A system for automatically measuring the sectional profile of an object and method thereof

Abstract

A system and method for automatically measuring cross-sectional profiles of an object is disclosed. The system includes a single line laser for emitting a single laser line, at least one planar mirror for reflecting the laser line to at least one non-exposed surface of a target object to be measured, a camera for capturing distortions of the laser line projected onto the surface of the target object, and an image processing unit for processing the distortions captured by the camera to generate a cross-sectional profile of the target object. The system of the present invention improves accuracy, repeatability, and efficiency through automated measurement and can become a highly efficient tool for applications requiring precise cross-sectional profile analysis of irregular and inaccessible surfaces.

Description

(19) State Intellectual Property Office (12) Invention Patent Application (10) Application Publication Number (43) Application Publication Date (21) Application Number 202510185250.5 (22) Application Date 2025.02.19 (71) Applicant University of Hong Kong Address Pok Fu Lam Road, Hong Kong, China (72) Inventors Su Qiliang Li Chaobin Zhao Zhongxu (74) Patent Agency Shenzhen Runlin Chuangyou Patent Agency Co., Ltd. 44223 Patent Attorney Liu Li (51) Int.Cl. G01B 11 / 24 (2006.01) G06T 7 / 13 (2017.01) (54) Invention Title A System and Method for Automatically Measuring the Cross-sectional Profile of an Object (57) Abstract This invention discloses a system and method for automatically measuring the cross-sectional profile of an object. The system includes: a single-line laser for emitting a single laser line; at least one planar mirror for reflecting the laser line onto at least one non-exposed surface of a target object; a camera for capturing distortions of the laser line projected onto the surface of the target object; and an image processing unit for processing the distortions captured by the camera to generate a cross-sectional profile of the target object. This system, through automated measurement, improves accuracy, repeatability, and efficiency, and can serve as a highly efficient tool for applications requiring precise cross-sectional profile analysis of irregular and inaccessible surfaces. Claims (2 pages), Description (13 pages), Drawings (6 pages), CN 122611808 A, 2026.08.21, CN 1 22 61 18 08 A. 1. A system for automatically measuring the cross-sectional profile of an object, characterized in that it comprises: a single-line laser for emitting a single laser line; at least one plane mirror for reflecting the laser line onto at least one non-exposed surface of the target object; a camera for capturing distortions of the laser line projected onto the surface of the target object; and an image processing unit for processing the distortions captured by the camera to generate the cross-sectional profile of the target object. 2. The system as claimed in claim 1, characterized in that: the target object comprises three uneven surfaces, namely an exposed surface, a first non-exposed surface, and a second non-exposed surface; the planar reflector comprises a first planar reflector and a second planar reflector, the first planar reflector corresponding to the first non-exposed surface such that the first non-exposed surface forms a first virtual image in the first planar reflector, and the second planar reflector corresponding to the second non-exposed surface such that the second non-exposed surface forms a second virtual image in the second planar reflector; the laser line emitted by the single-line laser is: directly projected onto the exposed surface, projected onto the first non-exposed surface through the first planar reflector, and projected onto the second non-exposed surface through the second planar reflector;The camera is used to capture the exposed surface, the first virtual image, and the second virtual image to capture the distortion of the laser line projected onto the three uneven surfaces; the image processing unit is used to reconstruct the 3D contour of each surface according to the distortion, and superimpose the laser contours of the three uneven surfaces to generate the complete cross-sectional contour of the target object. 3. The system of claim 2, wherein the laser line emitted by the single-line laser is: directly projected onto the exposed surface to form a third laser arc on the exposed surface, projected onto the first non-exposed surface through the first plane mirror to form a first laser arc on the first virtual image, and projected onto the second non-exposed surface through the second plane mirror to form a second laser arc on the second virtual image; the camera is used to capture the first laser arc, the second laser arc, and the third laser arc; the image processing unit is used to superimpose the arc lengths of the three laser arcs to generate the complete cross-sectional contour of the target object. 4. The system of any one of claims 1-3, wherein the light source of the single-line laser is a laser diode. 5. The system of any one of claims 1-3, wherein the single-line laser and the camera are aligned along the same vertical axis. 6. The system according to any one of claims 1-3, characterized in that: the image processing unit comprises: a distortion correction module for correcting the distortion; an image segmentation module for segmenting the corrected image into laser strips; a strip center extraction module for extracting the sub-pixel centers of the laser strips; a 3D point cloud generation module for aggregating the sub-pixel centers of the laser strips to generate a 3D point cloud; and a cross-sectional contour reconstruction module for extracting the cross-sectional contour from the 3D point cloud. 7. The system according to any one of claims 1-3, characterized in that: the placement position of the plane mirror satisfies the following conditions: (Claims 1 / 2 page 2 CN 122611808 A) Wherein, the extension of the line connecting the center of the target object and the center of the single-line laser intersects the extension of the plane mirror at point P1, LM represents the distance between point P1 and the initial point M of the plane mirror, θ represents the angle between the extension of the connecting line and the extension of the plane mirror, and Dmax represents the maximum equivalent circle diameter of the target object. 8. A method for automatically measuring the cross-sectional profile of an object, characterized by comprising the following steps: S1, emitting a single laser line using a single-line laser; S2, reflecting the laser line onto at least one non-exposed surface of the target object to be measured via at least one planar mirror; S3, capturing the distortion of the laser line projected onto the surface of the target object using a camera; and S4, processing the distortion captured by the camera using an image processing unit to generate the cross-sectional profile of the target object.9. The method as described in claim 8, wherein step S4 comprises: S41, correcting the distortion; S42, segmenting the laser stripe in the corrected image; S43, extracting the sub-pixel centers of the laser stripe; S44, aggregating the sub-pixel centers of the laser stripe to generate a 3D point cloud; S45, extracting the cross-sectional contour from the 3D point cloud. 10. The method as described in claim 9, wherein in step S42, the laser stripe is segmented using a U-Net model; in step S43, the sub-pixel centers of the laser stripe are extracted using the Steger algorithm; and in step S45, the cross-sectional contour of the target object is reconstructed through 3D point cloud post-processing. Claims 2 / 2 Page 3 CN 122611808 A A System and Method for Automatically Measuring the Cross-sectional Contour of an Object Technical Field

[0001] This invention relates to a system and method for automatically measuring the cross-sectional contour of an object. Background Art

[0002] Accurate measurement of the cross-sectional contour of irregular objects is crucial. Existing cross-sectional measurement methods have the following main problems: 1. Measurement is performed using simple manual instruments such as calipers. Calipers are limited to measuring only the maximum and minimum dimensions of the profile, often depending on personal experience and lacking repeatability. Furthermore, caliper measurements cannot accurately predict the dimensions of the cross-sectional profile. 2. Existing 3D laser scanners available on the market are usually expensive, and the process of using these devices to build an accurate model is time-consuming. In addition, these methods usually require rotating the measuring device or the object being measured, which is extremely inconvenient, and in some situations, such as when measuring the cross-sectional profile of a tree (where the object may be fixed or cannot be rotated), the measurement process may not be possible. Therefore, existing measurement methods become challenging or even impossible when the cross-sectional dimensions are relatively small or when parts of the object's surface are not visible. For example, the cross-sectional profile of corroded steel bars after the removal of a concrete ceiling is difficult to measure because the back section of the corroded steel bars is difficult to access.

[0003] Existing instruments generally lack the ability to perform accurate measurements while maintaining low cost and convenience.

[0004] It should be noted that the information disclosed in the above background section is only used for understanding the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The present invention provides a system and method for automatically measuring the cross-sectional profile of an object, which has advantages such as low cost, convenience, and measurement accuracy.

[0006] The present invention adopts the following technical solution:

[0007] In a first aspect, a system for automatically measuring the cross-sectional profile of an object is provided, comprising: a single-line laser for emitting a single laser line; at least one plane mirror for reflecting the laser line onto at least one non-exposed surface of the target object to be measured; a camera for capturing the distortion of the laser line projected onto the surface of the target object; and an image.A processing unit is used to process the distortion captured by the camera and generate the cross-sectional profile of the target object.

[0008] In a second aspect, a method for automatically measuring the cross-sectional profile of an object is provided, comprising the following steps: S1, emitting a single laser line using a single-line laser; S2, reflecting the laser line onto at least one non-exposed surface of the target object to be measured by at least one plane mirror; S3, capturing the distortion of the laser line projected onto the surface of the target object using a camera; and S4, processing the distortion captured by the camera using an image processing unit to generate the cross-sectional profile of the target object.

[0009] The beneficial effects of the present invention include: The system of the present invention is a portable single-line laser scanning system based on the principle of laser triangulation. This system is specifically designed for the automated and accurate measurement of irregular cross-sectional profiles of objects with an error of less than 0.1 mm, and has the following advantages:

[0010] 1. Compared with calipers, the present invention can measure the cross-sectional profile dimensions of irregular objects with non-exposed surfaces with higher accuracy. The system of the present invention allows for comprehensive profile analysis without direct access to all surfaces of the object, thereby overcoming the limitations inherent in conventional caliper-based measurements. Instruction Manual 1 / 13 Page 4 CN 122611808 A

[0011] 2. Compared with commercially available 3D laser scanners, the present invention is more cost-effective, faster, and more convenient. Unlike traditional 3D laser scanning methods that typically require rotating the measuring device or object to capture complete cross-sectional data, the present invention eliminates the need for relative rotation between the object and the device by mounting a plane mirror on the non-exposed surface of the corresponding target object. The plane mirror reflects the laser onto the inaccessible non-exposed surface, eliminating the need for such rotation. The present invention facilitates the identification and measurement of cross-sectional profiles without the logical complexity and time delay associated with rotating the object or scanning device, thus making the measurement process of the present invention more efficient and easier to use.

[0012] 3. The present invention incorporates advanced machine learning algorithms into its image processing flow to accurately locate the object surface, extract the center of the laser strip, and minimize errors induced by image distortion. By utilizing these sophisticated computational techniques, the present invention ensures high accuracy in the measurement of irregular objects, thereby providing reliable and detailed cross-sectional profiles that surpass the capabilities of existing measurement solutions.

[0013] 4. The present invention demonstrates significant flexibility and effectiveness in non-perpendicular situations. Even when the object is not perpendicularly aligned with the laser plane, the present invention can generate accurate cross-sectional profiles.

[0014] 5. The system of the present invention is also simple, portable, flexible, and economical, providing an attractive solution for industries seeking reliable, easy-to-use, and cost-effective measurement technologies.

[0015] 6. The system of the present invention can capture the entire cross-sectional shape of an object without requiring its rotation.

[0016] 7. The system of the present invention can quickly calculate cross-sectional profiles without the need for coordinate measuring devices. Furthermore, the system of the present invention requires only a simple, low-power, non-contact, and lightweight setup, making it highly practical and ideal for autonomous robot inspection and next-generation intelligent condition assessment. The system of the present invention ensures easy integration into robot platforms and supports efficient, real-time monitoring and assessment of object conditions without the need for complex or energy-intensive equipment.

[0017] In summary, the system of the present invention improves accuracy, repeatability, and efficiency through automated measurement, making it an efficient tool for applications requiring precise cross-sectional profile analysis of irregular and inaccessible surfaces.

[0018] Figure 1a is a front view of the measurement scenario of the system for automatically measuring the cross-sectional profile of an object in an embodiment of the present invention;

[0019] Figure 1b is a side view of the measurement scenario of the system for automatically measuring the cross-sectional profile of an object in an embodiment of the present invention;

[0020] Figure 1c is a laser image captured by a digital camera in the measurement scenario of the system for automatically measuring the cross-sectional profile of an object in an embodiment of the present invention;

[0021] Figure 2 is a schematic diagram of the principle of the camera capturing a single laser arc on the surface of the target object in an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of the principle of cross-sectional profile measurement in an embodiment of the present invention;

[0023] Figure 4 is a flowchart of cross-sectional profile measurement in an embodiment of the present invention;

[0024] Figure 5 is a flowchart of strip center extraction in an embodiment of the present invention;

[0025] Figure 6 is a 3D measurement model when (a) the laser is directly projected onto the exposed surface of the object and (b) the laser is indirectly projected onto the non-exposed surface of the object in an embodiment of the present invention;

[0026] Figure 7 is a schematic diagram of the experimental setup of the laser scanning device in an embodiment of the present invention;

[0027] Figure 8 shows (a) an external perspective view and (b) an internal perspective view of the experimental setup of the laser scanning device in an embodiment of the present invention;

[0028] Figure 9 is a schematic diagram of using a standard cylinder to verify the experimental results in an embodiment of the present invention; Specification 2 / 13 pages 5 CN 122611808 A

[0029] Figure 10 is a schematic diagram of the placement parameters of the plane mirror in an embodiment of the present invention. Detailed Description

[0030] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention. In the absence of conflict, the embodiments and features in the embodiments of this application can be combined with each other.

[0031] Commercially available measurement technologies generally do not support automated measurement of the cross-sectional profile dimensions of non-exposed surfaces. When dealing with objects with inaccessible or hidden areas (such objects have non-exposed surfaces), conventional methods cannot capture comprehensive dimensional data of the entire cross-section. Therefore, the present invention provides a method for automatically measuring objectsA system for profiled cross-sections of a body includes:

[0032] a single-line laser for emitting a single laser line;

[0033] at least one plane mirror for reflecting the laser line onto at least one non-exposed surface of a target object;

[0034] a camera for capturing distortions of the laser line projected onto the surface of the target object; and

[0035] an image processing unit for processing the distortions captured by the camera to generate a cross-sectional profile of the target object.

[0036] In some embodiments, the target object includes three uneven surfaces: an exposed surface, a first non-exposed surface, and a second non-exposed surface; the planar reflector includes a first planar reflector and a second planar reflector, the first planar reflector corresponding to the first non-exposed surface such that the first non-exposed surface forms a first virtual image in the first planar reflector, and the second planar reflector corresponding to the second non-exposed surface such that the second non-exposed surface forms a second virtual image in the second planar reflector; the laser line emitted by the single-line laser is: directly projected onto the exposed surface, projected onto the first non-exposed surface through the first planar reflector, and projected onto the second non-exposed surface through the second planar reflector; the camera is used to capture the exposed surface, the first virtual image, and the second virtual image to capture the distortion of the laser line projected onto the three uneven surfaces; the image processing unit is used to reconstruct the 3D contour of each surface according to the distortion, and superimpose the laser contours of the three uneven surfaces to generate the complete cross-sectional contour of the target object.

[0037] Wherein, when observed from an angle outside the direct emission path of the single-line laser, the surface directly reachable by the laser is the exposed surface, and the surface that the laser cannot directly illuminate is the non-exposed surface. For example, the front of the target object to be measured is the exposed surface, and the rear surface is the non-exposed surface.

[0038] In some embodiments, the laser line emitted by the single-line laser is: directly projected onto the exposed surface to form a third laser arc on the exposed surface, projected onto the first non-exposed surface through the first plane mirror to form a first laser arc on the first virtual image, and projected onto the second non-exposed surface through the second plane mirror to form a second laser arc on the second virtual image; the camera is used to capture the first laser arc, the second laser arc, and the third laser arc; the image processing unit is used to superimpose the arc lengths of the three laser arcs to generate the complete cross-sectional profile of the target object.

[0039] In some embodiments, the light source of the single-line laser is a laser diode.

[0040] In some embodiments, the single-line laser and the camera are aligned along the same vertical axis.

[0041] In some embodiments, the image processing unit includes:

[0042] A distortion correction module is used to correct the distortion;

[0043] An image segmentation module is used to segment the laser stripe in the corrected image; Specification 3 / 13 Page 6 CN 122611808 A

[0044] A stripe center extraction module is used to extract the sub-pixel center of the laser stripe;

[0045] A 3D point cloud generation module is used to aggregate the sub-pixel centers of the laser stripe to generate a 3D point cloud;

[0046] A cross-sectional contour reconstruction module is used to extract the cross-sectional contour from the 3D point cloud.

[0047] In some embodiments, the placement position of the plane mirror satisfies the following conditions:

[0048]

[0049] Wherein, the extension of the line connecting the center of the target object and the center of the single-line laser intersects the extension of the plane mirror at point P1, LM represents the distance between point P1 and the initial point of the plane mirror M, θ represents the angle between the extension of the line connecting and the extension of the plane mirror, and Dmax represents the maximum equivalent circle diameter of the target object.

[0050] A specific embodiment of the present invention also provides a method for automatically measuring the cross-sectional contour of an object, comprising the following steps:

[0051] S1, emitting a single laser line using a single-line laser;

[0052] S2, reflecting the laser line onto at least one non-exposed surface of the target object to be measured using at least one plane mirror;

[0053] S3, capturing the distortion of the laser line projected onto the surface of the target object using a camera; and

[0054] S4, processing the distortion captured by the camera using an image processing unit to generate the cross-sectional contour of the target object.

[0055] In some embodiments, step S4 includes:

[0056] S41, correcting the distortion;

[0057] S42, segmenting the corrected image into laser strips;

[0058] S43, extracting the sub-pixel centers of the laser strips;

[0059] S44, aggregating the sub-pixel centers of the laser strips to generate a 3D point cloud;

[0060] S45, extracting the cross-sectional contour from the 3D point cloud.

[0061] In some embodiments, the laser strip is segmented using a U-Net model in step S42; the sub-pixel centers of the laser strip are extracted using the Steger algorithm in step S43; and the cross-sectional profile of the target object is reconstructed through 3D point cloud post-processing in step S45.

[0062] Reinforced concrete (RC) has always been the most commonly used structural material worldwide, and steel corrosion is the main cause of premature deterioration of RC structures. Due to the high costs of maintenance and replacement, accurate structural inspection and RC deterioration prediction are crucial for developing effective measures to eliminate corrosion problems and ensure optimal allocation of limited resources. The present invention will be described in detail below using the measurement of the cross-sectional profile of corroded steel bars after the removal of a concrete ceiling as an example.

[0063] 1. Measurement Principle

[0064] The system includes a laser scanning device and an image processing unit that provide measurement results through laser triangulation. The laser scanning device includes a single-line laser with a line laser diode 3 projecting a single laser line as the light source, a camera 4, and two plane mirrors (first plane mirror 1 and second plane mirror 2). As shown in Figures 1a-1c, considering the measurement of the cross-sectional profile of the ceiling steel reinforcement, the line laser diode 3 and the camera 4 are aligned along the same vertical axis, and the two plane mirrors 1 and 2 are positioned at the oblique rear side of the target object 5 (steel reinforcement in this example). The two plane mirrors placed in key positions are configured to facilitate the reflection and projection of the laser onto the non-exposed surface of the target object. During the scanning process, observed from an angle outside the direct emission path of the laser diode, the beam emitted by the laser diode directly illuminates the front surface (exposed surface) of the irregular object and is projected onto the two oblique rear surfaces (two non-exposed surfaces) through specular reflection, thereby causing the single laser line to be distorted across three uneven surfaces. A camera captures virtual images of the exposed surface and two non-exposed surfaces reflected in a plane mirror to capture the distortion of a single laser line on three uneven surfaces (in this example, the front of the target object and two non-exposed sides). The laser diode used in this embodiment is a Class I laser (power less than 0.4 MW) conforming to 21 CFR 1040.10 and 1040.11. For image processing, a U-Net model is used for laser strip segmentation. The sub-pixel centers of the detected laser strips are extracted using the Steger algorithm and then aggregated to generate a three-dimensional point cloud based on the image model (3D measurement model), from which the cross-sectional profile is extracted. This advanced scanning method enhances the accuracy and reliability of profile analysis for irregular objects, thus providing an economical and practical alternative to existing measurement tools.

[0065] Figures 2 and 3 illustrate the measurement principle, demonstrating the cross-sectional profile measurement using the laser scanning device described above. As shown in Figure 2, the process of the camera capturing the laser point Am1_r by positioning via a plane mirror involves the following steps in Table 1:

[0066] Table 1 Pseudocode for laser point positioning via a plane mirror

[0067]

[0068] Based on the above equations, the camera 4 directly captures the laser arc on the exposed surface (front in this example) of the target object 5, the arc length of which is expressed as arc length As_rBs_r. The laser is also projected via the plane mirror onto the non-exposed surface of the target object (two rear sides in this example, page 5 / 13 of this specification, CN 122611808 A), thereby generating arc lengths Am1_vBm1_v and Am2_r on the virtual image formed in the plane mirror.The arcs of vBm2_v are then captured by the camera. The laser reflects off the surface of the plane mirror and is projected onto the non-exposed surface of the target object, thus producing arcs with arc lengths Am1_rBm1_r and Am2_rBm2_r. Assuming the plane mirror is perpendicular to the laser plane, the arc length of the arc of the virtual image of the non-exposed surface in the plane mirror is equal to the arc length of the arc projected by the laser reflection. Specifically, the arc length Am1_vBm1_v = Am1_rBm1_r and the arc length Am2_vBm2_v = Am2_rBm2_r. The camera captures three arcs of different arc lengths: As_rBs_r, Am1_vBm1_v, and Am2_vBm2_v. By superimposing these arc lengths, the cross-sectional profile of the target object can be accurately obtained.

[0069] As shown in Figure 3, in this example, a single laser is projected onto point A on the surface of the steel bar (target object 5) and imaged onto point A' on the image sensor by the lens of the camera 4. A measurement reference plane is established on a vertical section centered on the reinforcing bar. For a point on the cross-sectional profile of the reinforcing bar, its relative position with respect to the reference plane causes the laser beam to project the position onto point B on the measurement reference plane. Therefore, the captured laser point on the image sensor shifts from A' to B'. Assuming the imaging plane is parallel to the measurement reference plane and there is no camera imaging distortion, the depth (h) of the reinforcing bar cross-section at point O can be calculated based on the triangulation angle between the semiconductor laser module and the camera. The depth of the reinforcing bar cross-section is derived as follows:

[0070] Considering the similarity between ΔOAB and ΔOCD, it can be verified that:

[0071]

[0072] where H is the measurement distance between the camera and the measurement reference plane, and L is the equivalent horizontal distance between the camera and the laser diode, which is determined by the following formula:

[0073]

[0074] where L0 is the center distance between the camera 4 and the laser diode 3, and α is the camera tilt angle. Similarly, considering the similarity between ΔCAB and ΔCA′B′, the deformation displacement (A′B′) at the image sensor is calculated as:

[0075]

[0076] where f is the focal length of the camera.

[0077] The number of offset pixels (nop) is calculated as:

[0078]

[0079] where r is the horizontal resolution of the image sensor and s is the horizontal size of the image sensor. By combining equations (1)-(4), we can obtain:

[0080]

[0081] Equation (5) shows that the depth of the laser-scanned section is expressed based on line distortion pixels, which can be calculated using image processing techniques. Therefore, the measurement sensitivity expressed by nop / f reflects the ability to detect minute height changes.

[0082] Compared with the section depth h, the lateral distance d of the section is directly observed and calculated. Along the projection onto the reinforcing barLaser lines, identifying multiple laser points {m0,m1,m2,…,mn}, where the lateral distance of each laser point is represented as {d0,d1, d2,…,dn}. The depth distance {h0,h1,h2,…,hn} of these laser points can be calculated using the above formula. When the laser points are sufficiently dense, the arc length of each segment can be approximated by the hypotenuse of the corresponding triangular segment.

[0083] The approximation of the cross-sectional length of the reinforcing bar is as follows:

[0084]

[0085] Wherein, since m0 and mn correspond to laser points located at both ends of the cross-section, d0 and hn are equal. Therefore, the laser length of the captured cross-section on the image sensor is transformed into:

[0086]

[0087] This approximation helps to accurately reconstruct the cross-sectional profile by summing the individual arc segments derived from the dense laser point extraction.

[0088] 2. Image Processing Flow

[0089] Image processing mainly involves extracting the center of the laser strip from the scanned image (i.e., the image captured by the camera). By integrating a 3D measurement model, a correspondence between the measured 3D laser points and their corresponding image pixels was established. Subsequently, the position of the planar reflector was determined to facilitate the reconstruction and overlay of the cross-sectional contour. Finally, the actual cross-sectional contour was identified. Figure 4 illustrates the image processing process, using aligned 3D laser lines to achieve accurate calculation and reconstruction of the cross-sectional contour. The measurement method first applies distortion correction to the laser-scanned image, utilizing distortion parameters obtained through a comprehensive camera calibration procedure. After image correction, a U-Net convolutional neural network is used in the image segmentation task to accurately identify individual laser strips. Next, the Steger algorithm (i.e., the established method for strip center extraction) is used to accurately determine the center pixel coordinates (u,v) of the laser strip. Subsequently, an image model is used to establish the correlation between the center pixel coordinates (u,v) and the corresponding spatial 3D point cloud coordinates (Xc,Yc,Zc). This correlation helps to accurately transform the pixel-based measurement results into real-world spatial coordinates. Based on the transformed 3D coordinates of the distortion-free laser points, the relative positional relationship between the camera and the measured surface is effectively defined. By integrating and overlaying multiple spatial 3D point clouds, the cross-sectional contour of the measured object was constructed. Then, a minimization algorithm was used to calculate the final cross-sectional contour, which optimized the alignment and accuracy of the contour by minimizing the residual error between the modeled contour and empirical data. The image processing algorithm of the present invention is further described below.

[0090] (1) Strip center extraction

[0091] Accurate extraction of the laser strip center coordinates is crucial for capturing the contour and size of the object's cross-section. In order to ensure efficient and robust extraction while reducing interference, a strip center extraction algorithm based on a deep convolutional neural network (CNN) was designed. The flowchart of the strip center extraction algorithm is depicted in Figure 5.

[0092] (2) 3D measurement model

[0093] The image model establishes the relationship between the point P(Xc,Yc,Zc) of the laser strip in the light plane and its corresponding image pixel coordinates P′(u,v) (shown in Figure 6(a)). This relationship is defined by using geometric and trigonometric correlations. As mentioned above, for the non-exposed surfaces on both sides, the center pixel coordinates (u,v) are related to the spatial 3D coordinates (Xv,Yv,Zv) of the image of the object on the surface of the plane mirror. Using the same method, a separate measurement reference plane was established at the location of the plane mirror, thereby defining the relative positional relationship between the camera and the plane mirror surface:

[0094]

[0095] The spatial 3D coordinates (Xv, Yv, Zv) of the virtual image along the normal direction of the plane mirror can be correlated with the actual spatial 3D coordinates (Xc, Yc, Zc), which are calculated as follows:

[0096]

[0097] This process establishes the correspondence between (Xc, Yc, Zc) and P′(u, v) (shown in Figure 6(b)). A pinhole camera model is used to establish the positional association between the measurement point P and its image pixel P′. This model facilitates the accurate mapping of 3D spatial coordinates to their corresponding 2D image pixel coordinates, thereby ensuring the accurate reconstruction of the cross-sectional shape.

[0098] The positional association can be expressed as:

[0099]

[0100] where s represents a random scaling factor, and A is a 3×3 matrix of the camera's internal parameters, which includes the effective focal length, camera sensor size, and principal point. Equation (8) shows that any spatial point P along line Oc can be projected onto the same pixel position (P′) in the image. Furthermore, the 3D coordinates of the measured point P satisfy the equation of the laser plane πP, which is expressed as:

[0101]

[0102] where Ai, Bi, Ci and Di are physical coefficients describing the single laser plane equation. Considering that the camera model is an impractical linear imaging model, practical applications need to consider the nonlinear distortion caused by manufacturing and installation errors. The relationship between the ideal laser point P′u(xu,yu) and the captured laser point P′d(xd,yd) in the image coordinate system can be expressed as: Specification 8 / 13 pages 11 CN 122611808 A

[0103]

[0104] where the coefficients k1, k2, k3 correspond to the radial distortion of the camera, while p1, p2 represent the tangential distortion coefficients. The three laser strips are converted into point clouds using the established image model. The cross-sectional profile can be generated through point cloud post-processing (such as point cloud denoising, point cloud overlay, etc.).

[0105] 3. Measurement Experiment

[0106] The system was implemented and tested in laboratory and field environments. The experimental setup is shown in Figures 7 and 8. Specifically, Figure 8(a) presents an external view of the device, while Figure 8(b) shows an internal perspective view captured by internal imaging, andTable 2 lists the detailed specifications for each component of the experimental setup. The system can be adapted to meet the requirements of building scanning, enabling accurate measurements within a range of 10 mm to 2 m. The setup is equipped with different measurement ranges, allowing different target objects within each range to have their cross-sectional shapes measured using the corresponding measuring devices. This modular configuration ensures versatility and accuracy in capturing the geometric contours of various structural elements, thereby improving the overall efficiency and reliability of the scanning process. Table 3 details each measurement range, along with the corresponding object dimensions applicable to that range. The 3D measurement procedure was implemented using Python running on a computer with an Intel i7-6700, 3.40 GHz CPU and 16 GB of RAM. To ensure the accuracy and reliability of the experimental results, a standard cylinder (20 mm in diameter) was used for verification purposes, as shown in Figure 9. This standard cylinder served as a reference to verify the measurement and calculation predictions derived from the experimental setup. The implementation process first involved calibrating the camera, mirror, and laser plane, followed by placing the target object to be measured into the laser scanning device. A laser diode emits a beam that directly illuminates the front surface (exposed surface) of a cylinder and is projected onto two tilted rear surfaces (two unexposed surfaces) via specular reflection. A camera captures virtual images of the exposed surface and the two unexposed surfaces reflected in a plane mirror to capture the distortion of a single laser line across the three uneven surfaces. The images captured by the camera are processed using the RLL algorithm to reconstruct the point cloud of the three laser segments (specifically, this includes correcting the distortion; segmenting the laser strips in the corrected image using a U-Net model; extracting the sub-pixel centers of the laser strips using the Steger algorithm; and aggregating the sub-pixel centers of the laser strips to generate a 3D point cloud). Subsequently, post-processing of the point cloud yields the fitted cylindrical contour information (e.g., extracting the cross-sectional contour from the 3D point cloud through 3D point cloud denoising, overlaying 3D point clouds, etc., to reconstruct the cross-sectional contour of the target object). The diameter of the equivalent circle obtained by fitting the arc segment data points is 20.07 mm, which has an error of less than 0.1 mm compared to the actual diameter of the standard cylinder (20 mm), demonstrating the measurement accuracy of this method.

[0107] Table 2 Hardware Instrument Parameter Specifications 9 / 13 Page 12 CN 122611808 A

[0108]

[0109]

[0110] Table 3 Measurement Range and Corresponding Object Specifications 10 / 13 Page 13 CN 122611808 A

[0111]

[0112] As shown in Figure 10, it is necessary to define the positions of the intersection points P1 and P2 along the extension line of the plane mirror, especially the distance from these points to the center of the target object. The distance between point P1 and the center O of the target object is G. The center O of the target object andThe distance between points P2 is Q. Additionally, the angle θ between the plane mirror and line segment P1P2 needs to be determined; this angle θ is important for accurate profile measurement. Before initiating the measurement process, the operator must perform an initial assessment to approximate the equivalent diameter range category of the target object (a total of 6 categories). Standardization criteria have been established whereby the laser must illuminate at least 75% of the entire cross-section to enable the use of algorithms to accurately determine the cross-sectional profile. Under this standard (i.e., the laser illuminates at least 75% of the entire cross-section), permissible ranges of plane mirror angles corresponding to each specific measurement range have been determined, as detailed in Table 4.

[0113] Table 4 Measurement Ranges and Corresponding Plane Mirror Angle Ranges Specification 11 / 13 pages 14 CN 122611808 A

[0114]

[0115] The slash "\" in Table 4 indicates that there is no angle range that meets the standard (the laser illuminates at least 75% of the entire cross-section) at the corresponding Q and G values.

[0116] In order to prevent the target object from being projected multiple times between the plane mirrors, thereby adversely affecting image recognition, and to ensure that the target object is smoothly inserted into the laser scanning device, the positioning of the initial point of the plane mirror is extremely important. Specifically, after determining the appropriate plane mirror angle, the initial position of the plane mirror satisfies the following conditions:

[0117]

[0118] Wherein, the extension of the line connecting the center of the target object and the center of the single-line laser intersects the extension of the plane mirror at point P1, LM represents the distance between point P1 and the initial point M0 of the plane mirror M (the initial point M0 is on the end of the plane mirror M closer to point P1, that is, LM represents the distance between point P1 and the end of the plane mirror M closer to point P1), θ represents the angle between the extension of the line connecting the two sides and the extension of the plane mirror, and Dmax represents the maximum equivalent circle diameter of the target object.

[0119] The system proposed in the embodiments of the present invention has the following advantages:

[0120] (1) High efficiency and novelty. Current scanners are limited to capturing the external dimensions of exposed surfaces. The system proposed in this invention overcomes this limitation by reflecting the laser onto a planar mirror on a non-exposed surface, thereby enabling the measurement of the external dimensions of non-exposed surfaces. In addition, existing 3D scanning instruments cannot effectively reconstruct the dimensional parameters of a target object through specular reflection. The method of this invention provides significant advantages in reflection measurement based on a planar mirror, thereby enhancing the ability to capture the full surface profile. Specification 12 / 13 pages 15 CN 122611808 A

[0121] (2) Convenient and flexible. By utilizing a single laser line instead of multiple laser lines, the device has a simple and compact structure. The streamlined design facilitates handheld scanning and can be used with a handheld positioning rod, wherein the handheld positioning rodThe length is adjustable to allow for flexible measurement of various object contours (e.g., steel bars embedded in ceiling concrete, irregularly shaped cups, or even tree trunks).

[0122] (3) Robust, safe, and cost-effective. The algorithm integrates machine learning into the image processing flow to accurately locate the surface of an object, extract the center of the laser strip, and minimize errors caused by image distortion. Compared to pure vision systems such as stereo vision systems, the proposed algorithm demonstrates superior ability to handle textureless surfaces. The components used in the system, including a consumer-grade digital camera and a line laser, are both affordable and easily accessible. Furthermore, the system ensures safety by employing a Class I laser that is completely harmless to the human body. The method of the present invention presents a more cost-effective alternative to high-precision 3D laser scanners, thereby enabling advanced scanning technology to be used in a wider range of applications.

[0123] The above description is a further detailed description of the present invention in conjunction with specific / preferred embodiments, and it should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various substitutions or modifications can be made to the described embodiments without departing from the inventive concept, and all such substitutions or modifications should be considered within the scope of protection of this invention. In the description of this specification, the terms "an embodiment," "some embodiments," "preferred embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions, and modifications can be made herein without departing from the scope of protection of the patent application. Instruction manual, page 13 / 13, 16 CN 122611808 A, Figure 1a, Figure 1b; Instruction manual drawings, page 1 / 6, 17 CN 122611808 A, Figure 1c, Figure 2; Instruction manual drawings, page 2 / 6, 18 CN 122611808 A, Figure 3, Figure 4, Figure 5; Instruction manual drawings, page 3 / 6, 19 CN 122611808 A, Figure 6, Figure 7; Instruction manual drawings, page 4 / 6, 20 CN 122611808 A, Figure 8, Figure 9; Instruction manual drawings, page 5 / 6, 21CN 122611808 A Figure 10 Description of the Description Attached Drawings Page 6 / 6 22 CN 122611808 A Abstract The present application discloses a system for automatically measuring the sectional profile of an object and a method therefor. The system comprises: a single line laser configured to emit a single laser line; at least one plane mirror configured to reflect the laser line onto at least one non-exposed surface of a target object to be measured; a camera configured to capture distortions of the laser line projected onto the surfaces of the target object; and an image processing unit configured to process the distortions captured by the camera and generate the sectional profile of the target object. By automating the measurement process, the system of the present application improves accuracy, repeatability and efficiency, making it an effective tool for applications that require precise sectional profiling of irregular and inaccessible surfaces.

Claims

1. A system for automatically measuring the cross-sectional profile of an object, characterized in that, include: A single-line laser is used to emit a single laser line. At least one planar reflector is used to reflect the laser line onto at least one non-exposed surface of the target object to be measured; A camera used to capture the distortion of a laser line projected onto the surface of the target object; as well as An image processing unit is used to process the distortion captured by the camera and generate the cross-sectional profile of the target object.

2. The system as described in claim 1, characterized in that: The target object includes three uneven surfaces: an exposed surface, a first non-exposed surface, and a second non-exposed surface. The planar reflector includes a first planar reflector and a second planar reflector, the first planar reflector corresponding to the first non-exposed surface such that the first non-exposed surface forms a first virtual image in the first planar reflector, and the second planar reflector corresponding to the second non-exposed surface such that the second non-exposed surface forms a second virtual image in the second planar reflector; The laser line emitted by the single-line laser is projected directly onto the exposed surface, projected onto the first non-exposed surface through the first planar reflector, and projected onto the second non-exposed surface through the second planar reflector. The camera is used to capture images of the exposed surface, the first virtual image, and the second virtual image to capture the distortion of the laser lines projected onto the three uneven surfaces; The image processing unit is used to reconstruct the 3D contour of each surface based on the distortion, and superimpose the laser contours of the three uneven surfaces to generate the complete cross-sectional contour of the target object.

3. The system as described in claim 2, characterized in that: The laser line emitted by the single-line laser is: directly projected onto the exposed surface to form a third laser arc on the exposed surface; projected onto the first non-exposed surface through the first plane mirror to form a first laser arc on the first virtual image; and projected onto the second non-exposed surface through the second plane mirror to form a second laser arc on the second virtual image; the camera is used to capture the first laser arc, the second laser arc, and the third laser arc; the image processing unit is used to superimpose the arc lengths of the three laser arcs to generate the complete cross-sectional contour of the target object.

4. The system according to any one of claims 1-3, characterized in that: The light source of the single-line laser is a laser diode.

5. The system according to any one of claims 1-3, characterized in that: The single-line laser and the camera are aligned along the same vertical axis.

6. The system according to any one of claims 1-3, characterized in that: The image processing unit includes: A distortion correction module is used to correct the distortion. The image segmentation module is used to segment laser stripes into the corrected image; The strip center extraction module is used to extract the sub-pixel centers of the laser strip; The 3D point cloud generation module is used to aggregate the sub-pixel centers of the laser strip to generate a 3D point cloud. The cross-sectional contour reconstruction module is used to extract the cross-sectional contour from the 3D point cloud.

7. The system according to any one of claims 1-3, characterized in that: The placement of the plane mirror satisfies the following conditions: The extension of the line connecting the center of the target object and the center of the single-line laser intersects the extension of the plane mirror at point P1. M θ represents the distance between point P1 and the initial point of the plane mirror M, and θ represents the angle between the extension of the line connecting them and the extension of the plane mirror. max This represents the maximum equivalent circle diameter of the target object.

8. A method for automatically measuring the cross-sectional profile of an object, characterized in that, The steps include the following: S1. Use a single-line laser to emit a single laser line; S2. The laser line is reflected onto at least one non-exposed surface of the target object by at least one plane mirror; S3. Use a camera to capture the distortion of the laser line projected onto the surface of the target object; as well as S4. The distortion captured by the camera is processed by the image processing unit to generate the cross-sectional profile of the target object.

9. The method as described in claim 8, characterized in that, Step S4 includes: S41. Correct the distortion; S42. Segment the laser stripe in the corrected image; S43. Extract the sub-pixel center of the laser strip; S44. Aggregate the sub-pixel centers of the laser strips to generate a 3D point cloud; S45. Extract the cross-sectional profile from the 3D point cloud.

10. The method as described in claim 9, characterized in that, In step S42, the U-Net model is used to segment the laser strip; in step S43, the Steger algorithm is used to extract the sub-pixel centers of the laser strip; in step S45, the cross-sectional contour of the target object is reconstructed through 3D point cloud post-processing.

Images