Multi-frame three-dimensional point cloud splicing measurement method and device for high-temperature environment profile steel production and medium

Abstract

The invention belongs to the technical field of industrial measurement, and particularly discloses a multi-frame three-dimensional point cloud splicing measurement method and device for high-temperature environment profile steel production and a medium. The method disclosed by the invention comprises the following steps: firstly, acquiring three-dimensional point cloud data of the profile steel surface through a line laser scanner; then, denoising and normalization preprocessing operation is carried out on the three-dimensional point cloud data; thirdly, splicing the three-dimensional point cloud data by using an improved ICP (Inductively Coupled Plasma) algorithm, and carrying out Laplace smoothing and normal vector homogenization processing on the spliced three-dimensional point cloud data; thirdly, performing profile steel edge fitting on the spliced data; and finally, calculating the geometric dimension of the profile steel after edge fitting. According to the invention, the precision of point cloud splicing and the stability of a measurement result are greatly improved. The method can be widely applied to dimension measurement of the profile steel in the high-temperature environment.

Description

Technical Field

[0001] The invention belongs to the technical field of industrial measurement, and specifically relates to a multi-frame three-dimensional point cloud splicing measurement method, equipment and medium for high-temperature environment steel production. Background Art

[0002] With the advancement of Industry 4.0, the application of intelligent manufacturing in the steel industry is becoming more and more extensive, and the requirements for product quality in the production process are constantly increasing. Steel sections, especially H-section steel, are widely used in large structures such as buildings and bridges, and their dimensional accuracy is crucial to the safety and reliability of the overall structure.

[0003] The traditional method of measuring the profile size of steel mainly relies on manual operation, using measuring tools such as measuring rods, rulers and calipers for measurement. However, the manual measurement method has many limitations, especially in high-temperature production environments, where the accuracy and efficiency of manual operation are greatly limited.

[0004] In order to meet these challenges, in recent years, 3D non-contact profile dimension measurement technology based on point cloud information has received widespread attention. This technology obtains 3D point cloud data of steel sections and uses advanced algorithms for processing and analysis to achieve high-precision and high-efficiency dimension measurement. However, the existing 3D point cloud measurement technology still faces many technical difficulties in the production process of steel sections under high temperature environments, such as insufficient splicing accuracy of point cloud data and poor measurement stability. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production. The method introduces multiple optimizations to the ICP algorithm, greatly improving the accuracy of point cloud splicing, thereby improving the accuracy of contour dimension measurement of steel under high-temperature environment production conditions, and also improving the stability of measurement results;

[0006] The second object of the present invention is to provide a terminal device, which can realize a multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production when executing its own program;

[0007] The third object of the present invention is to provide a computer-readable storage medium for storing a corresponding computer program for a multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] A multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production includes the following steps performed in sequence:

[0010] S1, obtaining three-dimensional point cloud data of the steel surface through a line laser scanner;

[0011] S2, performing preprocessing operations of denoising and normalizing the three-dimensional point cloud data;

[0012] S3, using the improved ICP algorithm to stitch the 3D point cloud data;

[0013] S4, performing Laplace smoothing and normal vector homogenization processing on the spliced ​​three-dimensional point cloud data;

[0014] S5, performing edge fitting of the steel section on the data processed in step S4;

[0015] S6. Calculate the geometric dimensions of the steel section after edge fitting;

[0016] S7. End.

[0017] As a limitation, the improved ICP algorithm is run in the following order:

[0018] S31, before the ICP algorithm iteration, feature descriptor matching is introduced to calculate the local geometric features of each point in the point cloud to form a feature descriptor, and point pair matching is preferentially performed based on the similarity of the feature descriptors;

[0019] S32. During the iteration of the ICP algorithm, a weight factor is introduced for each point to affect the matching between point pairs, and an adaptive threshold control mechanism is introduced to dynamically adjust the distance threshold and the normal vector consistency threshold according to the current point pair error distribution;

[0020] S33, after the ICP algorithm is iterated, a candidate rigid body transformation model is calculated by the RANSAC algorithm, and the distance between the point pairs is estimated based on the least squares error. When the error is less than a preset value, the point pair is counted into the internal point set;

[0021] S34, repeating steps S31 to S33 until the iteration is completed, selecting the model with the most inliers in the inlier set, and calculating the objective function that maximizes the number of inliers.

[0022] As a limitation on feature descriptor matching, the calculation formula for feature descriptor matching is:

[0023]

[0024] In the formula, similarity(p i ,q j ) is point p i and dot q j The feature descriptor matching, f(p i ) and f(q j ) are points p i and dot q jThe feature descriptor of i )-f(q j )|| is f(p i ) and f(q j ), p i and q i are points in the source point cloud and target point cloud respectively, and λ is the standard deviation of feature distance.

[0025] As a limitation on the introduction of the weight factor, the calculation formula for introducing the weight factor includes:

[0026]

[0027] Where E'(R,t) is the objective function of the iterative process, R is the rotation matrix, t is the translation vector, is the weight factor of the i-th point, ‖p i -(R*q i +t)‖ is point p i and (R*q i +t), ‖p i -q i ‖ is the point p i and dot q i The Euclidean distance between them, σ is the standard deviation of the distance, θ i For point p i and dot q i The angle between the normal vectors of .

[0028] As a limitation of the adaptive threshold control mechanism, the adaptive threshold calculation formula is:

[0029] Thershold=μ+K·σ1

[0030] Where Thershold is the adaptive threshold, μ is the mean of the point pair error in the current iteration, σ1 is the standard deviation of the point pair error in the current iteration, and K is the adjustment coefficient.

[0031] As a further limitation, the objective function for maximizing the number of inliers is calculated as:

[0032] max|I|={i:‖Rp i +tq i ‖<∈}

[0033] Where ∈ is the preset value.

[0034] As a second limitation, step S5 is performed in the following order:

[0035] S51, for each point on the point cloud data, calculate the number of points whose distance from this point is less than a set radius, and define the points whose number is less than the set value as edge points;

[0036] S52, using a density clustering algorithm to group edge points according to the distance and density relationship between the points;

[0037] S53, calculating the center position of each group of edge points as the set center of the group;

[0038] S54, using the center of the set as a reference, fitting the plane of the steel section using the least square method;

[0039] S55, applying a moving average filter to smooth the fitted plane.

[0040] As a third limitation, the geometric dimensions of the steel section include flange width, web height and flange thickness.

[0041] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production is implemented.

[0042] A computer-readable storage medium stores a computer program, which, when executed by a processor, is used to implement the above-mentioned multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production.

[0043] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0044] (1) The present invention can efficiently and accurately complete the size measurement of steel sections under high temperature production conditions, ensuring quality control during the steel production process;

[0045] (2) The present invention introduces feature descriptor matching into the ICP algorithm, which can filter out a large number of noise points and abnormal matches before the ICP algorithm iteration, thereby improving the accuracy of point pair matching;

[0046] (3) The present invention introduces a weight factor into the ICP algorithm, which reduces the influence of points with large distance or normal vector deviation on the overall stitching result, and effectively improves the robustness of the algorithm in a noisy environment;

[0047] (4) The present invention introduces an adaptive threshold control mechanism into the ICP algorithm, which can maintain high stitching accuracy under different point cloud qualities and noise levels;

[0048] (5) The present invention achieves accurate and stable three-dimensional point cloud stitching effect by seeking the objective function of maximizing the number of internal points, reduces the cumulative error, and improves the robustness and adaptability of the algorithm;

[0049] (6) The present invention reduces the irregularity of the splicing area and improves the overall smoothness of the splicing result by performing Laplace smoothing and normal vector homogenization processing on the splicing transition area;

[0050] (7) The present invention calculates the center of the set of edge points to perform the least squares method to fit the plane, which greatly improves the accuracy of plane fitting;

[0051] (8) The present invention uses a moving average filter to smooth the fitted edge surface data, thereby eliminating possible noise and irregularities and improving the flatness of the steel surface.

[0052] The invention belongs to the technical field of industrial measurement. The method can improve the accuracy of point cloud splicing, thereby improving the accuracy of contour dimension measurement of steel sections under high-temperature production conditions, and also improving the stability of measurement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0054] In the attached picture:

[0055] Figure 1 This is a processing flow chart of Embodiment 1 of the present invention;

[0056] Figure 2 This is a schematic diagram of a medium-sized steel point cloud acquisition device according to Embodiment 1 of the present invention;

[0057] Figure 3 A schematic diagram of the scanning position of the online laser scanner of the point cloud collector in Example 1 of the present invention;

[0058] Figure 4 A point cloud collection diagram scanned by a line laser scanner in Example 1 of the present invention;

[0059] Figure 5 This is a mosaic result diagram of point cloud data in Example 1 of the present invention;

[0060] Figure 6 This is a graph showing the edge fitting results of the medium steel point cloud data in Example 1 of the present invention;

[0061] Figure 7 This is a diagram showing the overall edge fitting effect of the medium-shaped steel in Example 1 of the present invention.

[0062] In the figure: 1. shock absorption device, 2. point cloud collector, 3. line laser scanner. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] Example 1 A multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production

[0065] This embodiment takes H-beam as an example for measurement. Figure 1 As shown, this embodiment includes the following steps performed in sequence:

[0066] S1. Acquire three-dimensional point cloud data of the steel surface through the line laser scanner 3.

[0067] Due to the high temperature and steel movement in the H-beam production environment, the line laser scanner 3 has strong anti-interference ability and high precision advantages, and is very suitable for obtaining steel data in a high temperature environment.

[0068] like Figure 2 As shown, to ensure that the acquired data is comprehensive and accurate, the steel point cloud acquisition device of this embodiment is composed of two damping devices 1 and a point cloud collector 2. The point cloud collector 2 is located between the two damping devices 1. The damping device 1 is used to reduce the vibration of the steel to improve the accuracy of data acquisition. Figure 3 As shown, the point cloud collector 2 uses four line laser scanners 3 to scan the steel from different angles at the same time, and the point cloud data of the four positions of the steel are obtained as shown in FIG. Figure 4 shown.

[0069] S2. Perform denoising and normalization preprocessing operations on the three-dimensional point cloud data.

[0070] After collecting the point cloud data of the H-beam, the filtering algorithm is used to process the point cloud data to remove isolated points or noise points and complete the denoising operation; then the coordinate system is adjusted to align the point cloud data from different perspectives and unify them into the same coordinate system to complete the normalization process.

[0071] S3. Use the improved ICP algorithm to stitch the three-dimensional point cloud data.

[0072] After acquiring the point cloud data, the improved ICP algorithm is used to stitch the three-dimensional point cloud data.

[0073] The ICP algorithm is a general point cloud matching algorithm. In the case of high noise and data missing, directly using the point-to-point distance as the matching criterion may lead to incorrect pairing. In order to enhance the reliability of matching, feature descriptor matching is introduced before the ICP algorithm iteration. The local geometric features of each point in the point cloud are calculated to form a feature descriptor, and point pair matching is preferentially performed based on the similarity of the feature descriptors.

[0074] The calculation formula for feature descriptor matching is:

[0075]

[0076] In the formula, similarity(p i ,q j ) is point p i and dot q j The feature descriptor matching, f(p i ) and f(q j ) are points p i and dot q j The feature descriptor of ||f(p i )-f(q j )|| is f(p i ) and f(q j ), p i and q i are points in the source point cloud and target point cloud respectively, and λ is the standard deviation of feature distance.

[0077] In the traditional ICP algorithm, the matching of each point pair is treated equally, which is easily affected by noise points and abnormal points in the complex environment of high temperature and retrograde steel production. In order to enhance the robustness of the algorithm to abnormal points, this embodiment introduces a weight factor in the iterative process of the ICP algorithm to adjust the contribution of each point pair in the objective function, thereby affecting the matching between point pairs.

[0078] After the weight factor is introduced, the objective function of ICP and the calculation formula of the weight factor include:

[0079]

[0080]

[0081] Where E'(R,t) is the objective function of the iterative process, R is the rotation matrix, t is the translation vector, ω i is the weight factor of the i-th point, ‖p i -(R*q i +t)‖ is p i and (R*q i +t), ‖p i -q i ‖ is the point p i and dot q i The Euclidean distance between them, σ is the standard deviation of the distance, θ i For point p i and dot q i The angle between the normal vectors of .

[0082] The design of the weight factor gives smaller weights to point pairs with larger distances or normal vector deviations, thereby reducing the impact of these point pairs on the overall stitching result and effectively improving the robustness of the algorithm in noisy environments.

[0083] In a complex production environment, the distribution of noise and abnormal points is uncertain. In order to improve the adaptability of the algorithm, this embodiment introduces an adaptive threshold control mechanism in the ICP algorithm iteration process to dynamically adjust the distance threshold and normal vector consistency threshold according to the current point pair error distribution. The adjustment of the adaptive threshold enables the algorithm to maintain a high splicing accuracy under different point cloud qualities and noise levels.

[0084] The calculation formula of the adaptive threshold is:

[0085] Thershold=μ+K·σ1 (4)

[0086] Where Thershold is the adaptive threshold, μ is the mean of the point pair error in the current iteration, σ1 is the standard deviation of the point pair error in the current iteration, and K is the adjustment coefficient.

[0087] In order to further improve the stitching accuracy, after the ICP algorithm iteration, a candidate rigid body transformation model is first calculated by the RANSAC algorithm, that is, the optimal rotation matrix and translation vector are obtained, and the minimum value of the sum of the loss functions of the subsets of different points is calculated:

[0088]

[0089] Where k is the number of sampled point pairs and ρ(·) is the Huber loss function.

[0090] After calculating the candidate rigid body transformation model, the distance between point pairs is estimated based on the least squares error. When the error is less than a preset value, the point pair is counted into the internal point set.

[0091] The ICP algorithm requires continuous iteration to minimize the error between the two sets of point clouds. Repeat the above algorithm iteration until the iteration is completed. After the iteration is completed, the model with the most inliers is selected from the inlier set, and the objective function that maximizes the number of inliers is calculated. In this way, noise points and incorrect matching points can be effectively eliminated, making the spliced ​​point cloud more consistent with the real steel surface.

[0092] The objective function to maximize the number of inliers is calculated as:

[0093] max|I|={i:‖Rp i +tq i ‖<∈} (6)

[0094] Where ∈ is the preset value.

[0095] S4. Perform Laplace smoothing and normal vector homogenization processing on the spliced ​​3D point cloud data.

[0096] In order to avoid obvious discontinuity or breakage in the splicing transition area, smooth transition processing is added after splicing. By performing Laplace smoothing and normal vector homogenization on the splicing transition area, the irregularity of the splicing area is reduced and the overall smoothness of the splicing result is improved. The splicing result after processing is shown in the figure below. Figure 5 shown.

[0097] The Laplace smoothing formula is:

[0098]

[0099] In the formula, v i is the position of point i, v j is the position of point j, N(i) is the neighbor point set of point i, and α is the smoothing coefficient.

[0100] The normal vector homogenization formula is:

[0101]

[0102] In the formula, is the normal vector after homogenization, n j is the normal vector of point j, and u is the dimension of the space where the normal vector is located.

[0103] S5, performing edge fitting of the steel section on the data processed in step S4.

[0104] In order to measure more accurately, it is very important to obtain the edge profile of the steel section. Especially in high temperature environments, this edge information directly affects the accuracy of dimensional measurement and quality control.

[0105] First, for each point p on the point cloud data i , define a fixed radius r, calculate the distance point p i The number of points N that are smaller than the set radius r i , the number N i Points smaller than the set value T are defined as edge points.

[0106]

[0107] In the formula, δ(·) is the indicator function, p j is a point in the point cloud.

[0108] Next, the density clustering algorithm is used to group the edge points according to the distance and density relationship between the points, so as to identify different edge faces. Then, for each group of edge points, the average value on the coordinate axis is calculated as the center position of each group of edge points, which is the collection center of the group.

[0109] The calculation formula for the center position of the edge point is:

[0110]

[0111] In the formula, C n is the center position of the nth edge surface, G n is the nth group of edge points, |G n | is the number of edge points in the nth group, x i ,y i and z i For point p i The coordinates of .

[0112] Then, taking the center of the set as a reference, the plane of the steel section is fitted using the least squares method to minimize the sum of the distances from all edge points to the plane.

[0113] The error function calculation formula for plane fitting is:

[0114]

[0115] Where a, b, and c are the components of the plane normal vector, d is the intercept of the plane, and E is the error function, which represents the deviation between the fitted plane and the point cloud.

[0116] After fitting the plane, in order to improve the flatness of the steel surface and eliminate possible noise and irregularities, a moving average filter is used to smooth the fitted plane. The smoothed steel point cloud edge fitting result and the overall edge fitting effect are shown in the figure below. Figure 6 and Figure 7 shown.

[0117] The calculation formula for obtaining the center position of the smoothed edge surface by applying the moving average filter is:

[0118]

[0119] In the formula, is the center position of the smoothed edge surface, M is the filter window size, C j The center position of the window.

[0120] S6. Calculate the geometric dimensions of the steel section after edge fitting.

[0121] After completing the edge fitting of the steel, the next step is to accurately measure the geometric dimensions of the H-beam. The dimension measurement of the H-beam is based on the processing and fitting of point cloud data, obtaining its key sections and edge surfaces, and then calculating its main geometric dimensions, including flange width, web height and flange thickness. To ensure measurement accuracy, the extracted edge point set is smoothed before calculating the dimensions. A fitting smoothing method based on Gaussian filtering is used to reduce the influence of noise in the measurement.

[0122] Flange width W f Refers to the width of the upper and lower flanges of the H-beam. The measurement method is as follows:

[0123] Point cloud extraction: Extract the edge point sets of the upper flange and the lower flange from the previously fitted edge, denoted as p u and p l , corresponding to the edge points of the upper flange and the lower flange.

[0124] Curve and straight line fitting: The edge curves and straight lines of the upper and lower flanges have been obtained through the previous edge fitting. u and p l Project it onto the fitting curve to ensure the accuracy of the data.

[0125] Calculate the width: The flange width W is obtained by calculating the maximum horizontal distance between the projection points of the upper flange and the lower flange. f The calculation formula is as follows:

[0126]

[0127] In the formula, x i and x j is the coordinate of the projection point in the x direction.

[0128] Web height H w It refers to the overall height of the H-beam, that is, the vertical distance from the upper flange to the lower flange. The measurement method is as follows:

[0129] Get the center position of the edge point: Use the fitted edge curve to calculate the center position of the upper flange and the lower flange from the projection point set, denoted as C u and C l .

[0130] Calculate the web height: The web height H is obtained by calculating the vertical distance between the two center points. w The calculation formula is as follows:

[0131] H w =|y u -y i | (14)

[0132] In the formula, yu and i C u and C l The coordinate in the y direction.

[0133] Flange thickness T f Refers to the thickness of the upper and lower flanges of the H-beam. The measurement method is as follows:

[0134] Extract flange thickness edge points: directly extract the corresponding point sets from the edge fitting data of the upper and lower surfaces.

[0135] Calculate thickness: Use the normal vector information and distance formula in the point cloud edge fitting to calculate the vertical distance between the upper and lower surfaces to obtain the flange thickness. The calculation formula is as follows:

[0136]

[0137] Where a1, b1, c1 and d1 are the parameters of the upper surface plane, a2, b2, c2 and d2 are the parameters of the lower surface plane, and x, y and z are the three-dimensional coordinates of a point on the upper flange or the lower flange.

[0138] S7. End.

[0139] The final measurement results are shown in the table below, which includes all the key dimension data of the H-beam.

[0140] Table 1 H-beam key dimensions data table

[0141]

[0142]

[0143] In the table, the true value is the standard size setting of the H-shaped steel, the average value is the average value of the measured values ​​after multiple measurements using the method of this embodiment, the extreme value is the difference between the maximum and minimum values ​​of each parameter during the measurement process, and the average error is the average value of the absolute difference between the measured value and the true value of each parameter.

[0144] It should be noted that four line laser scanners 3 are used in this embodiment, and the scanning angle of each line laser scanner 3 is fixed. The number and scanning angle of the line laser scanners 3 can be adjusted according to actual conditions as long as the complete point cloud data can be obtained.

[0145] In summary, the method used in this embodiment can measure the contour dimensions of steel sections under high-temperature production conditions, and has good performance in terms of both measurement accuracy and measurement stability.

[0146] Embodiment 2 A terminal device

[0147] This embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it is used to implement a multi-frame three-dimensional point cloud stitching measurement method for high-temperature environment steel production in Example 1.

[0148] Embodiment 3 A computer readable storage medium

[0149] The computer-readable storage medium in this embodiment stores a computer program, and when the computer program is executed by the processor, it is used to implement a multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production in Example 1.

[0150] Among them, the computer-readable storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a computer-readable storage medium is coupled to a processor so that the processor can read information from the computer-readable storage medium and write information to the computer-readable storage medium. Of course, the computer-readable storage medium can also be a component of the processor. The processor and the computer-readable storage medium can be located in an application-specific integrated circuit (ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the computer-readable storage medium can also exist in a communication device as discrete components. Specifically, the computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk, etc. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

Claims

1. A multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production, characterized in that: The process includes the following steps: S1, obtaining three-dimensional point cloud data of the steel surface through a line laser scanner; S2, performing preprocessing operations of denoising and normalizing the three-dimensional point cloud data; S3, using the improved ICP algorithm to stitch the 3D point cloud data; S4, performing Laplace smoothing and normal vector homogenization processing on the spliced ​​three-dimensional point cloud data; S5, performing edge fitting of the steel section on the data processed in step S4; S6. Calculate the geometric dimensions of the steel section after edge fitting; S7. End.

2. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 1 is characterized in that: The improved ICP algorithm is run in the following order: S31, before the ICP algorithm iteration, feature descriptor matching is introduced to calculate the local geometric features of each point in the point cloud to form a feature descriptor, and point pair matching is preferentially performed based on the similarity of the feature descriptors; S32. During the iteration of the ICP algorithm, a weight factor is introduced for each point to affect the matching between point pairs, and an adaptive threshold control mechanism is introduced to dynamically adjust the distance threshold and the normal vector consistency threshold according to the current point pair error distribution; S33, after the ICP algorithm is iterated, a candidate rigid body transformation model is calculated by the RANSAC algorithm, and the distance between the point pairs is estimated based on the least squares error. When the error is less than a preset value, the point pair is counted into the internal point set; S34, repeating steps S31 to S33 until the iteration is completed, selecting the model with the most inliers in the inlier set, and calculating the objective function that maximizes the number of inliers.

3. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 2 is characterized in that: The calculation formula for the feature descriptor matching is: In the formula, similarity(p i ,q j ) is point p i and dot q j The feature descriptor matching, f(p i ) and f(q j ) are points p i and dot q j The feature descriptor of ||f(p i )-f(p j )|| is f(p i ) and f(q j ), p i and q i are points in the source point cloud and target point cloud respectively, and λ is the standard deviation of feature distance.

4. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 3 is characterized in that: The calculation formula for introducing the weight factor includes: Where E'(R,t) is the objective function of the iterative process, R is the rotation matrix, t is the translation vector, ω i is the weight factor of the i-th point, ‖p i -(R*q i +t)‖ is point p i and (R*q i +t), ‖p i -q i ‖ is the point p i and dot q i The Euclidean distance between them, σ is the standard deviation of the distance, θ i For point p i and dot q i The angle between the normal vectors of .

5. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 2 is characterized in that: The calculation formula of the adaptive threshold is: Thershold=μ+K·σ1 Where Thershold is the adaptive threshold, μ is the mean of the point pair error in the current iteration, σ1 is the standard deviation of the point pair error in the current iteration, and K is the adjustment coefficient.

6. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 4 is characterized in that: The calculation formula of the objective function of maximizing the number of internal points is: max|I|={i:||Rp i +t-q i ||<∈} In the formula, ∈ is the preset value.

7. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 1 is characterized in that: Step S5 is performed in the following order: S51, for each point on the point cloud data, calculate the number of points whose distance from this point is less than a set radius, and define the points whose number is less than the set value as edge points; S52, using a density clustering algorithm to group edge points according to the distance and density relationship between the points; S53, calculating the center position of each group of edge points as the set center of the group; S54, using the center of the set as a reference, fitting the plane of the steel section using the least square method; S55, applying a moving average filter to smooth the fitted plane.

8. The multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production according to claim 1 is characterized in that: The geometrical dimensions of the steel section include flange width, web height and flange thickness.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-frame three-dimensional point cloud stitching measurement method for high-temperature environment steel production as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, is used to implement a multi-frame three-dimensional point cloud splicing measurement method for high-temperature environment steel production as described in any one of claims 1 to 8.

Images