A prediction method for quantifying the development degree of erosion disease of rammed earth cultural relics

Abstract

A predictive method for quantifying the degree of erosion damage in rammed earth cultural relics, relating to the field of cultural relic protection, involves acquiring 3D laser point cloud data and close-up photogrammetric 3D model data of the rammed earth relics; fusing the 3D laser point cloud model and the close-up photogrammetric 3D model to construct a three-dimensional model of erosion damage in the rammed earth relics; extracting quantitative indicators of erosion damage development based on the three-dimensional model of erosion damage in the rammed earth relics, constructing and optimizing a predictive model of erosion damage development; and generating and outputting a quantitative report on the degree of erosion damage development in the rammed earth relics. This invention combines 3D laser scanning and close-up photogrammetry, designs differentiated flight paths for different height areas, aligns the cross-sectional contours of different models based on Fourier description, and uses Poisson surface reconstruction technology combined with a hole-filling algorithm to complete the missing parts of the model, intuitively presenting the spatial distribution and morphological characteristics of erosion damage, quantifying the changing patterns of erosion damage over time, and promoting a shift in protection work from passive restoration to proactive prevention.

Description

Technical Field

[0001] This invention relates to the field of cultural relic protection technology, and more specifically, to a method for predicting the degree of erosion damage in rammed earth cultural relics. Background Technology

[0002] Rammed earth cultural relics such as ancient city walls, imperial tombs, and ancient settlement ruins, constructed using local soil and simple ramming techniques, suffer from poor material strength and durability, making erosion one of their most common and destructive problems. Erosion is caused by natural or human factors that gradually erode and hollow out the interior or surface of the rammed earth, forming holes, grooves, honeycomb structures, and even through-cracks. This condition directly threatens the structural safety of the cultural relic and can easily lead to irreversible collapse.

[0003] Because rammed earth cultural relics are porous, highly porous, and sensitive to the environment, their internal cavities are difficult to pinpoint precisely, posing a significant challenge to the investigation and safety risk assessment of erosion damage. Currently, the industry lacks a quantitative indicator system that can accurately reflect the degree of erosion damage. In terms of quantitative prediction of the development degree of erosion damage, existing prediction methods are mostly based on single factors or simple models, failing to fully consider the complex coupling effect between natural environmental factors and the characteristics of rammed earth materials (such as particle composition and pore structure). This results in low accuracy and reliability of prediction results, making it difficult to provide precise guidance for the protection of rammed earth cultural relics in practical applications.

[0004] Meanwhile, for tall rammed earth sites, erosion damage may exist in different locations such as the top, middle, and bottom, and deep surface damage that is difficult to detect can easily occur, usually requiring high-altitude operations and large-scale exploration. Traditional manual survey methods have problems such as poor accessibility at high altitudes, low safety, incomplete damage records, and low quantitative accuracy; while ground-based three-dimensional laser scanning can achieve millimeter-level quantitative accuracy, it will have blind spots at high altitudes and a large number of data gaps; while UAV close-up photogrammetry can comprehensively grasp the overall situation of erosion damage, it has the limitation of limited detail accuracy.

[0005] Therefore, the existing technology has problems and needs further improvement and development. Summary of the Invention

[0006] (I) Purpose of the invention: In order to solve the problems existing in the prior art, the purpose of this invention is to provide a method for predicting the degree of erosion damage of rammed earth cultural relics, so as to solve the problems of incomplete investigation of erosion damage of rammed earth cultural relics, low quantitative accuracy and poor prediction accuracy in the prior art, and to achieve accurate quantitative description and reliable development prediction of erosion damage of rammed earth cultural relics, so as to provide scientific guidance for the safety protection and restoration of cultural relics.

[0007] (II) Technical Solution: To solve the above-mentioned technical problems, this technical solution provides a method for predicting the degree of erosion damage to rammed earth cultural relics, including:

[0008] Step 1: Obtain 3D laser point cloud data and close-up photographic 3D model data of rammed earth cultural relics;

[0009] Step 2: Fuse the 3D laser point cloud model with the close-up photographic 3D model to construct a three-dimensional model of the erosion damage of rammed earth cultural relics;

[0010] Step 3: Based on the three-dimensional model of erosion damage to rammed earth cultural relics, extract quantitative indicators of erosion damage development, and construct and optimize a erosion damage development prediction model.

[0011] Step 4: Generate and output a quantitative report on the degree of erosion damage to rammed earth cultural relics.

[0012] A method for predicting the degree of erosion damage in rammed earth cultural relics, wherein step 1 includes:

[0013] Step 101: Obtain data on rammed earth cultural relics;

[0014] Step 102: Normalize the coordinate system of the collected rammed earth cultural relics data.

[0015] A predictive method for quantifying the degree of erosion damage to rammed earth cultural relics, wherein the flight path of a drone is designed based on the minimum image resolution of rammed earth cultural relic damage being ≥0.5cm / pixel, and the flight trajectory must be equidistant from the surface of the site, with the forward overlap and lateral overlap of the flight path being greater than 80%.

[0016] A predictive method for quantifying the degree of erosion damage to rammed earth cultural relics involves importing denoised 3D point cloud data, performing a custom geographic transformation based on a calculated seven-parameter model, and then performing a projection transformation by selecting Gauss-Kruger projection parameters according to the location of the survey area, outputting 3D laser point cloud model data in the CGCS2000 coordinate system.

[0017] A method for predicting the degree of erosion damage in rammed earth cultural relics, wherein step 2 includes:

[0018] Step 201: Preprocess the 3D laser point cloud data and the close-up photographic 3D texture data to generate a 3D laser point cloud model and a close-up photographic 3D model.

[0019] Step 202: Extract the cross-sections of the 3D laser point cloud model and the close-up photographic 3D model and match them;

[0020] Step 203: Match and align multiple cross-sectional profiles to determine the optimal alignment position of the cross-sectional profiles;

[0021] Step 204: Fit the parameters of multiple cross-sections to generate a fusion model;

[0022] Step 205: Based on the fusion model, the surface morphology of the cultural relic is fitted using Poisson surface reconstruction technology, and the boundary of the eroded cavity is completed by combining the hole filling algorithm to generate a three-dimensional model of the erosion disease of the rammed earth cultural relic.

[0023] A method for predicting the degree of erosion damage to rammed earth cultural relics involves using the cross-sectional profile of a three-dimensional laser point cloud model to correspond to the cross-sectional profile of a close-range photogrammetric three-dimensional model with a height deviation less than or equal to the absolute elevation accuracy of the photogrammetry. The overlap is calculated using a shape matching algorithm to find the optimal alignment position.

[0024] A predictive method for quantifying the degree of erosion damage in rammed earth cultural relics, wherein, after locating the boundary of erosion damage by cutting plane, Poisson surface reconstruction technology is used based on the overall shape characteristics of the rammed earth cultural relics. With the overall shape of the cultural relics as a constraint, a global implicit function is constructed, and effective information from three-dimensional laser point cloud data and close-up photographic three-dimensional data is integrated to generate a triangular mesh surface. The erosion damage exists in the overall model in the form of cavities, and its boundary contour is naturally connected with the overall surface.

[0025] To address the tiny holes present after Poisson reconstruction, a hole-filling algorithm is used to automatically detect unclosed boundary loops in the triangular mesh surface, record the vertex coordinates and normal vectors of the hole boundaries, and use a local surface interpolation method to generate triangular patches to fill the holes, ensuring a smooth transition between the filled surface and the original surface.

[0026] The final model is rendered using 3D modeling software. The main body of the cultural relic is set to semi-transparent, while the eroded areas are set to solid color, showing the spatial distribution and morphological characteristics of the erosion inside the cultural relic, thus forming a three-dimensional model of the erosion damage of the rammed earth cultural relic.

[0027] A method for predicting the degree of erosion damage in rammed earth cultural relics, wherein step 3 includes:

[0028] Step 301: Based on the three-dimensional model of erosion damage in rammed earth cultural relics, extract the geometric characteristic indicators of erosion damage development.

[0029] Step 302: Based on the fusion data of geometric characteristic indicators of erosion disease development at different times in the same area, construct a erosion disease development prediction model.

[0030] Step 303: Optimize the prediction model for the development of erosion disease.

[0031] A method for predicting the degree of erosion damage in rammed earth cultural relics, wherein the geometric characteristic indicators of erosion damage development include erosion range, maximum erosion depth, and erosion damage volume; the erosion range refers to the boundary of the damage model; the maximum erosion depth refers to the difference between the minimum Z-axis value of the erosion damage model and the original surface or fitted surface; and the erosion damage volume refers to the volume of a three-dimensional irregular shape space.

[0032] A method for predicting the degree of erosion damage to rammed earth cultural relics, comprising collecting environmental data at different time points, including light intensity, temperature, wind speed, and rainfall, and calculating the actual rate of change of erosion damage indicators within each time interval:

[0033] Erosion rate = Δrange / time interval;

[0034] Erosion depth propagation rate = Δmaximum depth / time interval;

[0035] Erosion volumetric rate = Δvolume / time interval;

[0036] By comparing the actual disease data in time series with the simulated data of the erosion disease development prediction model, the rate of change is adjusted by multiple based on environmental influences, thereby adjusting the rate of change of the erosion range, maximum erosion depth, and erosion disease volume in the erosion disease development prediction model.

[0037] A method for predicting the degree of erosion damage to rammed earth cultural relics, wherein the quantitative report includes a three-dimensional model of erosion damage to rammed earth cultural relics, annual disease index change curves, a prediction model for the future development of erosion damage, and a risk level assessment table.

[0038] (III) Beneficial Effects: This invention provides a predictive method for quantifying the development degree of erosion damage in rammed earth cultural relics. Firstly, it combines the advantages of 3D laser scanning and close-up photogrammetry, designing differentiated flight paths for different height areas of tall rammed earth sites to ensure the comprehensiveness and accuracy of erosion damage data collection. Secondly, it proposes a multi-section fitting model fusion method, achieving high-precision alignment of cross-sectional contours of different models based on Fourier descriptor matching, avoiding global matching errors caused by surface roughness, and providing a high-quality data foundation for quantifying damage. Thirdly, it employs Poisson surface reconstruction technology to construct a three-dimensional model of erosion damage, combining it with a hole-filling algorithm to complete the missing parts of the model. Through visualization, the spatial distribution and morphological characteristics of erosion damage are presented intuitively, providing a clear reference for quantitative analysis. Finally, it constructs a time-based linear regression model to quantify the changes in erosion range, maximum depth, and erosion damage volume over time, while adjusting model parameters based on environmental factors to improve prediction accuracy. Attached Figure Description

[0039] Figure 1This is a flowchart of the steps in the present invention for predicting the degree of erosion damage to rammed earth cultural relics.

[0040] Figure 2 This is a schematic diagram of the photogrammetric flight path planning of the present invention;

[0041] Figure 3 This is a schematic diagram of a three-dimensional point cloud model closely resembling a photogrammetric model, based on the present invention.

[0042] Figure 4 This is a schematic diagram of the process for constructing a three-dimensional model of erosion damage to rammed earth cultural relics according to the present invention;

[0043] Figure 5 This is a schematic diagram illustrating the erosion damage of rammed earth cultural relics after multi-section fitting according to the present invention.

[0044] Figure 6 This is a schematic diagram of a three-dimensional model of the erosion damage to rammed earth cultural relics according to the present invention;

[0045] Figure 7 This is a schematic diagram illustrating the change of the erosion range of the present invention over time;

[0046] Figure 8 This is a schematic diagram illustrating the change of the maximum erosion depth of the present invention over time;

[0047] Figure 9 This is a schematic diagram illustrating the change in the volume of the erosion lesion over time according to the present invention;

[0048] Figure 10 This is a schematic diagram illustrating the influence of environmental factors on the rate of change of the prediction model for the development of erosion diseases. Detailed Implementation

[0049] The present invention will be further described in detail below with reference to preferred embodiments. More details are set forth in the following description in order to provide a full understanding of the present invention. However, the present invention can obviously be implemented in many other ways different from those described herein. Those skilled in the art can make similar extensions and derivations based on actual application situations without departing from the spirit of the present invention. Therefore, the scope of protection of the present invention should not be limited by the content of this specific embodiment.

[0050] The accompanying drawings are schematic diagrams of embodiments of the present invention. It should be noted that these drawings are for illustrative purposes only and are not drawn to scale, and should not be construed as limiting the actual scope of protection of the present invention.

[0051] A method for predicting the degree of erosion damage in rammed earth cultural relics, such as... Figure 1 As shown, the specific steps include:

[0052] Step 1: Obtain the 3D laser point cloud data and close-up photographic 3D model data of the rammed earth cultural relics.

[0053] Step 2: Integrate the 3D laser point cloud model with the close-up photographic 3D model to construct a three-dimensional model of the erosion damage of rammed earth cultural relics.

[0054] Step 3: Based on the three-dimensional model of erosion damage to rammed earth cultural relics, extract quantitative indicators of erosion damage development, and construct and optimize a erosion damage development prediction model.

[0055] Step 4: Generate and output a quantitative report on the degree of erosion damage to rammed earth cultural relics.

[0056] Specifically, step 1 includes:

[0057] Millimeter-precision 3D point cloud data of rammed earth cultural relics were obtained by ground-based 3D laser scanning, while UAV close-up photogrammetry technology was used to collect surface texture data of rammed earth cultural relics.

[0058] Step 101: Obtain data on rammed earth cultural relics.

[0059] Data was collected from rammed earth cultural relics using a stand-alone terrestrial 3D laser scanner to obtain high-precision 3D point cloud raw data of the relics. The raw data coordinate system is WGS84 and includes planar position and geodetic height H.

[0060] like Figure 2 As shown, close-up photogrammetry follows the basic idea of ​​starting from scratch and from coarse to fine. First, a generalized model is quickly established by measuring key points on site. Combined with flight path planning and attitude control models, the flight path and photography attitude are quickly planned. Then, high-definition images obtained by close-up photography of the object surface are used for photogrammetry to obtain the precise coordinates and fine shape of the object. Finally, 3D models, digital elevation models (DEMs), and digital orthophoto (DOM) data with sub-centimeter or even millimeter precision are obtained.

[0061] Based on the minimum identification accuracy requirement of rammed earth cultural relics with an image resolution ≥0.5cm / pixel, the drone flight path was designed, and the flight trajectory needed to be equidistant and parallel to the surface of the site, with a forward and lateral overlap of more than 80%. Aerial image data was collected according to the drone flight path, and at the same time, close-up photogrammetry trajectory planning software was used to intelligently plan the photographic target from multiple angles, and to take multiple vertical shots close to the cultural relic itself.

[0062] Where resolution = (sensor pixel size × flight altitude) / focal length.

[0063] Based on the camera's height from the ground, the data is divided into three categories: the bottom of rammed earth artifacts (approximately 2 meters), the middle section (2-15 meters), and the top section (over 15 meters). According to the vertical distance from the site surface to the camera, the data is further divided into four categories: bottom parallel flight path, middle parallel flight path, top grid, and edge encircling flight path.

[0064] The bottom parallel flight path targets a height of approximately 2 meters at the bottom of the rammed earth site. The camera is perpendicular to the surface of the rammed earth artifacts at the bottom, with a flight speed of 3–4 m / s, a directional overlap of 85%, and a relative height of 1.5–2 meters to the bottom surface. The waist parallel flight path targets a height of 2–15 meters at the waist of the rammed earth site. The camera is perpendicular to the surface of the rammed earth artifacts at the waist, with a flight speed of 4–5 m / s, a directional overlap of 85%, and a relative height of 2–2.5 meters to the waist surface. The top grid and edge circling flight path targets a height of 15 meters or more at the top of the rammed earth site. The camera is perpendicular to the surface of the rammed earth artifacts at the top, with a flight speed of 4–5 m / s, a directional overlap of 80%, and a relative height of 2.5–3 meters to the bottom surface.

[0065] Step 102: Normalize the coordinate system of the collected rammed earth cultural relics data.

[0066] The scanned point cloud data is imported into professional data processing software for automatic seamless stitching and noise reduction. With the help of known control point coordinates, the coordinate system is normalized.

[0067] Specifically, the denoised 3D point cloud data is imported into ArcGIS software, a custom geographic transformation is performed based on the solved seven-parameter model, and then the Gauss-Kruger projection parameters are selected according to the location of the survey area for projection transformation, outputting 3D laser point cloud model data in the CGCS2000 coordinate system.

[0068] More specifically, the collected 3D point cloud data is imported into professional data processing software for automatic seamless stitching and noise reduction. Isolated noise points during scanning are removed by filtering, and invalid height ranges are marked if there are scanning blind spots.

[0069] Select more than three high-precision common control points in the survey area, use a total station to measure the local coordinate system coordinates of the common control points, and obtain the CGCS2000 coordinates of the common control points through the foundation augmentation system; input the WGS84 original coordinates and CGCS2000 coordinates of the common control points into the data processing software, and use a seven-parameter model to calculate the coordinate transformation parameters, as shown in formula (1).

[0070] (1),

[0071] The seven-parameter model includes three translation parameters ΔX, ΔY, and ΔZ, three rotation parameters RX, RY, and RZ, and one scale factor m.

[0072] The common control points are known to have their CGCS2000 coordinates and WGS84 original coordinates. Precision control primarily involves ensuring the accuracy of the coordinates of the three common control points. Three common control points are selected, and their local coordinate system coordinates are measured using a high-precision total station. CGCS2000 coordinates are obtained using a ground augmentation system. The original coordinates and CGCS2000 coordinates of the common control points are then input, and seven parameters are calculated.

[0073] Select characteristic points and measure their CGCS2000 coordinates using a high-precision total station. Calculate the root mean square error. If the root mean square error is ≤5mm, it meets the surveying accuracy requirements. If the root mean square error exceeds the limit, the parameters need to be recalculated or the accuracy of the common control points needs to be checked.

[0074] By acquiring aerial imagery data and using close-up photogrammetry trajectory planning software to intelligently plan multi-angle shots of the photographic target, and taking multiple vertical angle shots close to the subject, the software is used to stitch, automatically perform aerial triangulation and modeling to generate high-precision close-up photogrammetric 3D texture data.

[0075] Specifically, such as Figure 4 As shown, step 2 specifically includes:

[0076] Step 201: Preprocess the 3D laser point cloud data and the close-up photographic 3D texture data to generate a 3D laser point cloud model and a close-up photographic 3D model.

[0077] like Figure 3 As shown, basic processing is performed on the acquired 3D laser point cloud model and the close-up photographic 3D model to ensure that the 3D laser point cloud model has completed denoising and coordinate system transformation.

[0078] The acquired aerial imagery data was imported into 3D modeling software for stitching, automatic aerial triangulation, and modeling processing to generate a textured triangular mesh-based close-up 3D model. The close-up 3D model underwent preprocessing to remove redundant triangular faces, retaining the macroscopic outline, and its geometric accuracy was ensured through actual size verification. The close-up 3D model has completed redundant triangular face removal and geometric accuracy verification, and the height coordinate system of the 3D laser point cloud model and the close-up 3D model is consistent, i.e., the z-axis is consistent.

[0079] Step 202: Extract the cross-section of the 3D laser point cloud model and the close-up 3D photographic model and match them.

[0080] In point cloud processing software, a cross-sectional section is cut into the 3D laser point cloud model to extract point sets at different heights. Then, point set fitting is used to transform the discrete points into polygonal contours. In 3D modeling software, a cross-sectional cutting tool is used to cut the close-up photogrammetric 3D model, obtaining polygonal contours at different heights. The cross-sectional contour of the 3D laser point cloud model at a certain height is then compared to the cross-sectional contour of the close-up photogrammetric 3D model at the corresponding height. A shape matching algorithm is used to calculate the overlap and find the optimal alignment position.

[0081] The profile height deviation must be less than or equal to the absolute elevation accuracy of the close-proximity photogrammetry; otherwise, the profile will deviate from the same actual height level. The absolute elevation accuracy of the close-proximity photogrammetry is typically at the centimeter level. Currently, the elevation accuracy obtained from the three control points is within 5cm. Therefore, the height threshold range should be within 5cm above and below the specified height deviation to avoid profile misalignment due to insufficient model height accuracy. In the point cloud processing software, the laser point cloud model is cross-sectionally cut every 0.5m to extract the laser contours of 20 height profiles, which are then fitted into polygonal contours. In the 3D modeling software, the close-proximity photogrammetric 3D model is cut at the same height to extract the polyline contours of 20 photogrammetric contours.

[0082] Extracting the target contours involves extracting the cross-sectional contours a of the 3D laser point cloud profile A and the close-up photographic 3D model profile B, which need to be aligned, respectively, through edge detection and threshold segmentation. i and b i This lays the foundation for alignment. Specifically, based on a 3D laser point cloud model, a set of polygonal contours A={a1,a2,...,a...} at different heights is extracted. n For each 3D laser point cloud profile contour a i Sampling yields an ordered pixel coordinate sequence Pa i ={(xa1,ya1),(xa2,ya2),...,(xa n ,ya n The number and height range of cross-sections to be extracted are clearly defined. For a close-up photographic 3D model, a set of polygonal contours B={b1,b2,...,b} corresponding to the height range of A is extracted. n}, for each photographic 3D model profile b i Sampling yields an ordered pixel coordinate sequence Qb i ={(xb1,yb1),(xb2,yb2),...,(xb n ,yb n )}.

[0083] Select the initial cross-section a1, and define the cross-sectional outline b of the close-up photographic 3D model at the same height ±5cm above and below a1. iSpecifically, press a. i With b i The standard is a height deviation of ≤5cm. For each laser profile, a corresponding photographic profile is matched to check for any instances where the 3D laser point cloud model's profile does not have a matching photographic 3D model profile that meets the deviation requirement. If any unmatched laser profiles are found, the coordinate system of the matching photographic model is readjusted, the z-axis coordinate is fine-tuned, and the verification is repeated until all 20 laser profiles have found valid matching photographic profiles, laying the foundation for subsequent matching.

[0084] The initial position cross-sections a1 and b i The closed contour is sampled into N ordered pixels, and keypoints are randomly sampled to obtain the coordinate sequence (x1, y1), (x2, y2), ..., (x...). N y N Perform a one-dimensional Fourier transform on the coordinate sequence to obtain the Fourier coefficients F in the x and y directions. x(k) and F y(k) , where k=0,1,...,N-1, and is normalized to retain only the first M low-frequency coefficients, while high-frequency components are filtered out to ignore local details.

[0085] In this invention, the low-frequency coefficient is selected as M=350. The specific number needs to be adjusted according to the complexity of the target. For complex contours, the number can be increased appropriately.

[0086] Step 203: Match and align multiple cross-sectional profiles to determine the optimal alignment position of the cross-sectional profiles.

[0087] Select an initial cross-section a1 from the 3D laser point cloud profile contour A, and then select contour b from the 3D photographic model profile contour B that has a height deviation from a1 within ±5cm. i .

[0088] For the initial position a1 and the corresponding b i The Euclidean distance between the two images is calculated using the normalized low-frequency coefficients. The smaller the Euclidean distance, the higher the similarity. The closer the correlation coefficient is to 1, the higher the similarity. The initial position a1 traverses the possible poses of the image b1 to be aligned, including rotation angles of 0~360° and scaling ratios of 0.8~1.2, and finds the pose with the highest similarity in the Fourier descriptors. The poses are then overlapped to find the optimal alignment position.

[0089] Optimal alignment position a i Outline and b i Outlines, calculate the pixel-level overlap between the two, overlap = (a i Outline and b i (Area of ​​the intersection region of the contours) / (a i Outline and b iIf the area of ​​the union region of the outlines is the highest, then the alignment is effective.

[0090] A multi-initial-position alignment comparison is adopted. Three to five different initial positions are set for A, and coarse and fine alignment mechanisms are executed to calculate the overlap. The results of each overlap rate are compared, and the position with the highest overlap rate is selected. Feature regions are used for verification. Points in key areas such as cracks and edges are given higher weights, forcing these areas to align. First, key areas such as cracks are aligned and matched, then the alignment is extended to the whole, and the overlap rate is calculated for each region. If the overlap rate of each region is ≥80%, it is considered the globally optimal position, and the overall alignment is considered complete.

[0091] The overlap calculation uses shape spectrum matching of the Fourier descriptor matching method, which focuses on macroscopic shape and is not sensitive to the translation, rotation and scaling of the contour. Even if there is a local break in the contour, as long as the overall shape is intact, the low frequency component can still stably reflect the features. The core is the overall contour alignment, which is best suited for cross-sectional matching of rammed earth cultural relics.

[0092] Step 204: Fit the parameters of multiple cross sections to generate a fusion model.

[0093] Matching parameters for a single height only reflect the alignment relationship of that plane. To avoid random errors from a single height, it is necessary to fit parameters from as many profiles as possible. By comprehensively fitting the matching results of multiple profiles, the overall superposition relationship of the two models can be obtained, thereby improving the reliability of model fusion.

[0094] For the 20 cross sections involved in the fitting, each section is based on the 3D laser point cloud profile. The measured parameters of the i-th section are obtained through precise coordinate measurement and calculation of feature points, with the translation in the x-direction being Δx. i The translation in the y-direction is Δy i and the rotation angle about the z-axis is θ i This ensures that each profile parameter reflects the local alignment relationship.

[0095] The least squares method is constructed to solve the objective function S, as shown in formula (2), and the overall superposition parameters are solved.

[0096] (2),

[0097] Where n represents the total number of cross-sections involved in the fitting; Δx i Δy represents the measured x-direction translation parameter of the i-th profile; i θ represents the measured y-direction translation parameter of the i-th profile; i This represents the measured rotation parameters of the i-th profile.

[0098] Find the values ​​of S respectively By taking the partial derivatives of the model and setting them to 0, we can obtain the global optimal fitting parameters, which are the parameters representing the overall superposition relationship between the two models.

[0099] The globally optimal parameters are solved. As parameters for the overall superposition of the 3D laser point cloud model and the close-up photogrammetric 3D model, coordinate transformation is performed on the close-up photogrammetric 3D model to align it with the laser point cloud model in 3D space. After superposition, a fused model is formed, providing a unified data foundation for subsequent reconstruction of erosion diseases.

[0100] Twenty height profile matching parameters were fitted, with the mean fluctuation of Δx, Δy, and θ ≤ 2mm / 0.5°. Coordinate transformation was performed on the close-up 3D model. The data was fused with an initial threshold of 10mm. Local repetition points were found at the top. The threshold was reduced to 8mm. The fusion result had no repetition points or unfused areas, and the model fusion was completed, generating a fused model with both high accuracy and integrity.

[0101] Specifically, theoretically, Δx, Δy, and θ at each height should be basically consistent. Based on the fitted superposition parameters, the coordinate transformation of the close-up photogrammetry 3D model is performed to align it with the laser point cloud model in 3D space. After superposition, the model can be further verified and the parameters optimized through height profiles. When there are two types of points in the neighborhood of a certain point, namely 3D laser scanning and close-up photogrammetry, the 3D laser scanning is retained. When there is only one type of point, it is directly retained, thereby achieving data fusion.

[0102] The high-density point cloud accuracy of the laser scanner is 2-5 mm, and the accuracy of the data acquired from the close-up photogrammetric 3D model is 3-8 mm. A threshold of 1-1.5 times the sum of these two data accuracies, typically 8-12 mm, is used as the initial threshold. This ensures that only truly overlapping points are retained by the laser scanner, avoiding the accidental deletion of valid supplementary points from the close-up photogrammetric 3D model. After overlaying and fusion, the model is observed for points with excessively high thresholds (repeated points at the same location) or points with excessively low thresholds (unfused points at points that should overlap). If repeated points exist in some areas, the threshold is reduced by 10%-20% each time. If unfused areas exist, the threshold is increased by 10%-20% each time. This iterative optimization and repeated verification continues until the fusion result meets the requirements.

[0103] Step 205: Based on the fusion model, the surface morphology of the cultural relic is fitted using Poisson surface reconstruction technology, and the boundary of the eroded cavity is completed by combining the hole filling algorithm to generate a three-dimensional model of the erosion disease of the rammed earth cultural relic.

[0104] Based on the differences in the characteristics of rammed earth cultural relics' diseases, and focusing on the core feature of erosion-induced cavities, the erosion-induced diseases are marked by combining the honeycomb-like features of the cavities inside the cross-section of the fused model. Figure 5 As shown, the fusion model after annotating the erosion disease is obtained.

[0105] The damage to the rammed earth cultural relics includes surface weathering and salt precipitation, which are uniform surface damage; cracks are linear fractures; biological coverage mostly stays on the surface or in the shallow layer; and human damage often shows regular or sudden signs of destruction.

[0106] Based on the fusion model annotated with erosion defects, the boundaries of erosion defects are located using a section plane. Specifically, based on the fusion model, and considering the irregular nature of erosion boundaries, the boundary line between normal and eroded areas is automatically identified using a point cloud density threshold. Multiple planar boundaries are superimposed to form an internal spatial bounding box, and a continuity requirement of ≥85% overlap between adjacent boundaries is proposed, providing precise constraints for surface reconstruction. The point cloud density threshold is 20 points / cm².

[0107] In normal rammed earth areas, the point cloud density is uniform, while the eroded cavities are devoid of point clouds. The boundary between these two areas represents the cross-sectional boundary of the erosion within the cutting plane. For unclosed boundaries within the triangular mesh surfaces of the eroded cavities, local surface interpolation is used to generate triangular facets to fill the holes. The normal vectors of the newly generated triangular facets are adjusted using a weighted average to ensure a smooth transition between the filled and original surfaces. By superimposing the cross-sectional boundaries of multiple cutting planes, a spatial internal boundary framework for the erosion damage is formed, creating a three-dimensional model of the erosion damage to the rammed earth cultural relic.

[0108] By integrating 3D laser point cloud data with close-up photogrammetric 3D models, the depth, extent, volume, and spatial location of the erosion can be quantified. After locating the boundary of the erosion by cutting planes, Poisson surface reconstruction technology is used based on the overall shape characteristics of the existing rammed earth artifacts. With the overall shape of the artifacts as constraints, a global implicit function is constructed. By integrating the effective information from 3D laser point cloud data and close-up photogrammetric 3D data, a triangular mesh surface is generated. The erosion exists in the overall model in the form of cavities, and its boundary contours are naturally connected to the overall surface.

[0109] Specifically, the formula for identifying the boundary of the eroded section is shown in formula (3), where 1 represents the eroded area and 0 represents the normal area.

[0110] (3),

[0111] Where ρ(x,y,z) represents the density function of the fused point cloud; B(x,y,z) represents the region determination result.

[0112] Extract the erosion boundary point set P={p1,p2,...,p n}, each point p i Each has a corresponding normal vector n iUsing the overall shape of the cultural relic as a constraint, a global implicit function covering the entire point cloud region is constructed. In the reconstruction of the Poisson surface, the normal vector information of the point cloud is regarded as the source term. The space is discretized by constructing an octree structure. The normal vectors of the surrounding point cloud are interpolated to construct a local vector field. The isosurface of f(x,y,z)=0 in the scalar field is extracted to obtain the surface model of the erosion disease of the rammed earth cultural relic.

[0113] To address the micro-holes present after Poisson reconstruction, a hole-filling algorithm was employed to automatically detect unclosed boundary loops in the triangular mesh surface. The vertex coordinates and normal vectors of the hole boundaries were recorded. Local surface interpolation was used to generate triangular facets to fill the holes. The normal vectors of the newly generated facets were adjusted using a weighted average to ensure a smooth transition between the filled and original surfaces. The final model was rendered using 3D modeling software, employing a contrast between transparency and solidity. The main body of the artifact was set to semi-transparent, while the eroded areas were set to solid color, visually demonstrating the spatial distribution and morphological characteristics of the erosion within the artifact, thus creating a three-dimensional model of the erosion damage to the rammed earth artifact. Figure 6 As shown.

[0114] Specifically, step 3 includes:

[0115] Step 301: Based on the three-dimensional model of erosion damage in rammed earth cultural relics, extract the geometric characteristic indicators of erosion damage development.

[0116] Based on the three-dimensional model of erosion damage to rammed earth cultural relics that has been obtained, the geometric characteristics of the main indicators of erosion damage development are extracted, including the erosion range, the maximum erosion depth, and the volume of erosion damage.

[0117] The erosion range refers to the boundary of the disease model, that is, the range of the eroded surface in the surface reconstruction. The maximum erosion depth refers to the difference between the minimum Z-axis value of the eroded disease model and the original surface or the fitted surface. The erosion volume refers to the volume calculation of a three-dimensional irregular shape space. It can be viewed from multiple perspectives such as top view, side view, and bottom view through operations such as rotation at any angle, local magnification, and cross-section cutting, to understand the specific situation of erosion with safety risks such as rammed layer detachment, tilting fracture, and internal disease extension.

[0118] Step 302: Based on the fusion data of geometric characteristic indicators of erosion disease development at different times in the same area, construct a prediction model for erosion disease development.

[0119] A linear regression model was constructed by fusing data on erosion disease development in the same region in chronological order. Based on the fused data of erosion disease at different time points in the same region, a predictive model for erosion disease development was built. This model mainly includes the changing patterns of three-dimensional morphological indicators of the disease at different time points, namely, the quantified changes in erosion range, maximum erosion depth, and erosion disease volume over time.

[0120] More specifically, such as Figure 7 - Figure 9 As shown, the data of erosion damage at different time points within multiple years of the same rammed earth cultural relics were obtained, and the erosion range L, maximum erosion depth d, and erosion damage volume V at each time point were extracted. With time t as the independent variable, a time series curve was constructed, and a linear regression model was used to describe the change law of the disease indicators with time, as shown in formulas (4) to (6):

[0121] L(t)=A1t+B1(4)

[0122] d(t)=A2t+B2(5)

[0123] V(t)=A3t+B3(6)

[0124] Where A1 is the annual expansion rate of the erosion range, in mm / year and cm / year; B1 is the initial erosion range at the start of monitoring, in mm and cm; A2 is the annual deepening rate of the maximum erosion depth, in mm / year and cm / year; B2 is the initial maximum erosion depth at the start of monitoring, in mm and cm; A3 is the annual growth rate of the erosion volume, in mm³ / year and cm³ / year; B3 is the initial erosion volume at the start of monitoring, in mm³ and cm³.

[0125] The intercepts B1, B2, and B3 in formulas (4) to (6) are collectively referred to as B below, as shown in formula (7):

[0126] (7),

[0127] in, It is the average value at a given time point; It is the average value of the corresponding disease indicators.

[0128] The intercepts A1, A2, and A3 in formulas (4) to (6) are collectively referred to as A below, as shown in formula (8):

[0129] (8),

[0130] Where n is the number of monitoring times, t i It is the i-th time point, y i It is the disease index value at the i-th time point.

[0131] According to formulas (4) to (8), input the future time to calculate the predicted value, extrapolate the boundary of the current erosion area according to the predicted rate, combine the predicted volume and adjust the morphology at the maximum depth to generate a erosion disease development prediction model.

[0132] Step 303: Optimize the prediction model for the development of erosion disease.

[0133] Under relatively stable environmental conditions, the changes in the erosion range L(t), maximum erosion depth d(t), and erosion volume V(t) of a cultural relic over the same time interval can be represented by a linear function with respect to time t. However, the actual environment of cultural relics is often affected by various factors such as light, temperature, wind, and rain. Therefore, when there are differences between actual and predicted / simulated damage, the impact of these environmental factors on the environment should also be considered. Figure 10 As shown, the prediction model for erosion disease development is optimized based on environmental data at different time points.

[0134] Collect environmental data at different time points, including light intensity, temperature, wind speed, and rainfall, and calculate the actual rate of change of erosion disease indicators within each time interval:

[0135] Erosion rate = Δrange / time interval;

[0136] Erosion depth propagation rate = Δmaximum depth / time interval;

[0137] Erosion volume rate = Δvolume / time interval.

[0138] By comparing the actual disease data in time series with the simulated data of the erosion disease development prediction model, the reasons for the differences are analyzed. The influence range of environmental factors is defined by combining the actual situation of each environmental factor in the environmental data. The rate of change is adjusted by an appropriate multiple according to the environmental impact. The rate of change of erosion range, maximum erosion depth and erosion disease volume in the erosion disease development prediction model are adjusted.

[0139] Specifically, step 4 includes:

[0140] Input a future time, calculate the predicted value through the optimized model, extrapolate the boundary of the current erosion area according to the predicted rate, and adjust the shape by combining the predicted erosion volume and maximum depth to complete the prediction of the development degree of erosion of rammed earth cultural relics, and generate and output a quantitative report on the development degree of erosion of rammed earth cultural relics.

[0141] The quantitative report includes a three-dimensional model of erosion damage to rammed earth cultural relics, annual disease index change curves, a future erosion damage development prediction model, and a risk level assessment table.

[0142] The three-dimensional visualization model of the erosion damage uses a contrast rendering of transparency and solidity, with the main body of the cultural relic set to semi-transparent and the eroded areas set to solid color.

[0143] The annual disease index change curves include trend graphs of the erosion range L(t), maximum depth d(t), and volume V(t) over time.

[0144] The future time erosion disease development prediction model and risk level assessment table are marked with high risk areas in red, medium risk areas in yellow, and low risk areas in green. The risk level is determined by the predicted annual growth rate of erosion disease volume > 5 cm³ / year, medium risk areas are 2 cm³ to 5 cm³ / year, and low risk areas are < 2 cm³ / year.

[0145] This invention presents a predictive method for quantifying the development degree of erosion damage in rammed earth cultural relics. First, it combines the advantages of 3D laser scanning and close-up photogrammetry, designing differentiated flight paths for different height areas of tall rammed earth sites. This solves the problems of poor accessibility at high altitudes in traditional manual surveys, blind spots in high-altitude laser scanning, and insufficient detail accuracy in photogrammetry, ensuring the comprehensiveness and accuracy of erosion damage data collection. Second, it proposes a multi-section fitting model fusion method, achieving high-precision alignment of cross-sectional contours of different models based on Fourier descriptor matching. This avoids global matching errors caused by surface roughness and is suitable for the stable macroscopic morphology and complex microscopic details of rammed earth cultural relics, providing a high-quality data foundation for quantifying damage. Third, it employs Poisson surface reconstruction technology to construct a three-dimensional model of erosion damage, combining it with a hole-filling algorithm to complete missing parts of the model. Through visualization, the spatial distribution and morphological characteristics of erosion damage are presented intuitively, providing a clear reference for quantitative analysis. Finally, a time-based linear regression model was constructed to quantify the changes in the erosion range, maximum depth, and erosion damage volume over time. At the same time, environmental factors were incorporated to adjust the model parameters, improve prediction accuracy, and promote the transformation of rammed earth cultural relic protection from passive restoration to proactive prevention.

[0146] The above description illustrates preferred embodiments of the present invention and helps those skilled in the art to more fully understand the technical solution of the present invention. However, these embodiments are merely illustrative and should not be construed as limiting the specific implementation of the present invention to these embodiments. For those skilled in the art, several simple deductions and modifications can be made without departing from the inventive concept, and all such modifications should be considered within the protection scope of the present invention.

Claims

1. A method for predicting the degree of erosion damage in rammed earth cultural relics, characterized in that, include: Step 1: Obtain 3D laser point cloud data and close-up photographic 3D model data of rammed earth cultural relics; Step 2: Preprocess the 3D laser point cloud data and the close-up photographic 3D texture data to generate a 3D laser point cloud model and a close-up photographic 3D model. Extract the cross sections of the 3D laser point cloud model and the close-up photographic 3D model. Achieve high-precision alignment of multiple cross section contours based on Fourier descriptors. Use the cross section contour of the 3D laser point cloud model to correspond to the cross section contour of the close-up photographic 3D model whose height deviation is less than or equal to the absolute accuracy of the elevation measured by the close-up photograph. Calculate the overlap degree through the shape matching algorithm, find the optimal alignment position, fit the parameters of multiple cross sections, and realize the fusion of the 3D laser point cloud model and the close-up photographic 3D model. Use Poisson surface reconstruction technology to fit the surface morphology of the cultural relics, and combine it with the hole filling algorithm to complete the boundary of the eroded cavity, and construct a three-dimensional model of the erosion disease of the rammed earth cultural relics. Step 3: Based on the three-dimensional model of erosion damage to rammed earth cultural relics, extract quantitative indicators of erosion range, maximum erosion depth, and erosion volume. Use a linear regression model to construct a time-series prediction model for erosion damage development, and optimize the prediction model for erosion damage development by combining environmental data. Step 4: Generate and output a quantitative report on the degree of erosion damage to rammed earth cultural relics. The quantitative report includes a three-dimensional model of erosion damage to rammed earth cultural relics, annual disease index change curves, a future erosion damage development prediction model, and a risk level assessment table.

2. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 1, characterized in that, Step 1 includes: Step 101: Obtain data on rammed earth cultural relics; Step 102: Normalize the coordinate system of the collected rammed earth cultural relics data.

3. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 2, characterized in that, The drone flight path is designed based on the minimum image resolution of ≥0.5cm / pixel for identifying defects in rammed earth cultural relics. The flight path must be equidistant from the surface of the site, and the directional and lateral overlap of the flight path must be greater than 80%.

4. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 2, characterized in that, Import the denoised 3D point cloud data, perform a custom geographic transformation based on the solved seven-parameter model, and then perform a projection transformation according to the Gauss-Kruger projection parameters selected based on the location of the survey area, outputting 3D laser point cloud model data in the CGCS2000 coordinate system.

5. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 1, characterized in that, After locating the boundary of the erosion disease by cutting the plane, based on the overall shape characteristics of the rammed earth cultural relic, the Poisson surface reconstruction technology is adopted. With the overall shape of the cultural relic as a constraint, a global implicit function is constructed, and the effective information of the three-dimensional laser point cloud data and the close-up photogrammetric three-dimensional data is integrated to generate a triangular mesh surface. The erosion disease exists in the overall model in the form of a cavity, and its boundary contour is naturally connected with the overall surface. To address the tiny holes present after Poisson reconstruction, a hole-filling algorithm is used to automatically detect unclosed boundary loops in the triangular mesh surface, record the vertex coordinates and normal vectors of the hole boundaries, and use a local surface interpolation method to generate triangular patches to fill the holes, ensuring a smooth transition between the filled surface and the original surface. The final model is rendered using 3D modeling software. The main body of the cultural relic is set to semi-transparent, while the eroded areas are set to solid color, showing the spatial distribution and morphological characteristics of the erosion inside the cultural relic, thus forming a three-dimensional model of the erosion damage of the rammed earth cultural relic.

6. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 1, characterized in that, Step 3 includes: Step 301: Based on the three-dimensional model of erosion damage in rammed earth cultural relics, extract the geometric characteristic indicators of erosion damage development. Step 302: Based on the fusion data of geometric characteristic indicators of erosion disease development at different times in the same area, construct a erosion disease development prediction model. Step 303: Optimize the prediction model for the development of erosion disease.

7. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 6, characterized in that, The erosion range refers to the boundary of the disease model; the maximum erosion depth refers to the difference between the minimum Z-axis value of the eroded disease model and the original surface or the fitted surface; the erosion volume refers to the volume of a three-dimensional irregular shape space.

8. The method for predicting the degree of erosion damage in rammed earth cultural relics according to claim 6, characterized in that, Environmental data at different time points were collected, including light intensity, temperature, wind speed, and rainfall, and the actual rate of change of erosion disease indicators within each time interval was calculated. By comparing the actual disease data in time series with the simulated data of the erosion disease development prediction model, the rate of change is adjusted by multiple based on environmental influences, thereby adjusting the rate of change of the erosion range, maximum erosion depth, and erosion disease volume in the erosion disease development prediction model.

Images