Tunnel orthoimage high-precision generation processing method, system and platform based on adaptive fitting and fixed-distance equal division, and storage medium

Abstract

The invention discloses a tunnel orthoimage high-precision generation processing method, system and platform based on adaptive fitting and fixed-distance equal division and a storage medium. According to the scheme, first data corresponding to a tunnel are generated and obtained, the first data are preprocessed, and second data corresponding to the first data are generated; processing the second data through longitudinal calibration and circumferential calibration respectively, and generating third data corresponding to the second data; wherein the third data is geometric distortion-free tunnel ortho-gray image data, and the system, the platform and the storage medium corresponding to the method can realize high-precision longitudinal and circumferential calibration at the same time, are suitable for tunnels in any shape, and can generate ortho-images which have a strict geometric mapping relation and support centimeter-level precision measurement. According to the scheme, the problem of longitudinal and circumferential distortion in the orthoimage generation of the tunnel in any shape can be solved, and high-precision geometric measurement is realized.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of surveying and tunnel engineering detection technology, and particularly relates to a tunnel orthographic image high-precision generation processing method, system, platform and storage medium based on adaptive fitting and constant-distance equal division. BACKGROUND

[0002] With the rapid growth of tunnel mileage in China, safety monitoring and maintenance during the tunnel operation period become crucial. Cracks, leakage, spalling and other diseases may occur in tunnels during long-term operation, requiring efficient and accurate detection means.

[0003] Currently, image-based tunnel disease detection methods mainly rely on area array or linear array cameras. However, these methods have obvious limitations: they rely on complex external lighting systems, and image quality is easily affected by uneven lighting; they cannot directly obtain three-dimensional topographic information of the tunnel surface; and linear array cameras are prone to image distortion due to carrier shaking during high-speed movement, posing challenges to subsequent image rectification, denoising and stitching techniques.

[0004] Laser scanning technology, especially mobile laser scanning technology, has gradually been applied in tunnel detection due to its high precision, high efficiency and immunity to environmental lighting. The point cloud data obtained by this technology contains both spatial coordinates and reflectivity information, which can be used to generate grayscale images of the tunnel inner wall. However, the grayscale images generated directly from the original point cloud data have significant distortion: first, in the longitudinal direction (tunnel forward direction), due to track irregularities and carrier shaking, the scanning platform speed is uneven, resulting in uneven density of scanning lines obtained at a fixed rotation speed, causing longitudinal stretching or compression of the image and inaccurate mileage information. Second, in the circumferential direction (tunnel cross-section direction), the scanning center of the laser radar does not coincide with the geometric center of the tunnel cross-section, and the shape of the tunnel cross-section is complex and variable (such as circular, horseshoe-shaped, irregular, etc.), so direct expansion projection or only circular / elliptical fitting will cause severe projection distortion, making the generated image unsuitable for accurate length and area measurement.

[0005] Existing technologies, such as patent CN117132709A, although propose a method of grayscale correction through polygon fitting and centroid angle, which is suitable for non-circular tunnels, but it has obvious limitations. In longitudinal calibration, it only relies on simple linear interpolation, which cannot effectively handle nonlinear speed errors caused by vibration, etc. In circumferential calibration, its method based on centroid angle normalization may cause the problem of equal angle and unequal arc length on non-circular cross-sections, i.e., the mapping relationship between pixels and real physical distance is not strictly consistent, limiting the absolute accuracy of area measurement.

[0006] Therefore, in view of the above technical problems and defects, there is an urgent need to design and develop a tunnel orthographic image high-precision generation processing method, system, platform and storage medium based on adaptive fitting and constant-distance equal division. Summary of the Invention

[0007] To overcome the shortcomings and difficulties of the existing technology, the present invention provides a method, system, platform and storage medium for high-precision generation and processing of tunnel orthophotos based on adaptive fitting and fixed-distance equal division, so as to solve the longitudinal and circumferential distortion problems in the generation of orthophotos of tunnels of arbitrary shapes and realize high-precision geometric measurement.

[0008] The first objective of this invention is to provide a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division; the second objective of this invention is to provide a high-precision generation and processing system for tunnel orthophotos based on adaptive fitting and fixed-distance equal division; the third objective of this invention is to provide a high-precision generation and processing platform for tunnel orthophotos based on adaptive fitting and fixed-distance equal division; and the fourth objective of this invention is to provide a computer-readable storage medium.

[0009] The first objective of this invention is achieved as follows: the method includes the following steps: generating and acquiring first data corresponding to the tunnel, preprocessing the first data, and generating second data corresponding to the first data; wherein, the first data is the original point cloud data inside the tunnel, and the point cloud data includes spatial coordinate information data and reflection intensity information data; the second data is initial grayscale image data;

[0010] The second data is processed by longitudinal and circumferential calibration respectively, and a third data corresponding to the second data is generated; wherein, the third data is geometrically distortion-free tunnel orthophoto grayscale image data; the longitudinal calibration includes combining dynamic filtering and cubic spline interpolation methods to correct and adjust the mileage data and scan line distribution state corresponding to the first data; the circumferential calibration includes using a tunnel cross-section adaptive fitting method to fit and generate a polygon corresponding to the tunnel cross-section; and using a fixed-distance equal division method to divide the perimeter of the polygon into equal parts at fixed distances, and generating centroid angles corresponding to the division points and centroids.

[0011] The second objective of this invention is achieved as follows: the system is used to implement the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division. The system includes: a first data processing and generation unit, used to generate and acquire first data corresponding to the tunnel, and preprocess the first data to generate second data corresponding to the first data; wherein the first data is the original point cloud data inside the tunnel, and the point cloud data includes spatial coordinate information data and reflection intensity information data; the second data is initial grayscale image data; a second data processing and generation unit, used to process the second data longitudinally and circumferentially, respectively, and generate third data corresponding to the second data; wherein the third data is geometrically distortion-free tunnel orthophoto grayscale image data; the longitudinal calibration includes combining dynamic filtering and cubic spline interpolation methods to correct and adjust the mileage data and scan line distribution state corresponding to the first data; the circumferential calibration includes using an adaptive fitting method for tunnel cross-sections to fit and generate polygons corresponding to the tunnel cross-sections; and using a fixed-distance equal division method to divide the perimeter of the polygons at fixed distances, and generating centroid angles corresponding to the division points and centroids.

[0012] The third objective of this invention is achieved as follows: the platform includes a processor, a memory, and a control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division; wherein the control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division is executed on the processor, and the control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division is stored in the memory, and the control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division implements the method for high-precision generation and processing of tunnel orthophotos based on adaptive fitting and equal interval division.

[0013] The fourth objective of this invention is achieved as follows: the computer-readable storage medium stores a control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division, and the control program for the high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division implements the method for high-precision generation and processing of tunnel orthophotos based on adaptive fitting and equal interval division.

[0014] This invention generates and acquires first data corresponding to a tunnel through a method, and preprocesses the first data to generate second data corresponding to the first data. The first data is raw point cloud data inside the tunnel, including spatial coordinate information and reflection intensity information. The second data is initial grayscale image data. The second data is processed by longitudinal and circumferential calibration to generate third data corresponding to the second data. The third data is geometrically distortion-free tunnel orthophoto grayscale image data. The longitudinal calibration includes combining dynamic filtering and cubic spline interpolation to correct and adjust the mileage data and scan line distribution corresponding to the first data. The circumferential calibration includes using an adaptive fitting method for the tunnel cross-section to generate a polygon corresponding to the tunnel cross-section; and using a fixed-distance equal division method to divide the perimeter of the polygon at fixed distances, generating centroid angles corresponding to the division points and centroids. The invention also includes a corresponding system, platform, and storage medium, enabling high-precision longitudinal and circumferential calibration simultaneously. It is applicable to tunnels of any shape and can generate orthophotos with strict geometric mapping relationships and support centimeter-level precision measurements.

[0015] In other words, by integrating longitudinal calibration with dynamic Kalman filtering and cubic spline interpolation, and circumferential calibration based on adaptive polygon fitting and equidistant division, this invention successfully reconstructs a geometrically distortion-free tunnel orthophoto image from a moving laser scanning point cloud. This achieves a strict mapping between pixels and physical distances, enabling the measurement accuracy of the length and area of ​​tunnel surfaces of arbitrary shapes to reach the centimeter level. It completely solves the problems of image distortion and measurement inaccuracy caused by uneven speed, projection center offset, and shape limitations in traditional methods. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a technical roadmap for a high-precision generation and processing method of tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0018] Figure 2 This is a schematic diagram of a mobile detection device for a tunnel laser scanning system, which is based on an adaptive fitting and fixed-distance equal division method for high-precision generation and processing of tunnel orthophotos according to the present invention.

[0019] Figure 3This is a schematic diagram illustrating outlier removal in a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance division according to the present invention.

[0020] Figure 4 This is a schematic diagram of point cloud sparsification in a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0021] Figure 5 This is a schematic diagram of the tunnel point cloud coordinate system of the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0022] Figure 6 This is a schematic diagram of the Canny edge detection and Hough line detection results of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0023] Figure 7 This is a schematic diagram of point cloud vertical calibration for a high-precision generation and processing method of tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0024] Figure 8 This is a schematic diagram of the tunnel cross-section adaptive method of the present invention, which is a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division.

[0025] Figure 9 This is a schematic diagram of the projection center transformation of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0026] Figure 10 This is a schematic diagram of the interval division algorithm for a high-precision generation and processing method of tunnel orthophotos based on adaptive fitting and interval division according to the present invention.

[0027] Figure 11 This is a schematic diagram comparing images of tunnels with different cross-sectional shapes before and after correction, based on a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0028] Figure 12 This is a detailed comparison diagram of the images before and after correction in the high-precision generation and processing method of tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0029] Figure 13 This is a schematic diagram comparing the area measurement values ​​before and after correction of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0030] Figure 14This is a schematic diagram of the image region measurement results of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0031] Figure 15 This is a schematic diagram of the process steps of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0032] Figure 16 This is a schematic diagram of the architecture of a high-precision generation and processing system for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0033] Figure 17 This is a schematic diagram of the architecture of a high-precision generation and processing platform for tunnel orthophotos based on adaptive fitting and fixed-distance equal division according to the present invention.

[0034] Figure 18 This is a schematic diagram of a computer-readable storage medium architecture in one embodiment of the present invention. Detailed Implementation

[0035] To facilitate a clearer understanding of the objectives, technical solutions, and advantages of this invention, the invention will be further described below in conjunction with the accompanying drawings and specific embodiments. Those skilled in the art can easily understand other advantages and effects of this invention from the content disclosed in this specification.

[0036] This invention can also be implemented or applied through other different specific examples, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of this invention.

[0037] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0038] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Secondly, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0039] Preferably, the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division of the present invention is applied in one or more terminals or servers. The terminal is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0040] The terminal can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal can interact with the customer via a keyboard, mouse, remote control, touchpad, or voice control device.

[0041] This invention provides a method, system, and platform for high-precision generation and processing of tunnel orthophotos based on adaptive fitting and fixed-distance equal division.

[0042] like Figure 15 The diagram shown is a flowchart of a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division provided by an embodiment of the present invention.

[0043] In this embodiment, the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance division can be applied to terminals or fixed terminals with display functions. The terminals are not limited to personal computers, smartphones, tablets, desktop computers or all-in-one computers with cameras, etc.

[0044] The high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division can also be applied to a hardware environment consisting of a terminal and a server connected to the terminal via a network. The network includes, but is not limited to, a wide area network (WAN), a metropolitan area network (MAN), or a local area network (LAN). The high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division in this embodiment of the invention can be executed by a server, by a terminal, or by both a server and a terminal.

[0045] For example, for terminals requiring high-precision generation and processing of tunnel orthophotos based on adaptive fitting and equal interval division, the high-precision generation and processing function of tunnel orthophotos based on adaptive fitting and equal interval division provided by the method of this invention can be directly integrated into the terminal, or a client for implementing the method of this invention can be installed. Alternatively, the method provided by this invention can also run on servers or other devices in the form of a Software Development Kit (SDK), providing an interface for the high-precision generation and processing function of tunnel orthophotos based on adaptive fitting and equal interval division in the form of an SDK. Terminals or other devices can then implement the high-precision generation and processing function of tunnel orthophotos based on adaptive fitting and equal interval division through the provided interface. The invention will be further described below with reference to the accompanying drawings.

[0046] like Figures 1-15 As shown, this invention provides a high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division. The method includes the following steps: S001, generating and acquiring first data corresponding to the tunnel, and preprocessing the first data to generate second data corresponding to the first data; wherein, the first data is the original point cloud data inside the tunnel, and the point cloud data includes spatial coordinate information data and reflection intensity information data; the second data is initial grayscale image data; the data preprocessing includes at least one of point cloud denoising, point cloud thinning, and laser intensity correction; the laser intensity correction uses a piecewise linear stretching algorithm to convert the intensity information of the lidar into grayscale values ​​and enhance the corresponding contrast. S002. The second data is processed by longitudinal and circumferential calibration respectively, and a third data corresponding to the second data is generated; wherein, the third data is a geometrically distortion-free tunnel orthophoto grayscale image data; the longitudinal calibration includes combining dynamic filtering and cubic spline interpolation methods to correct and adjust the mileage data and scan line distribution state corresponding to the first data; the circumferential calibration includes using a tunnel cross-section adaptive fitting method to fit and generate a polygon corresponding to the tunnel cross-section; and using a fixed-distance equal division method to divide the perimeter of the polygon into equal parts at a fixed distance, and generating centroid angles corresponding to the division points and centroids.

[0047] The process of longitudinal and circumferential calibration of the second data, and generating third data corresponding to the second data, further includes: S0021, creating a first mapping relationship between image row direction pixels and tunnel longitudinal mileage, and using the inherent sleeper structure in the tunnel as a geometric reference, correcting the uneven distribution of scan lines and mileage accumulation errors caused by fluctuations in the carrier's moving speed through dynamic filtering and interpolation algorithms; S0022, creating a second mapping relationship between image column direction pixels and tunnel circumferential arc length, adaptively fitting the tunnel cross-sectional contour into a polygon, calculating the centroid of the polygon as the projection center, and segmenting the perimeter of the polygon based on a fixed-distance equal division algorithm.

[0048] The process of creating a first mapping relationship between image row direction pixels and tunnel longitudinal mileage, and using the inherent sleeper structure within the tunnel as a geometric reference, corrects for uneven scan line distribution and mileage accumulation errors caused by fluctuations in the carrier's moving speed through dynamic filtering and interpolation algorithms. This process further includes: S00211, automatically detecting and locating the edges or centerlines corresponding to the sleepers based on the second data, and constraining and adjusting the detected sleeper position sequences based on the standard sleeper spacing or known actual spacing, and constructing a high-precision sleeper mileage sequence corresponding to the sleepers; S00212, using the sleeper mileage sequence as observation values, combining the carrier motion model, and dynamically correcting the original mileage values ​​corresponding to the lidar scanning sections using a Kalman filtering algorithm; S00213, combining a cubic spline interpolation function, and based on the corrected mileage point data, resampling the scan lines, and generating a point cloud section sequence that is uniformly distributed longitudinally and has accurate mileage.

[0049] The process of creating a second mapping relationship between image column direction pixels and tunnel circumferential arc length, adaptively fitting the tunnel cross-section contour into a polygon, calculating the centroid of the polygon as the projection center, and segmenting the polygon perimeter based on a fixed-distance equal division algorithm, further includes: S00221, recursively simplifying the single-section point cloud data corresponding to the tunnel, and dynamically adjusting the distance threshold for determining whether to retain points during the recursion; wherein, in each recursion, for the current point set to be simplified, the perpendicular distance from all points to the lines connecting their first and last endpoints is calculated, and only the point with the largest distance is retained as the key vertex of the polygon, and the point set is divided into two subsets for recursive processing based on this point; the threshold gradually decreases as the recursion depth increases; S00222, generating and acquiring fourth data corresponding to the tunnel, and smoothing and optimizing the vertex positions corresponding to the fourth data by minimizing an energy function; wherein, the fourth data is the initial polygon vertex data obtained in the recursive simplification; the energy function includes at least one term to constrain the consistency between the area of ​​the optimized polygon and the coverage area of ​​the original point cloud, and one term to constrain the smoothness of the polygon side length.

[0050] The process of creating a second mapping relationship between image column direction pixels and tunnel circumferential arc length, adaptively fitting the tunnel cross-section contour into a polygon, calculating the centroid of the polygon as the projection center, and segmenting the polygon perimeter based on a fixed-distance equal division algorithm, further includes: S00223, generating and acquiring fifth data corresponding to the tunnel cross-section contour, and generating corresponding sixth data based on a preset target resolution; wherein, the fifth data is the total perimeter data of the polygon obtained by adaptive fitting; the target resolution is the actual physical length represented by each pixel; the sixth data is the height of the circumferentially calibrated image. S00224. Starting from a vertex of the polygon, the cumulative arc length is divided equally along the polygon boundary according to the preset physical length, and the corresponding seventh data is generated by linear interpolation; wherein, the seventh data is the coordinate data of a series of equally divided points; S00225. The centroid angle corresponding to the line connecting the adjacent equally divided points and the centroid of the polygon is calculated and generated, and each point in the original cross-sectional point cloud is mapped to the corresponding equally divided interval according to the normalized angle corresponding to the line connecting the centroid and the centroid, and the reflection intensity information of the point is assigned to the corresponding pixel of the equally divided interval.

[0051] Specifically, in this embodiment of the invention, the limitations of existing technologies, such as poor applicability, limitation to circular cross-sections, and inability to perform area measurement, are systematically analyzed. The technical approach is mainly divided into two modules: longitudinal calibration and circumferential calibration. Longitudinal calibration uses sleeper spacing as a benchmark, integrating dynamic Kalman filtering and cubic spline interpolation, combined with carrier speed and lidar rotation speed, to achieve uniform distribution of scan lines and correction of mileage errors. Circumferential calibration first uses the tunnel cross-section adaptive fitting method (TSAM) innovatively proposed in this scheme to generate the cross-section profile. This method significantly improves the robustness of fitting complex shapes through dynamic threshold simplification and vertex smoothing optimization. Subsequently, through projection center transformation and the fixed-distance equal division algorithm proposed in this scheme, the perimeter of the fitted polygon is precisely divided by distance. This fixed-distance equal division algorithm strictly achieves accurate pixel mapping based on perimeter equal division, avoiding the approximate error of traditional angle projection, ensuring that each pixel corresponds to a fixed physical distance, and supporting high-precision area measurement. This method is applicable to the generation of high-resolution orthophotos of tunnels of arbitrary shapes. The generated images have low distortion, high clarity, and area measurement capabilities, and can be widely used in tunnel management scenarios such as tunnel defect detection, ancillary facility statistics, and structural performance evaluation. Figure 1 The diagram shown is a technical roadmap for this solution.

[0052] In this solution, regarding laser point cloud data acquisition and preprocessing, the acquisition of point cloud data is based on the fundamental work of laser tunnel deformation and defect measurement. This solution utilizes a self-developed tunnel laser scanning system track inspection vehicle, with Faro, Leica, and Z+F lidars as the main measuring equipment. A schematic diagram of the equipment is shown below. Figure 2 As shown, during the mobile measurement process, as the inspection vehicle moves forward, the lidar scans along a two-dimensional spiral path to capture point cloud information of the tunnel cross-section. The data acquired by the equipment is all relative measurement data in the scanner's coordinate system, requiring no absolute coordinate transformation. Because the tunnel inner wall point cloud acquired by the tunnel laser scanning system contains inherent intensity information, the tunnel cross-section can be presented in image form based on this intensity information.

[0053] Tunnel laser scanning system mobile detection equipment hardware such as Figure 2 As shown, the algorithm implementation in this scheme uses data collected by the Z+FPROFILER9012 lidar. This lidar model exhibits excellent performance and is highly representative. The parameters of the lidar mounted on the mobile detection device are shown in Table 1.

[0054] Table 1 LiDAR Parameters

[0055]

[0056] For data preprocessing, after acquiring the raw point cloud, it is essential to preprocess it to ensure data accuracy and usability. Preprocessing mainly includes point cloud denoising, point cloud thinning, and piecewise linear stretching. Point cloud denoising aims to remove outliers and erroneous points, thereby improving the accuracy and recognizability of the generated orthophoto grayscale image. Point cloud thinning focuses on eliminating redundant and repetitive information to improve data processing efficiency. Laser intensity correction, through piecewise linear stretching, converts the LiDAR intensity information into a grayscale image, enhancing contrast and highlighting image details.

[0057] like Figure 3 The diagram illustrates the outlier removal process based on the distance thresholding method and the K-means clustering algorithm. Specifically, the distance thresholding method calculates the distance from each point on each cross-section to the origin and removes outliers exceeding a preset threshold; the K-means clustering algorithm is used to identify and remove abnormal points that are too close to the tunnel surface. Figure 3 As shown, this process effectively eliminates obvious outliers and improves the quality of the point cloud. Excessive point cloud density significantly reduces the algorithm's computation speed, therefore, thinning is necessary. This scheme combines point cloud density and target grayscale image resolution to thin the original point cloud, employing a thinning algorithm based on uniformly distributed random numbers. Data simplification is achieved by randomly selecting a subset. A schematic diagram of the algorithm is shown below. Figure 4As shown in the experiment, the impact of the thinning ratio on the reconstruction effect is as follows: a high thinning ratio (40%) can significantly improve the calculation speed, but it will lead to a decrease in accuracy, loss of reconstruction details, and an increase in error; a medium thinning ratio (20%) achieves a good balance between calculation efficiency and accuracy, with the smallest reconstruction error; a low thinning ratio (10%) produces a reconstruction effect closest to the original data, but the improvement in calculation speed is limited. If the thinning ratio is too high (>40%), the point cloud data will be insufficient, which may amplify the errors in distance and area calculations, thereby reducing the accuracy of the reconstructed image. Therefore, the thinning ratio should be comprehensively weighed according to the specific accuracy requirements.

[0058] After converting point cloud intensity into grayscale values, this scheme employs a piecewise linear stretching algorithm to enhance the image. This algorithm constructs a piecewise linear transformation function to achieve targeted adjustment of grayscale values: enhancing the dynamic range of low-contrast regions while compressing excessive stretching in high-contrast regions, thereby optimizing overall contrast and improving detail recognition. This method is particularly suitable for scenarios with non-uniform distribution of laser intensity signals, effectively highlighting tunnel surface texture and defect features. Subsequently, the grayscale image matrix is ​​filled according to the storage order of the original uncorrected point cloud to generate an initial grayscale image of the tunnel inner wall, providing basic data for subsequent longitudinal and circumferential calibration. Although this initial image retains the original intensity information of the point cloud, it does not eliminate the effects of scan line unevenness and projection distortion.

[0059] For the point cloud calibration algorithm, this scheme first needs to clarify the coordinate system definition of the tunnel point cloud to facilitate calculation. For example... Figure 5 As shown, the point cloud collected by the lidar in one rotation is considered a cross-section, with the forward direction as the Y-axis and the X and Z axes forming the cross-sectional plane. The Y value is the mileage value, and all points within the same cross-section have the same Y value. The Y value of the original point cloud is assigned based on the carrier's motion speed, but due to factors such as track irregularities, track surface roughness, and carrier swaying, the actual motion speed often deviates from the preset uniform speed, leading to accumulated errors in the mileage value. This not only distorts the true distance mapping between cross-sections in the grayscale image generated from the point cloud but also reduces the accuracy of area measurement. Therefore, the point cloud mileage information must be calibrated, i.e., longitudinal calibration. Simultaneously, given the large number of points and irregular shapes in each cross-section of the original tunnel point cloud, accurately obtaining the true distance mapping between grayscale pixels while calibrating the image is also a key challenge, requiring circumferential calibration. Therefore, the technical approach of this solution is divided into two main modules: longitudinal point cloud calibration and circumferential point cloud calibration. Its core objective is to establish a quantitative correspondence between the pixels in the grayscale image generated by the lidar point cloud and the actual spatial geometric distance. Specifically, longitudinal calibration employs automated sleeper marking, dynamic Kalman filtering, and cubic spline interpolation to obtain accurate mileage values ​​between scan lines; circumferential calibration first automatically generates the cross-sectional polygonal contour based on the Tunnel Cross-Section Adaptive Fitting Method (TSAM), and finally achieves accurate correction of the circumferential point cloud through a fixed-distance equal division algorithm.

[0060] In the longitudinal point cloud calibration of tunnels, this solution addresses the limitations of existing longitudinal calibration methods, such as low efficiency due to manual annotation, neglect of nonlinear errors by linear interpolation, and uneven scan line distribution. It proposes an optimized method based on dynamic filtering and automated geometric reference. Through fully automated sleeper edge detection, automatic sleeper identification, dynamic Kalman filtering, and cubic spline interpolation calibration, a longitudinally uniform and distortion-free orthophoto grayscale image is generated, suitable for complex track environments such as mining tunnels. After optimization, calibration efficiency is improved by more than 10 times, longitudinal error is reduced from ±5mm to <1mm, and the mapping accuracy between image pixels and actual mileage is significantly improved, supporting tunnel defect detection and area measurement.

[0061] To overcome the subjective errors and low efficiency of manual sleeper labeling, this module uses image processing technology to automatically extract sleeper edges as geometric references. The standard sleeper spacing is 0.6m, with an allowable deviation of ±20mm. The processing flow begins by applying Gaussian filtering to the uncorrected grayscale image (generated from point cloud intensity) to smooth noise and preserve sleeper edge features. The filter kernel size is 3×3, the standard deviation σ = 1.5, and the filtering formula is:

[0062]

[0063] Where I(x,y) represents the original pixel gray value, G(x,y) is the filtered pixel value, and σ is the standard deviation of the Gaussian distribution. Gaussian filtering enhances the continuity of the sleeper edge by smoothing noise such as uneven illumination or intensity fluctuations. Next, the Canny

[38] edge detection algorithm is used to extract the sleeper edge, and the Sobel operator is used to calculate the image gradient:

[0064]

[0065] Among them, I x and I y The gradients in the horizontal and vertical directions are respectively. The Canny algorithm generates continuous sleeper edge lines through threshold filtering and edge connection. Then, the Hough transform

[39] is used to identify the straight line features in the edge, and the formula is:

[0066] ρ=xcosθ+ysinθ(3)

[0067] Where ρ represents the distance from the line to the image origin, and θ represents the angle between the line and the horizontal axis. The Hough transform converts edge points from image space (x, y) to parameter space (ρ, θ), searches for voting peaks in the parameter space, identifies points conforming to sleeper geometry, and records the sleeper center position, thereby generating a sequence of mileage points {y1, y2, ... y}. n}, each y iThis indicates the longitudinal position of a sleeper. This automated process significantly improves efficiency, generates accurate sleeper mileage sequences, and provides a reliable benchmark for subsequent calibration.

[0068] Dynamic Kalman filtering and spline interpolation calibration are used in this system. The inspection vehicle moves at a nominal speed, while the lidar scans at a fixed rotation speed, generating a fixed number of scan lines at regular intervals. Each line corresponds to a point cloud cross-section, recording the (x,z) coordinates of the tunnel cross-section and the longitudinal mileage (y) coordinate. However, track irregularities causing vibrations or wheel spin lead to speed fluctuations, resulting in uneven scan line numbers and spacing. This inaccurate mileage values ​​for the original scan lines, with errors reaching several centimeters, cause longitudinal distortion in the grayscale image, affecting defect detection and area measurement. Dynamic Kalman filtering corrects nonlinear errors by fusing motion models and sleeper observations, and combines this with cubic spline interpolation to assign accurate mileage values ​​to the existing scan lines. Simultaneously, the scan line distribution is adjusted to ensure uniformity, generating an orthophoto image where pixels precisely correspond to the actual mileage.

[0069] Dynamic Kalman filtering

[40] estimates the mileage and speed of the detection vehicle through prediction and correction steps, with the input including the state vector [y t ,v t ], where y t Distance (m), v t Velocity (m / s). Prediction based on motion model:

[0070] y t =y t-1 +v t-1 ·Δt+w t (4)

[0071] The input is the state [y] from the previous time step. t-1 ,v t-1 [Time step Δt (LiDAR sampling interval) and process noise w] t The output is the predicted mileage y at the current time. t Process noise w t The covariance matrix of the mileage and speed disturbances caused by simulated vibration and idling is:

[0072]

[0073] Here, Q describes the disturbance amplitude of mileage (standard deviation 0.01m) and speed (standard deviation 0.1m / s), used to update the state covariance matrix P in the prediction step, ensuring that the filter dynamically adapts to nonlinear errors. The value of Q is based on physical analysis: a mileage disturbance standard deviation of 0.01m corresponds to centimeter-level deviations caused by vibration or idling, and a speed disturbance standard deviation of 0.1m / s corresponds to a 20% fluctuation of the nominal speed of 0.5m / s, which conforms to the common disturbance range in tunnel environments (such as track unevenness or wheel slippage). The observation input is the automatically labeled sleeper mileage z. t The model is:

[0074] z t =m t +v t (6)

[0075] The input is the actual mileage m. t The output is the observed sleeper mileage z. t v t To observe the noise, the covariance R = [0.01] 2 The symbol ] indicates that the standard deviation of the sleeper mileage markings is 0.01m. Formula z t =m t +v t Its purpose is to establish the relationship between observed values ​​and true mileage, indicating that z... t It is for m t The noisy estimate provides external observation data (sleeper positions) to correct the prediction results. Prediction formula y t =y t-1 +v t-1 ·Δt+w t It provides state prediction based on motion models, and combines the two by fusing information through Kalman filtering to calculate a weighted average:

[0076]

[0077] The formula input is the predicted mileage y. t Observation mileage z t The output consists of covariance matrices Q and R, and is the corrected mileage. n is the number of sleepers, and K is the Kalman gain.

[0078] Cubic spline interpolation assigns accurate mileage values ​​to existing scan lines using a corrected sleeper mileage sequence and adjusts the distribution to ensure uniformity. The input includes the corrected sleeper mileage sequence. Nominal speed v, lidar rotation speed f, and sleeper spacing w. The theoretical number of scan lines is:

[0079]

[0080] Due to vibration and idling, the actual number and spacing of scan lines are uneven, resulting in inaccurate original mileage values. Interpolation is performed by constructing a smoothing function for each sleeper interval [y i ,y i+1 Generate a uniformly distributed scanline mileage sequence {y1, y2...y}. num}:

[0081] s i (y)=a i (yy i ) 3 +b i (yy i ) 2 +c i (yy i )+d i (9)

[0082] Where y represents any mileage position within the interval, and yi represents the starting mileage of the interval. The coefficients ai, bi, ci, and di are obtained by satisfying interpolation conditions, continuity conditions, and natural boundary conditions. The mileage points of a uniformly distributed scan line are calculated using the following formula:

[0083]

[0084] Calculate, where y i,j Let be the mileage value of the j-th scan line. When there are fewer than num scan lines, the spline function s is used. i (y) Generate num uniformly distributed mileage points to replace the original scan line mileage values. Then, perform bilinear interpolation on the original point cloud intensity along the image row direction to complete the pixel grayscale of the newly added scan lines. When there are more than num lines, or exactly num lines but unevenly distributed, uniform sampling is used to select the num scan lines closest to the mileage value, retaining their point cloud intensity as pixel grayscale, and removing the redundant scan lines. The smoothness of cubic spline interpolation is achieved by fitting the continuity of the first and second derivatives to the nonlinear motion trajectory of the detected vehicle caused by vibration or idling, generating a uniformly distributed mileage point sequence. This eliminates the non-uniform distribution and mileage error of the original scan lines, ensuring accurate mileage and grayscale mapping of the distortion-free orthophoto grayscale image. See the diagram for a detailed longitudinal calibration. Figure 7 As shown.

[0085] For tunnel point cloud circumferential point cloud calibration, this scheme approximates the true contour of the cross-section by fitting polygons to obtain accurate distance information. Then, it obtains the mapping relationship between pixels and true distances through projection center transformation and equal-distance division, generating a distortion-free orthophoto grayscale image with distance information.

[0086] To address the challenge of generating orthorectified grayscale images of tunnels of arbitrary shapes from moving laser scanning point clouds, this paper proposes an innovative polygon fitting method called "Tunnel Section Adaptive Method (TSAM)." This method simplifies point cloud data through dynamic thresholding and vertex smoothing, significantly improving fitting accuracy and robustness. Compared to traditional methods' sensitivity to noise and complex cross-sections (such as horseshoe-shaped or irregular tunnels), the TSAM method can adapt to various tunnel cross-section shapes, including circular, horseshoe, and irregular shapes. This ensures that the generated orthorectified grayscale image more closely approximates the actual tunnel cross-section in terms of geometry and area calculation, meeting the high requirements for tunnel defect detection and area measurement.

[0087] The TSAM method's processing flow begins with a point cloud dataset, denoted as P = {p1, p2, ..., p...}. N}, where p i =(x i ,y i ,z i (v1, v2, ..., v3) represents a point in a point cloud, containing three-dimensional spatial coordinates. The goal is to simplify the point cloud into a polygon Poly = {v1, v2, ..., v4}. M}, where v i =(x i ,y i The points are polygon vertices, with the number M being much smaller than N, to preserve the main geometric features of the tunnel cross-section while reducing computational complexity. First, the TSAM method downsamples the point cloud using a dynamic threshold simplification strategy to generate an initial polygonal outline. Then, it optimizes the distribution of polygonal vertices through vertex smoothing adjustment to make it more consistent with the actual cross-sectional shape.

[0088] In the dynamic threshold simplification stage, the TSAM method determines whether to retain a point by calculating the perpendicular distance from each point in the point cloud to the line segment. Given two endpoints A = (x A ,y A ) and B = (x B ,y B For point p in the point cloud i =(x i ,y i The formula for calculating the vertical distance d is:

[0089]

[0090] Where × represents the cross product of vectors, |·| represents the magnitude of the vector, and the input is a point cloud point p. i Given the endpoints A and B of the line segment, the output is the perpendicular distance d from each point to the line. The TSAM method only retains the point with the maximum distance (i.e., d = max{d...}). i |di Using ∈ as the new vertex, the line segment is divided into two sub-segments, and the recursion continues. If no point satisfies d > ∈ , the point within that segment is deleted. This strategy ensures that only one keypoint is added in each recursion, avoiding an excessive number of vertices. The dynamic threshold uses an adaptive strategy: the initial threshold ∈ 0 = R / 100, where R is the diagonal length of the point cloud bounding box; after each recursive simplification, the threshold is proportionally reduced and updated to ∈ . k+1 =∈ k • 0.98, to gradually improve the accuracy of the retained points; when the polygon area A poly With point cloud area A cloud The recursion stops when the error meets the following condition:

[0091] |A poly -A cloud |<0.01m 2 (12)

[0092] Polygon area A poly Calculate using the cross product formula of vertex coordinates:

[0093]

[0094] Where, x i ,y i Given the coordinates of the vertices of a polygon, the input is the vertex set V = {v1, v2, ..., v...}. M The output is the area A of the polygon. poly Point cloud area A cloud It can be calculated using the triangulation method of the original point cloud.

[0095] To further optimize the polygon vertex distribution, the TSAM method introduces a vertex smoothing adjustment step. This step uses an energy function to adjust vertex positions, resulting in a smoother polygon shape and an area that more closely approximates the actual cross-section. For the polygon vertex set V = {v1, v2, ..., v...} M The energy function E is defined as follows:

[0096]

[0097] Where λ1 and λ2 are weighting coefficients, controlling the consistency of area and the smoothness of distance between vertices, respectively. The first term |A poly -A cloud | Ensure the polygon area is close to the point cloud area, the second term Σ|v i+1 -v i Calculate the sum of Euclidean distances between adjacent vertices. Iteratively update vertex positions using gradient descent:

[0098]

[0099] Where η is the learning rate. Let E be the current vertex position, and E be the gradient of the energy function. In the vertex smoothing phase of the TSAM method, the gradient E is the energy function E with respect to vertex v. i =(x i ,y i The partial derivatives of the equation indicate the direction and magnitude of vertex movement, reducing the energy function E and thus optimizing the polygon vertex set. This represents the updated vertex position.

[0100] Gradient descent iteratively updates the initial set of vertices, Poly, step by step to Poly. opt The process continues until the energy function E converges. The TSAM method simplifies the generation of the initial polygon through dynamic thresholding, and combines this with vertex smoothing to ultimately generate a polygon with the same height as the actual tunnel cross-section. A schematic diagram of the TSAM method is shown below. Figure 8 As shown.

[0101] For projection center transformation, during data acquisition, the center of the LiDAR is not at the center of the tunnel cross-section. Without correction, projection distortion will occur in the image, affecting the quality of the tunnel grayscale image and related image processing such as area calculation. Therefore, it is necessary to transform the projection center from the LiDAR center to the cross-section center to make the tunnel image closer to orthographic projection. To accommodate various cross-section shapes, the centroid of a fitted polygon is used as the cross-section center. The formula for calculating the polygon centroid coordinates is as follows:

[0102]

[0103] Where N is the number of sides of the polygon (x i ,y i ) represents the coordinates of the polygon's vertices, (c x ,c y Let S be the centroid coordinates of the polygon and S be the area of ​​the polygon. The centroid coordinates calculated from the fitted polygon and those obtained from the original cross-section show an error within the millimeter range, meeting the accuracy requirements. After obtaining the centroid of the fitted polygon, it can be used as the projection center. Next, the cross-sectional point cloud is divided into equal intervals. A schematic diagram of the projection center transformation is shown below. Figure 9 As shown.

[0104] To achieve accurate circumferential calibration of tunnel cross-sections of arbitrary shapes, this scheme proposes a fixed-distance equal-segmentation algorithm for correcting cross-sectional point clouds. This algorithm precisely segments the fitted polygon along its contour, calculates the coordinates of the division points and their corresponding centroid angles, and corrects the grayscale distribution of the original point cloud accordingly. The core of this algorithm lies in uniformly parameterizing the perimeter of the polygon into a sequence of fixed distances *d*, ensuring that each pixel corresponds to the true physical distance, avoiding projection distortion, and supporting area measurement. Compared to angle-based approximation methods, this algorithm achieves strict equal-distance segmentation through cumulative arc-length interpolation, making it suitable for circular, horseshoe-shaped, and irregular cross-sections. The following details the algorithm flow, formulas, and implementation steps, closely connected to the polygon fitting and projection center transformation logic described earlier.

[0105] To fit a polygon into equal parts, first, calculate the total perimeter S of the polygon using the following formula:

[0106]

[0107] Where, p i and p i+1 Let pi+1 be the coordinates of adjacent vertices of the polygon (when i = N, pi+1 = p1), where N is the number of vertices of the polygon, |p i+1 -p i | represents the Euclidean distance of the i-th edge. The perimeter S represents the actual circumferential length of the tunnel cross-section.

[0108] Based on actual project requirements (such as accuracy requirements and cross-sectional dimensions), the true distance d represented by each pixel is set. The height H of the corrected grayscale image (i.e., the number of rows in the image matrix) is determined by the following formula:

[0109]

[0110] Ensure the total pixel height covers the entire perimeter S. The height H determines the circumferential resolution. To ensure continuous equal division along the contour, start processing from the endpoints p1 and p2 of the initial edge L1, and calculate the remaining length r of the first edge:

[0111] r = |L1| mod d (21)

[0112] Where |L1|=|p2-p1| is the length of L1, and mod represents the modulo operation (remaining margin). The remaining length r is used as the starting offset of the next edge L2 to maintain the continuity of the division points. The specific processing steps are as follows: For the k-th edge Lk (from endpoint pk to pk+1), the effective length of the current edge is calculated starting from the remaining offset rk-1 of the previous edge, which is the current edge length minus rk-1. Within the effective length, a division point is generated every d distance. After processing, the remaining offset of the next edge is updated, which is the remainder of the current effective length divided by d, and it is passed to Lk+1. Each edge processes its own part, but the remaining distance is not discarded, but accumulated and passed to ensure that the entire perimeter S is completely covered as a sequence of segments of length d, the total number of division points is equal to H, and there are no gaps or repetitions between edges. The coordinates of the specific division point qj are calculated by linear interpolation:

[0113]

[0114] Among them, s j =j·d is the cumulative arc length parameter at the j-th division point. Let tj be the starting arc length of the current edge, 0 ≤ tj ≤ 10, ensuring that qj falls inside the edge. This step outputs an ordered sequence of equally divided points {qj}. j}

[0115] Calculation of coordinates and centroidal angle of the points to be divided at a fixed distance, and coordinates q of the points to be divided. j This has been obtained through the interpolation described above. To establish the angular mapping between pixels and physical distances, the centroidal angle θm between adjacent equally spaced points is calculated. For adjacent equally spaced points qm and qm+1, the distance a = |qm| from each point to the centroid c is first calculated. m -c| and b = |q m+1 -c|. Then, the centroidal angle θm is calculated using the law of cosines:

[0116]

[0117] Among them, |q m+1 -q m | represents the straight-line distance between adjacent equally divided points (i.e., the chord length of the equally divided segment). This formula is based on a simple triangle formed by the centroid c and the two points, and directly solves for the included angle using the lengths of the three sides. It is easy to calculate and does not require complex coordinate transformations.

[0118] Pixel correction redistributes the grayscale values ​​of the original cross-sectional point cloud into the H-row image matrix, ensuring that each pixel corresponds to an equally divided interval [θ]. j-1 ,θ j The average physical distance d. First, calculate the sum of all centroidal angles P:

[0119]

[0120] The corrected total number of pixels in the circumferential direction is H, and the normalized cumulative angle is:

[0121]

[0122] Each point Q in the original point cloud i =(x i ,z i (Y is fixed as the current cross-sectional mileage) Centroid φ i Calculated using the arctangent:

[0123]

[0124] Normalization Adjust to the [0,1] interval). Traverse each pixel interval [θ′ j-1 ,θ′ j ], matching the original point {Q} that falls into k |φ′ k ∈[θ′ j-1 ,θ′ j If there is only one point in the interval, its gray level gj = gk (gk is the gray level value converted from point intensity); if there are multiple points, take the average:

[0125]

[0126] Where K j For matching point sets, if an interval is empty, gj is filled in from adjacent pixels using bilinear interpolation. All H intervals are traversed to generate a grayscale matrix with a fixed height H and a width equal to the vertical scan lines. The image height H needs to be set considering the cross-sectional size, carrier speed, and accuracy to optimize the correction effect. See the schematic diagram of the fixed-distance equal division algorithm. Figure 10 As shown.

[0127] Through the above steps, a distortion-free orthophoto grayscale image is generated, with each pixel corresponding to a real distance d, enabling area measurement. Combined with longitudinal calibration, the image supports centimeter-level identification of tunnel defects. It is suitable for complex cross-sections.

[0128] In summary, the proposed method is as follows: First, the point cloud is preprocessed with denoising and thinning to generate an initial grayscale image. Then, intelligent high-precision software is used for annotation, combined with algorithms such as automated sleeper edge detection, dynamic Kalman filtering, and cubic spline interpolation to complete longitudinal pixel calibration, ensuring uniform distribution of scan lines and eliminating mileage errors. Circumferential calibration sequentially employs the TSAM method for efficient polygon fitting of the cross-section, projection center transformation based on the centroid of the fitted polygon, and the proposed fixed-distance equal-division algorithm for pixel calibration. The core of this algorithm lies in dividing the perimeter of the fitted polygon into equal parts at specified distances, calculating the coordinates of the division points and their corresponding centroid angles, and traversing all pixels. Based on the correspondence between the normalized centroid angle and the pixel, distortion-free correction of the circumferential point cloud is achieved.

[0129] For accuracy verification, to validate the method's accuracy and feasibility, this scheme involved on-site scanning of point cloud data from three sections of Chongqing Metro Line 16. Chongqing Metro Line 16 employs a combination of mining and tunneling methods. The two sections constructed using the mining method have large, representative cross-sectional areas, and their cross-sectional shapes change midway, simulating common geometric variations in real-world engineering. The tunneling section used was a curved section to verify the image generation quality under curved conditions. Specifically, the two mining method test sections had horseshoe and irregular cross-sectional shapes, respectively, while the tunneling section had a standard circular cross-section with staggered segment splicing, a ring width of 1.5m, and an inner diameter of 5.4m. The carrier's operating speed was 1.8km / h (0.5m / s), and the lidar rotation speed was 100 rpm. Calculations showed that the actual distance between longitudinal sections was 5mm.

[0130] In the horseshoe-shaped scanning section, the tunnel cross-sectional shape changes, with the perimeter of the fitted polygons being 30m before and 33m after the change; the perimeter of the fitted polygon for the irregular cross-section is 32m. In this experiment, the image height for the mining method tunnel was set to 5000 pixels, and the image height for the shield tunnel was set to 4000 pixels. Calculations showed that the pixel resolutions for the horseshoe-shaped cross-section were 5mm×6mm and 5mm×6.6mm, the irregular cross-section was 5mm×6.4mm, and the circular cross-section of the shield tunnel was 5mm×4.2mm. Calculations indicate that the actual distances represented by pixel length and width are similar, resulting in minimal image distortion. Therefore, setting the image height parameter is reasonable. The appropriate LiDAR model, rotation speed setting, carrier speed, and image height parameter should be selected based on project requirements and accuracy specifications.

[0131] Before experimental verification, the Canny edge detection algorithm was first used to extract sleeper edge features from the original point cloud image, and Hough transform was combined to identify straight line parameters, thereby achieving automated annotation. This method effectively improves the efficiency and accuracy of annotation, avoids the subjective bias of manual annotation, and provides a reliable geometric benchmark for longitudinal calibration. To verify the effectiveness of the algorithm, three typical tunnel cross-sections were selected for testing: a horseshoe-shaped test section, an irregular-shaped test section, and a shield curve test section. The TSAM polygon fitting method proposed in this scheme was used to automatically adjust the contour polygons of each cross-section, followed by longitudinal calibration. This calibration is based on dynamic Kalman filtering and cubic spline interpolation to ensure uniform distribution of scan lines, and circumferential calibration is achieved through projection center transformation and the fixed-distance equal division algorithm proposed in this scheme. This achieves precise distance segmentation of the perimeter of the fitted polygon and accurate mapping of pixels to physical distances, ultimately generating an orthophoto grayscale image of the tunnel inner wall point cloud. This fixed-distance equal division algorithm, as a unique innovation of this scheme, can strictly divide the perimeter of arbitrary-shaped cross-sections equally, avoids the approximate error of traditional angle projection, ensures that each pixel corresponds to a fixed physical distance, and supports high-precision area measurement.

[0132] like Figure 11 As shown, the original and calibrated images of three cross-sections are compared: the left sub-image shows significant distortion in the original image along the velocity-mileage and cross-sectional directions, resulting in disproportionate length ratios, blurred object edges, and poor identification of tunnel wall ancillary facilities, failing to meet the needs of defect identification or area measurement; the right sub-image presents the calibrated image, with uniform scan line distribution, sharp contours, and significantly enhanced identification, clearly reflecting the geometric details and surface features of the tunnel wall. The test sections of this scheme are mainly straight sections or gentle curves with small curvature, so curve offset has a limited impact on the experimental results. However, subway tunnels commonly have sections with small turning radii and large curvature. In such scenarios, relying solely on relative measurements of laser point clouds easily introduces attitude errors. To overcome this limitation, the tunnel laser scanning system motion detection equipment developed in this scheme has reserved an IMU interface and installation space, supporting multimodal data fusion. In subsequent schemes, an IMU will be integrated based on the existing platform to achieve absolute attitude correction, and a dedicated point cloud calibration algorithm will be developed for small-radius, large-curvature sections, thereby comprehensively meeting the monitoring needs of complex tunnel environments.

[0133] like Figure 12As shown, this image compares the local details of the tunnel interior walls before and after calibration for three different cross-sectional shapes (horseshoe, irregular, and circular shield tunnels). The differences between the measured length and area values ​​and the actual values ​​are marked. Before calibration, the measured results showed significant deviations and blurred details due to uneven longitudinal mileage and circumferential projection distortion, making it impossible to accurately identify ancillary facilities or damaged areas. After calibration, the geometric consistency of the image was significantly improved, and the measured results closely matched the actual values, validating the algorithm's effectiveness.

[0134] Meanwhile, to further verify the accuracy, this scheme was evaluated based on two different sets of tunnel ancillary facility area measurement data. The first set of data included 20 samples with known areas, covering horseshoe-shaped tunnels, irregular-shaped tunnels, and shield-type tunnels. The results before and after correction were compared with the true values ​​to quantitatively evaluate the calibration effect of the algorithm. Statistical analysis showed that the mean absolute error of the uncorrected image was 0.1246m. 2 The maximum absolute error reached 0.3930m. 2 The average percentage error was approximately 17.83%, which was large and unstable, failing to meet the requirements for accurate measurement; after correction, the average absolute error was significantly reduced to 0.0099m. 2 The maximum absolute error is 0.0254m. 2 The measured values ​​are very close to the true values, and the overall accuracy is improved by about 14 times. For example... Figure 13 The bar chart shown highlights the significant improvement in accuracy after calibration, further confirming the algorithm's effectiveness in eliminating longitudinal and circumferential distortion. The second set of data, also based on 20 samples, performs statistical analysis on the measurement results, such as... Figure 14 The dotted line diagram shown indicates that the area measurement difference of this algorithm is within 0.0255m. 2 Within 0.0116m, the standard deviation of the error is 0.0116m. 2 With an average percentage error of only 1.27%, the overall accuracy is superior to traditional cylindrical projection methods. These statistical indicators focus on the quantitative analysis of absolute accuracy, demonstrating the good stability of this algorithm under different cross-sectional shapes. It performs well across all cross-sectional shapes, exhibiting a uniform error distribution without systematic bias, and achieving centimeter-level area measurement accuracy (equivalent to pixel-level geometric precision). Therefore, the orthophotos generated using this method are of excellent quality, have high calibration accuracy, and possess reliable area measurement capabilities. They can be widely applied in engineering scenarios such as tunnel defect identification, ancillary facility statistics, and structural performance evaluation.

[0135] This solution focuses on the core challenges of constructing orthophoto grayscale images of tunnels of arbitrary shapes using mobile laser scanning point cloud data, and proposes a high-precision reconstruction and calibration algorithm framework. This framework integrates data preprocessing, longitudinal calibration, and circumferential calibration modules to achieve distortion-free image construction and area measurement. Experimental results verify the algorithm's robustness and centimeter-level accuracy under complex cross-sections. The main features of this application are: proposing a longitudinal calibration method based on automated sleeper annotation, dynamic Kalman filtering, and cubic spline interpolation; using sleeper spacing as a geometric reference; achieving uniform distribution of scan lines and correction of mileage errors; supporting distortion-free longitudinal mapping; and improving the accuracy of defect length measurement. A circumferential calibration algorithm based on the Tunnel Cross-Section Adaptive Fitting Method (TSAM) and the fixed-distance equal division method is proposed. The innovations of this scheme are TSAM and the fixed-distance equal division method: TSAM recursively retains key points and adaptively adjusts the threshold through a dynamic threshold simplification strategy, combined with energy function-driven vertex smoothing optimization, to generate polygonal contours suitable for circular, horseshoe-shaped, and irregular cross-sections; the fixed-distance equal division method, as the core of circumferential calibration, calculates the perimeter of the fitted polygon and generates a sequence of equal division points by interpolating the accumulated arc length along the contour, combined with centroid angle calculation and bilinear interpolation to allocate gray values, achieving accurate mapping between pixels and real space, supporting high-precision area measurement, and applicable to image generation for various cross-sectional shapes. Field testing was conducted on a multi-section section of Chongqing Metro Line 16, and the results showed that image distortion was eliminated after calibration, and the absolute error of area measurement was <0.01m. 2 With a percentage error of <1.3%, the centimeter-level defect identification accuracy meets the needs of tunnel ancillary facility statistics and leakage survey, improving the efficiency of service status assessment.

[0136] To achieve the above objectives, the present invention also provides a high-precision generation and processing system for tunnel orthophotos based on adaptive fitting and fixed-distance equal division, such as... Figure 16As shown, the system is used to implement the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and fixed-distance equal division. The system specifically includes: a first data processing and generation unit, used to generate and acquire first data corresponding to the tunnel, preprocess the first data, and generate second data corresponding to the first data; wherein the first data is the original point cloud data inside the tunnel, the point cloud data including spatial coordinate information data and reflection intensity information data; the second data is initial grayscale image data; a second data processing and generation unit, used to process the second data longitudinally and circumferentially, respectively, and generate third data corresponding to the second data; wherein the third data is geometrically distortion-free tunnel orthophoto grayscale image data; the longitudinal calibration includes combining dynamic filtering and cubic spline interpolation methods to correct and adjust the mileage data and scan line distribution state corresponding to the first data; the circumferential calibration includes using an adaptive fitting method for the tunnel cross-section to fit and generate a polygon corresponding to the tunnel cross-section; and using a fixed-distance equal division method to divide the perimeter of the polygon at a fixed distance, and generating centroid angles corresponding to the division points and centroids.

[0137] The data preprocessing includes at least one of point cloud denoising, point cloud thinning, and laser intensity correction; the laser intensity correction uses a piecewise linear stretching algorithm to convert the intensity information of the lidar into grayscale values ​​and enhance the corresponding contrast.

[0138] The second data processing and generation unit further includes: a first creation processing module, used to create a first mapping relationship between image row direction pixels and tunnel longitudinal mileage, and using the inherent sleeper structure in the tunnel as a geometric reference, to correct and process the uneven distribution of scan lines and mileage accumulation errors caused by fluctuations in the carrier's moving speed through dynamic filtering and interpolation algorithms; and a second creation processing module, used to create a second mapping relationship between image column direction pixels and tunnel circumferential arc length, and to adaptively fit the tunnel cross-sectional contour into a polygon, calculate the centroid of the polygon as the projection center, and segment the perimeter of the polygon based on a fixed-distance equal division algorithm.

[0139] And / or, the first creation processing module further includes: a first processing module, configured to automatically detect and locate the edge or centerline corresponding to the sleeper based on the second data, and constrain and adjust the detected sleeper position sequence based on the standard spacing or known actual spacing of the sleepers, and construct a high-precision sleeper mileage sequence corresponding to the sleepers; a second processing module, configured to use the sleeper mileage sequence as the observation value, combine it with the carrier motion model, and dynamically correct the original mileage value corresponding to the lidar scanning section through the Kalman filtering algorithm; a third processing module, configured to combine the cubic spline interpolation function, and based on the corrected mileage point data, resample the scanning line, and generate a point cloud section sequence that is uniformly distributed in the longitudinal direction and has accurate mileage.

[0140] And / or, the second creation processing module further includes: a fourth processing module, used for recursively simplifying the single-section point cloud data corresponding to the tunnel, and dynamically adjusting the distance threshold for determining whether to retain points during the recursion process; wherein, in each recursion, for the current set of points to be simplified, the perpendicular distance from all points to the lines connecting their first and last endpoints is calculated, only the point with the largest distance is retained as the key vertex of the polygon, and the point set is divided into two subsets for recursive processing based on this point; the threshold gradually decreases with the increase of the recursion depth; a fifth processing module, used to generate and obtain fourth data corresponding to the tunnel, and to smoothly optimize the vertex positions corresponding to the fourth data by minimizing an energy function; wherein, the fourth data is the initial polygon vertex data obtained in the recursive simplification; the energy function includes at least one term for constraining the consistency between the area of ​​the optimized polygon and the coverage area of ​​the original point cloud, and one term for constraining the smoothness of the polygon side length; the first generation module A block is used to generate and acquire fifth data corresponding to the tunnel cross-section contour, and generate corresponding sixth data based on a preset target resolution; wherein, the fifth data is the total perimeter data of the polygon obtained by adaptive fitting; the target resolution is the actual physical length represented by each pixel; the sixth data is the height data of the image after circumferential calibration, i.e., the total number of pixel rows; a sixth processing module is used to perform cumulative arc length equal division along the polygon boundary according to the preset physical length, starting from one vertex of the polygon, and generate corresponding seventh data through linear interpolation; wherein, the seventh data is the coordinate data of a series of equal division points; a second generation module is used to calculate and generate the centroid angle corresponding to the line connecting adjacent equal division points and the centroid of the polygon, and map each point in the original cross-section point cloud to the corresponding equal division interval according to the normalized angle corresponding to its connection with the centroid, and at the same time assign the reflection intensity information of the point to the corresponding pixel of the equal division interval.

[0141] In the system embodiment of the present invention, the specific details of the method steps involved in the high-precision generation and processing of tunnel orthophotos based on adaptive fitting and fixed-distance equal division have been described above. That is to say, the functional modules in the system are used to implement the steps or sub-steps in the above method embodiment, which will not be repeated here.

[0142] To achieve the above objectives, the present invention also provides a high-precision generation and processing platform for tunnel orthophotos based on adaptive fitting and fixed-distance equal division, such as... Figure 17 As shown, the system includes a processor, a memory, and a control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division. The processor executes the control program, which is stored in the memory. This control program implements the steps of the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and equal interval division. The specific details of these steps have been described above and will not be repeated here.

[0143] In this embodiment of the invention, the built-in processor of the high-precision generation and processing platform for tunnel orthophotos based on adaptive fitting and fixed-distance division can be composed of integrated circuits. For example, it can be composed of a single packaged integrated circuit, or it can be composed of multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor connects to various components through various interfaces and lines, and executes programs or units stored in the memory, and calls data stored in the memory to perform various functions and process data for the high-precision generation and processing of tunnel orthophotos based on adaptive fitting and fixed-distance division. The memory is used to store program code and various data, is installed in the high-precision generation and processing platform for tunnel orthophotos based on adaptive fitting and fixed-distance division, and achieves high-speed and automatic access to programs or data during operation. The memory includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0144] To achieve the above objectives, the present invention also provides a computer-readable storage medium, such as... Figure 18As shown, the computer-readable storage medium stores a control program for a high-precision generation and processing platform of tunnel orthophotos based on adaptive fitting and equal interval division. This control program implements the steps of the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and equal interval division. In other words, a computer-readable storage medium is provided, storing a computer program that, when executed by a processor, implements the steps of the high-precision generation and processing method for tunnel orthophotos based on adaptive fitting and equal interval division.

[0145] Specifically, the storage medium is a non-transitory computer-readable medium, which may be solid-state flash memory, SD card, or hard disk, and has a shock-resistant design to adapt to the tunnel vibration environment. The storage medium stores an executable instruction set, which includes: image data preprocessing code, laser intensity correction code, longitudinal and circumferential calibration code, and equal division processing code. When the instruction set is loaded and executed by the processor, steps S001-S002 can be implemented sequentially, and configuration parameters can be adjusted via INI file. The storage medium is compatible with embedded operating systems and has an error handling mechanism to ensure reliability, facilitating engineering replication and maintenance.

[0146] In the description of embodiments of the present invention, it should be noted that any process or method description in the flowcharts or otherwise described herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the present invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order according to the functions involved, as should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0147] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processing module, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, a “computer-readable medium” can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, the computer-readable medium can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0148] This invention generates and acquires first data corresponding to a tunnel through a method, and preprocesses the first data to generate second data corresponding to the first data. The first data is raw point cloud data inside the tunnel, including spatial coordinate information and reflection intensity information. The second data is initial grayscale image data. The second data is processed by longitudinal and circumferential calibration to generate third data corresponding to the second data. The third data is geometrically distortion-free tunnel orthophoto grayscale image data. The longitudinal calibration includes combining dynamic filtering and cubic spline interpolation to correct and adjust the mileage data and scan line distribution corresponding to the first data. The circumferential calibration includes using an adaptive fitting method for the tunnel cross-section to generate a polygon corresponding to the tunnel cross-section; and using a fixed-distance equal division method to divide the perimeter of the polygon at fixed distances, generating centroid angles corresponding to the division points and centroids. The invention also includes a corresponding system, platform, and storage medium, enabling high-precision longitudinal and circumferential calibration simultaneously. It is applicable to tunnels of any shape and can generate orthophotos with strict geometric mapping relationships and support centimeter-level precision measurements.

[0149] In other words, by integrating longitudinal calibration with dynamic Kalman filtering and cubic spline interpolation, and circumferential calibration based on adaptive polygon fitting and equidistant division, this invention successfully reconstructs a geometrically distortion-free tunnel orthophoto image from a moving laser scanning point cloud. This achieves a strict mapping between pixels and physical distances, enabling the measurement accuracy of the length and area of ​​tunnel surfaces of arbitrary shapes to reach the centimeter level. It completely solves the problems of image distortion and measurement inaccuracy caused by uneven speed, projection center offset, and shape limitations in traditional methods.

[0150] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A high-precision generation processing method for tunnel orthographic images based on adaptive fitting and equidistant division, characterized in that, The method comprises: generating and acquiring first data corresponding to the tunnel, preprocessing the first data, and generating second data corresponding to the first data; wherein the first data is original point cloud data inside the tunnel, and the point cloud data comprises spatial coordinate information data and reflection intensity information data; and the second data is initial gray image data; respectively longitudinally and circumferentially calibrating the second data, and generating third data corresponding to the second data; wherein the third data is tunnel orthographic gray image data without geometric distortion; the longitudinal calibration comprises combining dynamic filtering processing and a cubic spline interpolation method to correct and adjust the mileage data and the scanning line distribution state corresponding to the first data; the circumferential calibration comprises using a tunnel section self-adaptive fitting method to fit and generate a polygon corresponding to the tunnel section; and using a fixed-distance equal-division method to divide the perimeter of the polygon at a fixed distance and generate a centroid angle corresponding to the division points and the centroid.

2. The high-precision generation processing method of tunnel orthographic images based on adaptive fitting and equidistant division according to claim 1, characterized in that, The data preprocessing comprises at least one of point cloud denoising, point cloud thinning, and laser intensity correction; The laser intensity correction uses a piecewise linear stretching algorithm to convert the intensity information of the laser radar into a gray value and enhance the corresponding contrast. 3.The tunnel orthographic image high-precision generation processing method based on adaptive fitting and equidistant division according to claim 1, characterized in that, The longitudinal and circumferential calibration of the second data to generate the third data corresponding to the second data further comprises: creating a first mapping relationship between image row direction pixels and tunnel longitudinal mileage, and taking the inherent sleeper structure inside the tunnel as a geometric reference to correct and process the uneven scanning line distribution and mileage accumulation error caused by the fluctuation of the carrier movement speed through dynamic filtering and interpolation algorithms; creating a second mapping relationship between image column direction pixels and tunnel circumferential arc length, adaptively fitting the tunnel section profile into a polygon, calculating the centroid of the polygon as the projection center, and dividing the perimeter of the polygon based on the fixed-distance equal-division algorithm.

4. The high-precision generation processing method of tunnel orthographic images based on adaptive fitting and equidistant division according to claim 3, characterized in that, The creation of the first mapping relationship between the image row direction pixels and the tunnel longitudinal mileage, and the taking of the inherent sleeper structure inside the tunnel as a geometric reference to correct and process the uneven scanning line distribution and mileage accumulation error caused by the fluctuation of the carrier movement speed through dynamic filtering and interpolation algorithms further comprises: automatically detecting and locating the edges or center lines corresponding to the sleepers according to the second data, and based on the standard spacing or the known actual spacing of the sleepers, respectively constraining and adjusting the detected sleeper position sequence, and constructing a high-precision sleeper mileage sequence corresponding to the sleepers; taking the sleeper mileage sequence as the observation value, combining the carrier motion model, and dynamically correcting the original mileage value corresponding to the laser radar scanning section through the Kalman filtering algorithm; combining the cubic spline interpolation function and based on the corrected mileage point data, resampling the scanning line and generating a point cloud section sequence that is uniformly distributed in the longitudinal direction and has accurate mileage.

5. The high-precision generation processing method of tunnel orthographic images based on adaptive fitting and equidistant division according to claim 3, characterized in that, The second mapping relationship between the image column direction pixel and the tunnel ring direction arc length is created, the tunnel section profile is adaptively fitted as a polygon, the centroid of the polygon is calculated as a projection center, and the polygon perimeter is segmented based on a constant distance equal division algorithm, and the system further comprises: Recursive simplification processing of single-section point cloud data corresponding to the tunnel is performed, and in the recursive process, a distance threshold for judging whether to retain a point is dynamically adjusted; wherein in each recursion, for a current point set to be simplified, the perpendicular distance of all points to the connecting line of the first and last points is calculated, only the point with the maximum distance is retained as a key vertex of the polygon, and the point set is segmented into two subsets based on the point for recursive processing; the threshold gradually decreases with the increase of the recursion depth; Fourth data corresponding to the tunnel is generated and obtained, and the vertex position corresponding to the fourth data is smoothed and optimized by minimizing an energy function; wherein the fourth data is the initial polygon vertex data obtained by recursive simplification; the energy function at least includes a term for constraining the consistency of the polygon area after optimization and the original point cloud coverage area, and a term for constraining the smoothness of the polygon side length.

6. The high-precision generation processing method of tunnel orthographic images based on adaptive fitting and equidistant division according to claim 3 or 5, characterized in that, The second mapping relationship between the image column direction pixel and the tunnel ring direction arc length is created, the tunnel section profile is adaptively fitted as a polygon, the centroid of the polygon is calculated as a projection center, and the polygon perimeter is segmented based on a constant distance equal division algorithm, and the system further comprises: Fifth data corresponding to the tunnel section profile is generated and obtained, and sixth data corresponding to the target resolution is generated based on a preset target resolution; wherein the fifth data is the total perimeter data of the polygon obtained by adaptive fitting; the target resolution is the actual physical length represented by each pixel; the sixth data is the height data of the ring direction calibrated image, i.e. the total number of pixel rows; A vertex of the polygon is taken as a starting point, cumulative arc length equal division processing is performed along the polygon boundary according to the preset physical length, and corresponding seventh data is generated by linear interpolation calculation; wherein the seventh data is coordinate data of a series of equal division points; The centroid angle corresponding to the connecting line between the adjacent equal division points and the polygon centroid is calculated, and each point in the original section point cloud is mapped into the corresponding equal division interval according to the normalized angle corresponding to the connecting line between the point and the centroid, and the reflection intensity information of the point is assigned to the pixel corresponding to the equal division interval.

7. A high-precision generation processing system for tunnel orthographic images based on adaptive fitting and equidistant division, characterized in that, The system is used to realize the tunnel orthographic image high-precision generation processing method based on adaptive fitting and constant distance equal division according to any one of claims 1-6, and the system comprises: A first data processing generation unit is configured to generate and obtain first data corresponding to a tunnel, and to preprocess the first data to generate second data corresponding to the first data; wherein the first data is original point cloud data inside the tunnel, and the point cloud data includes spatial coordinate information data and reflection intensity information data; and the second data is initial gray scale image data; The second data processing generation unit is configured to longitudinally calibrate and circularly calibrate the second data respectively, and generate third data corresponding to the second data; the third data is tunnel orthographic gray scale image data without geometric distortion; the longitudinal calibration comprises dynamic filtering processing and a cubic spline interpolation method, and adjusts and processes mileage data and scanning line distribution states corresponding to the first data; the circular calibration comprises tunnel section adaptive fitting to generate a polygon corresponding to a tunnel section; and a constant distance equal division method is used to divide the polygon periphery at a fixed distance, and a centroid angle corresponding to a centroid and a division point is generated.

8. The high-precision generation processing system of the tunnel orthographic image based on adaptive fitting and equidistant division according to claim 7, characterized in that, The data preprocessing comprises at least one of point cloud denoising, point cloud thinning and laser intensity correction; The laser intensity correction uses a piecewise linear stretching algorithm to convert intensity information of the laser radar into a gray scale value and enhance the corresponding contrast; The second data processing generation unit further comprises: A first creation processing module is configured to create a first mapping relationship between image row direction pixels and tunnel longitudinal mileage, and take inherent track structure in the tunnel as a geometric reference, and correct and process uneven scanning line distribution and mileage accumulation error caused by carrier movement speed fluctuation through dynamic filtering and interpolation algorithms; A second creation processing module is configured to create a second mapping relationship between image column direction pixels and tunnel circular arc length, and adaptively fit tunnel section profiles to polygons, calculate a centroid of the polygon as a projection center, and divide the polygon periphery based on a constant distance equal division algorithm; And / or, the first creation processing module further comprises: A first processing module is configured to automatically detect and locate edges or center lines corresponding to the track based on the second data, and respectively constrain and adjust detected track position sequences based on standard track spacing or known actual track spacing, and construct track mileage sequences corresponding to the track and having high precision; A second processing module is configured to take the track mileage sequences as observation values, combine a carrier motion model, and dynamically correct original mileage values corresponding to laser radar scanning sections through a Kalman filtering algorithm; A third processing module is configured to combine a cubic spline interpolation function, and resample the scanning lines based on corrected mileage point data, and generate point cloud section sequences that are uniformly distributed in the longitudinal direction and have accurate mileage; And / or, the second creation processing module further comprises: A fourth processing module is configured to recursively simplify single section point cloud data corresponding to the tunnel, and dynamically adjust a distance threshold for judging whether to retain points in the recursive process; in each recursion, for a point set to be simplified, the perpendicular distance of all points to a line connecting the first and last points is calculated, only the point with the largest distance is retained as a key vertex of the polygon, and the point set is divided into two subsets based on the point for recursive processing; the threshold gradually decreases with the increase of the recursive depth; The fifth processing module is configured to generate and acquire fourth data corresponding to the tunnel, and to perform smoothing optimization on vertex positions corresponding to the fourth data by minimizing an energy function; wherein the fourth data is initial polygon vertex data obtained by recursive simplification; and the energy function includes at least one term for constraining consistency between an area of the optimized polygon and an area covered by the original point cloud, and one term for constraining smoothness of a length of the polygon edge; The first generating module is configured to generate and acquire fifth data corresponding to a tunnel section profile, and to generate corresponding sixth data based on a preset target resolution; wherein the fifth data is total perimeter data of the polygon obtained by adaptive fitting; the target resolution is an actual physical length represented by each pixel; and the sixth data is height data of the image after ring direction calibration, i.e., a total number of pixel rows; The sixth processing module is configured to take a vertex of the polygon as a starting point, perform cumulative arc length equal division processing along a polygon boundary at the preset physical length, and generate corresponding seventh data by linear interpolation calculation; wherein the seventh data is coordinate data of a series of equal division points; The second generating module is configured to calculate and generate a centroid angle corresponding to a line connecting a neighboring equal division point and a polygon centroid, and to map each point in the original section point cloud into a corresponding equal division interval according to a normalized angle corresponding to the line connecting the point and the centroid, while assigning reflection intensity information of the point to a pixel corresponding to the equal division interval.

9. A high-precision generation processing platform for tunnel orthographic images based on adaptive fitting and equidistant division, characterized in that, The computer readable storage medium stores a tunnel orthographic image high-precision generation processing platform control program based on adaptive fitting and fixed-distance equal division, and the tunnel orthographic image high-precision generation processing platform control program based on adaptive fitting and fixed-distance equal division realizes the tunnel orthographic image high-precision generation processing method based on adaptive fitting and fixed-distance equal division as claimed in any one of claims 1 to 6.

10. A computer readable storage medium, characterized in that, The computer readable storage medium stores a tunnel orthographic image high-precision generation processing platform control program based on adaptive fitting and fixed-distance equal division, and the tunnel orthographic image high-precision generation processing platform control program based on adaptive fitting and fixed-distance equal division realizes the tunnel orthographic image high-precision generation processing method based on adaptive fitting and fixed-distance equal division as claimed in any one of claims 1 to 6.

Images