordinates in a certain scanner coordinate system. global These are the transformed global coordinates, and Tij is the transformation matrix obtained by solving.

[0028] In the specific solution process, multiple sets of corresponding point constraint equations are established using observation data from each scanner at the same physical edge of the calibration plate. The rotation matrix R and translation vector t are then solved using the least squares method or singular value decomposition (SVD) algorithm. To ensure calibration accuracy, multiple sets of data from different locations on the calibration plate can be collected for joint optimization to further improve the accuracy of the transformation matrix solution.

[0029] To achieve dynamic compensation, the system needs to establish a standard template database.

[0030] Standard template data acquisition: A standard, qualified CG glass cover plate is placed at the center of the measurement area as a "standard template". The data acquisition synchronization control module synchronously triggers four scanners to acquire data. Using the transformation matrix Tij calculated in step two, all local point clouds are transformed into the global reference coordinate system in real time and stitched together to form a complete object contour point cloud.

[0031] Theoretical boundary position recording: Accurately record the theoretical boundary position of the standard template in the field of view of each laser contour scanner. Specifically, record the theoretical X coordinate of the left edge of the template in the global coordinate system as X. left_template X on the right right_template The upper boundary is Y top_template The lower boundary is Y bottom_template .

[0032] Geometric center calculation: Based on the boundary data above, calculate the geometric center Ctemplate of the standard template: C template =( , ); The theoretical boundary location information and geometric center data of the standard template are stored in the template storage module of the central processing unit as a reference benchmark for subsequent matching.

[0033] When the object to be measured (the actual CG glass cover) enters the measurement area along the conveyor belt, the system begins to execute the actual measurement process.

[0034] Synchronous Trigger Acquisition: Based on the pulse signal from the conveyor belt encoder or an external trigger signal, the data acquisition synchronization control module synchronously starts four scanners to acquire data when the object under test moves to the preset trigger position. Each scanner is only responsible for acquiring the local contour point cloud of the corresponding side of the object.

[0035] Point cloud data preprocessing: For test objects made of highly transparent or highly reflective materials (such as CG glass covers), a point cloud preprocessing step tailored to the material characteristics is performed before point cloud stitching. Filtering: Set a filter window that matches the material properties to filter the original point cloud, suppressing ambient light interference and high reflectivity noise. Specifically, statistical filtering or radius filtering methods can be used to remove outliers that are far from the main point cloud.

[0036] Threshold segmentation: A threshold segmentation strategy matching the material properties is employed to separate the point cloud data of the target object from the background. Based on the reflectivity of the glass surface, a reasonable intensity threshold is set to distinguish the point cloud of the glass object from the background point cloud.

[0037] Chamfer interference point removal: By utilizing the normal vector change characteristics or curvature characteristics of the point cloud at the chamfered edge, false contour points generated by the chamfered or curved surface of the object's edge are identified and removed, while retaining the true main body edge point cloud of the object. Specifically, the normal vector of each point cloud is calculated. If the change in the normal vector in the neighborhood of a point exceeds a preset threshold, the point is identified as an interference point in the chamfered area and removed.

[0038] Point cloud stitching: The central processing unit receives data from all scanners and uses the transformation matrix T calculated in step two. ij It converts all local point clouds to a unified global reference coordinate system in real time and stitches them together to form a complete object outline image.

[0039] In the global point cloud, the system extracts the actual edge coordinates of the object under test through an edge detection algorithm and calculates the deviation between the actual edge and the template edge.

[0040] Point Cloud Boundary Extraction: This implementation uses the AlphaShape method to find and extract the boundaries of the object under test from the complete contour point cloud. This method reconstructs the latent geometry from a discrete point set, controlling the "compactness" of the boundary through an adjustable parameter α, thereby establishing a continuous geometric abstraction level between the convex hull of the point cloud and the original point set. The specific implementation steps are as follows: Step 1: Construct Triangulation. Perform Delaunay triangulation on the input point cloud data S. Delaunay triangulation has a key property: the circumcircle of any triangle does not contain any other points in the point set S. This property naturally aligns with the empty circle criterion of AlphaShape.

[0041] Step 2: Edge Filtering. Traverse all triangles obtained from the triangulation and filter the edges that form the AlphaShape based on the α parameter. For a triangle composed of vertices Pa, Pb, and Pc, calculate its circumradius Rcircum. The filtering criterion is: if Rcircum ≤ α, then Pa, Pb, and Pc are all candidate edges.

[0042] Step 3: Generate Boundary. Collect all edges that meet the conditions, remove duplicate edges, and collectively form the boundary contour of the point cloud. The parameter α controls the compactness of the boundary: the smaller the α value, the closer the boundary is to the original point cloud; the larger the α value, the closer the boundary is to the convex hull. In this embodiment, the α value can be set to 2 to 5 times the average point distance of the point cloud, depending on the point cloud density and measurement accuracy requirements.

[0043] Position deviation calculation calculates the deviation between the actual edge and the template edge. Specifically, it calculates the deviation between the actual left boundary and the template left boundary on the X-axis: ΔX left =X left_actual X left_template ; Similarly, calculate: ΔX right =X right_actual X right_template ; ΔY top =Y top_actual Y top_template ; ΔY bottom =Y bottom_actual Y bottom_template ; Rotation angle deviation calculation: Edge lines in the complete contour point cloud are detected using Hough transform. The angle between the detected edge lines and the corresponding edge lines of the standard template object is calculated, and this angle is taken as the rotation angle deviation θ of the object under test. Specifically, a Hough space transform is performed on the extracted boundary point cloud. The peak values ​​of the lines are detected in the parameter space (ρ,θ) to determine the equations of each edge line, and then the angles between the corresponding edge lines are calculated. Simultaneously, the translation deviation is obtained by calculating the offset between the actual geometric center of the object under test and the geometric center Ctemplate of the standard template object.

[0044] By using the extracted actual boundary coordinates and combining them with the pose deviation, the actual size of the object under test is calculated through dynamic compensation.

[0045] Width Calculation: The actual width of the object to be measured can be calculated directly. W actual =∣X right_actual X left_actual | Alternatively, fine-tune the compensation based on the template size Wtemplate: W actual =W template +(ΔX right ΔX left Where Wtemplate is the width of the standard template object.

[0046] Length Calculation: Similarly, the formula for calculating the actual length (Lactual) of the object to be measured is: L actual =∣Y bottom_actual Y top_actual | Alternatively, fine-tune the compensation based on the template size Ltemplate: L actual =L template +(ΔY bottom ΔY top Where Ltemplate is the length of the standard template object.

[0047] Rotation Compensation: When the object under test has a rotational angle deviation θ, the actual boundary coordinates are first rotated to compensate before calculating the width and length. Specifically, the actual boundary point cloud is rotated around the geometric center by an angle of -θ to eliminate the rotational deviation before dimensional calculation. Through the above compensation mechanism, even if the object undergoes translation or slight rotation on the conveyor belt, the system can compensate for these errors through mathematical correction and output high-precision measurement results.

[0048] When this measurement system needs to be linked with a robotic arm (e.g., to guide the robotic arm in grasping or positioning operations), the measured point cloud coordinates need to be converted to the robotic arm's base coordinate system. The specific calibration steps are as follows: Calibration tool: Use a standard sphere with a known diameter and extremely high sphericity (e.g., a standard gauge ball with a diameter of Φ10mm) and rigidly connect it to the end effector of the robot arm.

[0049] Data Acquisition: A robotic arm is controlled to move a standard sphere to four or more different spatial positions within the measurement area. At each position, a laser profile scanner scans the standard sphere to obtain the three-dimensional coordinates (x, y, y) of its center in the sensor coordinate system. si ,y si ,z si Simultaneously, the theoretical pose (x, y) of the end effector in the robot's base coordinate system is read from the robot control system. ri ,y ri ,z ri ).

[0050] Coordinate transformation calculation: Using the acquired coordinates of multiple corresponding points, the rotation matrix R and translation vector T from the laser sensor coordinate system to the robot's base coordinate system are solved using the Singular Value Decomposition (SVD) algorithm. This transformation relationship satisfies the following formula: ; Calibration and Verification: After calibration, the robot arm is moved to a new verification position, and the coordinates of the sphere's center calculated by the laser sensor are compared with the theoretical coordinates of the robot arm. If the error between the two is within 0.05mm, the calibration is considered successful. This calibration method achieves precise mapping between the 3D point cloud data of the measurement system and the robot arm's motion coordinates, providing high-precision position information support for subsequent automated grasping and positioning.

[0051] The present invention also provides a high-precision object size measurement system based on multiple laser profile scanners, which mainly includes: N (N≥2, preferably 4 in this embodiment) high-precision laser profile scanners (denoted as S1, S2, S3, S4), a data acquisition synchronization control module, a central processing unit (usually an industrial control computer), a high-precision motion device (such as a conveyor belt system), and a PLC motion scanning control system.

[0052] Each laser contour scanner is fixedly installed around the measurement area at a preset angle, forming a surrounding layout of the target object. In this embodiment, the four laser contour scanners are respectively positioned above, below, to the left, and to the right of the object to be measured, with an angle of 90 degrees between adjacent scanners. Specifically, the upper scanner S1 is responsible for acquiring the upper edge contour of the object, the lower scanner S2 for acquiring the lower edge contour, the left scanner S3 for acquiring the left edge contour, and the right scanner S4 for acquiring the right edge contour. The laser emission direction of each scanner is directed towards the center of the measurement area, ensuring that when the object passes through the measurement area, each scanner can independently and completely capture the edge contour data of its corresponding side.

[0053] The data acquisition synchronization control module is connected to each laser contour scanner and sends synchronization trigger signals to all scanners to ensure that each scanner acquires data at the same time or based on the same spatial position reference. The central processing unit is connected to each laser contour scanner and the data acquisition synchronization control module and is responsible for running algorithms such as coordinate transformation, point cloud stitching, edge feature extraction, and size calculation.

[0054] The motion control system is integrated with each 3D contour scanner to scan the edges, while the data acquisition card simultaneously collects data and transmits it to the industrial control computer. After scanning, the central processing unit processes the collected point cloud data, locates the point cloud boundary, calculates the coordinate set of the boundary points, and calculates the length and width of the product according to different requirements.

[0055] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0056] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0057] In yet another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform any of the methods described above.

[0058] It is understood that the systems, devices, and storage media provided in the embodiments of the present invention correspond to the methods provided in the embodiments of the present invention, and the explanations, examples, and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

[0059] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state disk (SSD)).

[0060] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0061] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0062] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-precision object size measurement method based on a multi-laser contour scanner, characterized in that, Includes the following steps: Calibration steps: Using standard calibration objects, calibrate the spatial position transformation relationship between N laser profile scanners, where N≥2, and use the coordinate system of one of the scanners as the global reference coordinate system; Template creation steps: Using a standard template object, the complete point cloud data of the standard template object in the global reference coordinate system is collected by the N laser contour scanners, and the theoretical boundary position information of the standard template object is recorded; Acquisition and stitching steps: When the object to be measured enters the measurement area, the N laser contour scanners are simultaneously triggered to acquire local contour point clouds of different parts of the object to be measured, and the spatial position transformation relationship is used to transform all local contour point clouds to the global reference coordinate system and stitch them together to form the complete contour point cloud of the object to be measured. Deviation calculation steps: Extract the actual boundary position of the object to be measured from the complete contour point cloud, and calculate the pose deviation between the actual boundary position and the theoretical boundary position of the standard template object. The pose deviation includes translation deviation and / or rotation deviation. Size compensation step: Based on the extracted actual boundary position and combined with the pose deviation, the actual size of the object to be measured is calculated through dynamic compensation.

2. The high-precision object size measurement method based on a multi-laser contour scanner according to claim 1, characterized in that, In the calibration steps: The standard calibration object is a standard rectangular calibration plate, whose standard length and width dimensions are known. Each laser contour scanner extracts the edge contour coordinates of the standard rectangular calibration plate in its respective field of view, and fits the boundary line using the least squares method. Using the coordinate system of one of the laser profile scanners as the global reference coordinate system, and employing the fitted boundary lines, the transformation matrix T of the coordinate systems of the other laser profile scanners relative to the global reference coordinate system is calculated. ij The transformation matrix T ij It includes rotation component R and translation component t, used to describe the spatial transformation relationship between coordinate systems, satisfying: P global =T ij ·P local ; Among them, P local P represents the local point cloud coordinates. global These are the transformed global coordinates.

3. The high-precision object size measurement method based on a multi-laser contour scanner according to claim 1, characterized in that, In the deviation calculation step, the actual boundary position of the object to be measured is found and extracted from the complete contour point cloud using the AlphaShape method, specifically including: Perform Delaunay triangulation on the input point cloud data S; Traverse all triangles obtained from triangulation, and filter the edges that form an AlphaShape according to a preset parameter α. The filtering criterion is: if the radius of the circumcircle of the triangle is R... circum If ≤α, then all three sides of the triangle are candidate sides; Collect all edges that meet the conditions and remove duplicate edges to form the boundary contour of the object under test. The parameter α is used to control the compactness of the boundary. The smaller the value of α, the closer the boundary is to the original point cloud. The larger the value of α, the closer the boundary is to the convex hull.

4. The high-precision object size measurement method based on a multi-laser contour scanner according to claim 1, characterized in that, In the size compensation step, the actual width W of the object to be measured actual and actual length L actual The calculation uses a fine-tuning compensation formula based on template size: The width compensation formula is: W actual =W template +(ΔX right -ΔX left ); The length compensation formula is: L actual =L template +(ΔY bottom -ΔY top ); Among them, W template L is the width of the standard template object. template The length of the standard template object; ΔX left Let ΔX be the deviation of the left boundary of the object to be measured from the left boundary of the template on the X-axis. right This is the deviation of the right boundary of the object to be measured from the right boundary of the template on the X-axis. ΔY top ΔY is the deviation between the upper boundary of the object to be measured and the upper boundary of the template on the Y-axis. bottom This represents the deviation between the lower boundary of the object under test and the lower boundary of the template on the Y-axis.

5. The high-precision object size measurement method based on a multi-laser contour scanner according to claim 1, characterized in that, In the deviation calculation step, the rotational deviation θ is calculated in the following way: Edge lines in the complete contour point cloud are detected by Hough transform; Calculate the angle between the detected edge line and the corresponding edge line of the standard template object, and use this angle as the rotation angle deviation θ of the object under test; Furthermore, by calculating the actual geometric center of the object under test and the geometric center C of the standard template object... template The offset between the two values ​​is used to obtain the translation deviation.

6. The high-precision object size measurement method based on a multi-laser contour scanner according to claim 1, characterized in that, When the object under test is a workpiece made of a material with high light transmittance or high reflectivity, a point cloud preprocessing step based on material characteristics is included after the acquisition and stitching step and before the deviation calculation step: Set a filter window that matches the material properties to filter the original point cloud and suppress ambient light interference and high reflectivity noise; A threshold segmentation strategy that matches the material properties is used to separate the point cloud data of the target object from the background; The logic for eliminating chamfered interference points is executed. By utilizing the normal vector change characteristics or curvature characteristics of the point cloud at the chamfered edge, false contour points generated by the chamfered edge or curved surface of the object are identified and eliminated, while the true main body edge point cloud of the object is preserved.

7. A high-precision object size measurement system based on a multi-laser contour scanner, characterized in that, The system for performing the method according to any one of claims 1 to 6 comprises: N laser contour scanners, where N≥2, are fixedly installed around the measurement area to form a surrounding distribution layout of the object to be measured. Each laser contour scanner is used to collect local contour point cloud data of the corresponding side of the object to be measured. The data acquisition synchronization control module is connected to each of the N laser contour scanners and is used to send synchronization trigger signals to all laser contour scanners to ensure that each scanner acquires data at the same time or according to the same spatial position reference. A central processing unit is connected to the N laser contour scanners and the data acquisition synchronization control module, respectively. The central processing unit includes: The calibration module is used to calculate the transformation matrix of each laser profile scanner coordinate system relative to the global reference coordinate system using standard calibration objects; The template storage module is used to store the theoretical boundary position information and geometric center data of standard template objects in the global reference coordinate system; The point cloud stitching module, connected to the calibration module, is used to transform the local contour point clouds collected by each laser contour scanner to the global reference coordinate system in real time using the transformation matrix, and stitch them together to form a complete object contour point cloud. The pose calculation module is connected to the point cloud stitching module and the template storage module. It is used to extract the actual edge coordinates of the object under test from the complete object contour point cloud, and to calculate the translational deviation and rotational angle deviation between the object under test and the standard template object. The size compensation module, connected to the pose calculation module, is used to calculate and output the actual size of the object under test based on the actual edge coordinates and the translation and rotation angle deviations using a dynamic compensation formula.

8. The high-precision object size measurement system based on a multi-laser contour scanner according to claim 7, characterized in that, The number of N laser contour scanners is 4, which are respectively set above, below, left and right of the object to be measured. The included angle between adjacent scanners is 90 degrees, forming an independent full-coverage measurement architecture for the four edges of the object to be measured. The upper scanner is responsible for capturing the upper edge contour of the object under test, the lower scanner is responsible for capturing the lower edge contour, the left scanner is responsible for capturing the left edge contour, and the right scanner is responsible for capturing the right edge contour.

9. The high-precision object size measurement system based on a multi-laser contour scanner according to claim 7, characterized in that, The system also includes: A conveyor belt motion control system is used to carry and transport the object to be measured through the measurement areas of the N laser contour scanners. An encoder, connected to the conveyor belt motion control system and the data acquisition synchronization control module, is used to generate pulse signals according to the movement distance of the conveyor belt. The data acquisition synchronization control module, based on the pulse signal from the encoder, synchronously triggers the N laser contour scanners to acquire data when the object under test moves to a preset trigger position, thereby achieving online dynamic measurement synchronized with the movement of the conveyor belt.

10. The high-precision object size measurement system based on a multi-laser contour scanner according to claim 7, characterized in that, The central processing unit also includes a hand-eye calibration module for establishing the coordinate mapping relationship between the measurement system and the external robotic arm; The hand-eye calibration module is configured as follows: Obtain the laser sensor coordinates (x, y) of a standard sphere at multiple different spatial locations. si ,y si ,z si At the same time, the theoretical coordinates (x, y) of the end effector of the robot arm in the robot arm's base coordinate system are obtained. ri ,y ri ,z ri ); Using the collected coordinates of multiple corresponding points, the rotation matrix R and translation vector T from the laser sensor coordinate system to the robot's base coordinate system are solved using the singular value decomposition algorithm, satisfying the following relationship: ； The obtained rotation matrix R and translation vector T are output as calibration results, which are used to convert the point cloud coordinates of the measurement system into the motion coordinates of the robot, realizing robot guidance and positioning based on vision measurement.

Images