A data processing method based on underwater concrete scanning reconstruction

Abstract

The application belongs to the field of data processing and relates to a data processing method based on underwater concrete scanning reconstruction, which comprises the following steps: obtaining scanning data based on underwater concrete, performing adaptive noise filtering on the scanning data, performing point cloud density equalization on the scanning data after adaptive noise filtering, performing geometric feature enhancement on the scanning data after point cloud density equalization, performing data completion and repair on the scanning data after geometric feature enhancement, performing multi-view registration on the scanning data after data completion and repair, and performing surface reconstruction on the scanning data after multi-view registration. Through surface reconstruction, the scanning data after registration is converted into a continuous and smooth three-dimensional surface, thereby realizing accurate digital reconstruction of the underwater concrete structure. The method not only improves the data processing efficiency and accuracy, but also enhances the realism and practicality of the reconstructed model, thereby providing strong technical support for underwater engineering detection, maintenance and design.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a data processing method based on underwater concrete scanning reconstruction. BACKGROUND

[0002] The detection and evaluation of underwater concrete structures are important issues in the fields of marine engineering, water conservancy engineering, and bridge engineering. With the development of three-dimensional scanning technology, underwater concrete detection methods based on acoustic or optical scanning have been widely applied. However, due to the particularity of the underwater environment, there are many problems with the data obtained by scanning, resulting in insufficient reconstruction accuracy and affecting subsequent analysis and evaluation.

[0003] Current data processing for underwater concrete scanning reconstruction mainly has the following defects: severe noise interference: suspended solids, bubbles, and water flow disturbances in the water medium result in a large number of noise points in the scanning data; uneven point cloud density: due to changes in scanning angle and distance, as well as the attenuation effect of water on the signal, the point cloud density distribution is extremely uneven; blurred geometric features: the water medium leads to a decrease in scanning resolution, and key features such as cracks and holes on the concrete surface become blurred; severe data loss: water absorption, reflection, and multi-path effects result in data loss in some areas; insufficient registration accuracy: traditional ICP algorithms have poor adaptability in underwater environments, and multi-view scanning data registration errors are large; low reconstruction efficiency: the existing processing flow has high computational complexity, making it difficult to meet the real-time requirements of engineering. SUMMARY

[0004] To solve the above technical problems, the present application provides a data processing method based on underwater concrete scanning reconstruction, which adopts the following technical solutions:

[0005] Obtain scanning data based on underwater concrete, and perform adaptive noise filtering on the scanning data;

[0006] Perform point cloud density equalization on the scanning data after adaptive noise filtering;

[0007] Perform geometric feature enhancement on the scanning data after point cloud density equalization;

[0008] Perform data completion and repair on the scanning data after geometric feature enhancement;

[0009] Perform multi-view registration on the scanning data after completion and repair;

[0010] Perform surface reconstruction on the scanning data after multi-view registration.

[0011] Preferably, the step of obtaining scanning data based on underwater concrete and performing adaptive noise filtering on the scanning data specifically includes:

[0012] Acquire scanning data based on underwater concrete;

[0013] The DBSCAN algorithm is used to classify noise based on the characteristics of the water medium. ,in For point Energy value, For point and Euclidean distance, The attenuation coefficient of the water medium. Distance weighting factor;

[0014] By combining wavelet transform and statistical filtering, noise is eliminated at different scales. ,in These are wavelet coefficients. The original signal, For wavelet basis functions, For scale parameters, These are displacement parameters;

[0015] Bilateral filtering is used to protect geometric features. ,in The intensity of the filtered point. As the normalization factor, For the spatial domain Gaussian kernel, For the range Gaussian kernel, It is a set of neighboring points.

[0016] Preferably, the step of performing point cloud density equalization on the scan data after adaptive noise filtering specifically includes:

[0017] Establish a density field model based on kernel density estimation: ,in For position Density estimation at [location] The number of sample points. For bandwidth parameters, For data dimensions, For kernel functions;

[0018] Non-uniform resampling based on density field: ,in For the newly generated points, These are the weighting coefficients. The number of neighboring points;

[0019] Preserving geometric characteristics using the Laplacian operator: ,in For point Laplace coordinates, for The neighborhood point set.

[0020] Preferably, the step of performing geometric feature enhancement on the scanned data after point cloud density equalization specifically includes:

[0021] Integrating curvature, normal, and texture features: ,in For comprehensive eigenvalues, Curvature characteristics Characteristics of normal changes For texture features, These are the weighting coefficients;

[0022] Feature enhancement is performed using an improved U-Net network: ,in For the total loss function, To rebuild the losses, To preserve loss at the edge, For structural similarity loss, For loss weights;

[0023] Feature-constrained surface optimization: ,in For the optimized vertex set, For the original vertex, For feature constraint operators, These are the balancing parameters.

[0024] Preferably, the step of data completion and repair of the scanned data after geometric feature enhancement specifically includes:

[0025] Predicting missing data using underwater acoustic propagation models: ,in Distance The sound intensity at that location, The initial sound intensity, The absorption coefficient is... For reference distance;

[0026] Non-rigid registration based on bio-inspired algorithms is employed: ,in For transformation function, For regularization terms, For smoothing parameters;

[0027] Hole-filling algorithm combining structural features: ,in For the area with holes, It is the normal vector. For geometric distance.

[0028] Preferably, the step of multi-view registration of the scan data after completion and repair specifically comprises:

[0029] Initial registration using FPFH features: wherein is the FPFH feature of point is the angle feature, is the distance feature, is the scale parameter; Adaptive ICP algorithm under water:

[0030] wherein is the rotation matrix, is the translation vector, is the confidence weight; Global optimization based on pose graph:

[0031] wherein is the node pose, is the relative pose measurement, is the covariance matrix. Preferably, the step of surface reconstruction of the scan data after multi-view registration specifically comprises:

[0032] Spatial division based on point cloud density:

[0033] wherein is the decision of node, is the number of points in the node, is the threshold value; Improved Poisson reconstruction algorithm:

[0034] wherein is the indicator function, is the vector field; Hierarchical mesh optimization:

[0035] wherein is the mesh of the layer, is the point cloud data, is the Laplace operator. In order to solve the above technical problems, the application also provides a data processing device based on underwater concrete scanning reconstruction, which adopts the technical scheme as follows:

[0036] Filtering module, for obtaining scanning data based on underwater concrete, and performing adaptive noise filtering on the scanning data;

[0037]

[0038] ​An equalization module is configured to perform point cloud density equalization on the scanning data after adaptive noise filtering;

[0039] An enhancement module is configured to perform geometric feature enhancement on the scanning data after point cloud density equalization;

[0040] A repair module is configured to perform data completion and repair on the scanning data after geometric feature enhancement;

[0041] A registration module is configured to perform multi-view registration on the scanning data after completion and repair;

[0042] A reconstruction module is configured to perform surface reconstruction on the scanning data after multi-view registration.

[0043] To solve the above technical problems, the present application also provides a device, which adopts the technical solution as follows: a memory and a processor, the memory stores computer readable instructions, and the processor executes the computer readable instructions to realize the steps of the data processing method based on underwater concrete scanning reconstruction.

[0044] To solve the above technical problems, the present application also provides a computer readable storage medium, which adopts the technical solution as follows: the computer readable storage medium stores computer readable instructions, and the computer readable instructions are executed by a processor to realize the steps of the data processing method based on underwater concrete scanning reconstruction.

[0045] Compared with the prior art, the present application has the following beneficial effects:

[0046] The scanning data of underwater concrete is obtained by high-precision equipment, and then adaptive noise filtering technology is used to effectively remove random noise in the data and improve data quality;

[0047] Point cloud density equalization processing is performed to ensure consistent density distribution of scanning data in different areas, laying a solid foundation for subsequent processing;

[0048] Through geometric feature enhancement, the edges and key shapes of the concrete structure can be highlighted, providing more accurate reference for data completion and repair, and the data completion and repair technology accurately restores the missing or damaged parts to ensure the integrity and accuracy of the scanning data;

[0049] Through multi-view registration, scanning data from different angles is accurately aligned to form a globally consistent three-dimensional model, which is crucial for constructing a three-dimensional model of a complex structure;

[0050] The registered scanning data is converted into continuous and smooth three-dimensional curved surface by curved surface reconstruction technology, so as to realize accurate digital reconstruction of the underwater concrete structure.

[0051] The data processing efficiency and precision are improved, the reality and practicability of the reconstructed model are significantly enhanced, and powerful technical support is provided for underwater engineering detection, maintenance and design. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the scheme in the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0053] Figure 1 is a flow chart of an embodiment of the data processing method based on underwater concrete scanning reconstruction of the present application;

[0054] Figure 2 is a structural schematic diagram of an embodiment of the data processing device based on underwater concrete scanning reconstruction of the present application;

[0055] Figure 3 is a structural schematic diagram of an embodiment of the device of the present application. DETAILED DESCRIPTION

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs; the terminology used in the specification of the application only for the purpose of describing specific embodiments, and is not intended to limit the present application; the specification of the present application and the claims and the above description of drawings, the terms "include" and "have" and any variations thereof, are intended to cover non-exclusive inclusion. The specification of the present application and the claims or the above description of drawings, the terms "first", "second" and the like are used to distinguish different objects, not to describe a specific order.

[0057] In this paper, the "embodiment" means that the specific features, structures or characteristics described in conjunction with the embodiment can be included in at least one embodiment of the present application. The phrase appears in the specification does not necessarily refer to the same embodiment, nor is it independent or alternative to other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0058] In order to make the person skilled in the art better understand the technical scheme of the present application, the technical scheme in the embodiments of the present application will be described clearly and completely in conjunction with the drawings.

[0059] It should be noted that the data processing method based on underwater concrete scanning reconstruction provided by the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the data processing device based on underwater concrete scanning reconstruction is generally arranged in the server / terminal device.

[0060] It should be understood that the number of terminal devices, networks and servers is only illustrative. Any number of terminal devices, networks and servers can be provided according to the implementation needs.

[0061] Embodiment one:

[0062] Please refer to Figure 1 , a flowchart of one embodiment of the data processing method based on underwater concrete scanning reconstruction of the present application is shown. The data processing method based on underwater concrete scanning reconstruction comprises the following steps:

[0063] Step S1, obtaining scanning data based on underwater concrete, and performing adaptive noise filtering on the scanning data.

[0064] In this embodiment, the electronic device (for example, a server / terminal device) on which the data processing method based on underwater concrete scanning reconstruction is run can receive a data processing request based on underwater concrete scanning reconstruction through a wired connection mode or a wireless connection mode. It should be noted that the wireless connection mode can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAXX connection, Zigbee connection, UWB (ultra wideband) connection, and other now known or future developed wireless connection modes.

[0065] In this embodiment, step S1, obtaining scanning data based on underwater concrete, and performing adaptive noise filtering on the scanning data specifically comprises the following steps:

[0066] S11, obtaining scanning data based on underwater concrete.

[0067] The scanning data based on underwater concrete can be obtained by using an underwater concrete apparent laser line three-dimensional scanning measurement method.

[0068] A clean water replacement device is used, which is a sealed structure composed of glass blocks containing clean water, for placing a laser scanning imaging device. This device can replace the turbid water between the camera and the measured object in the laser scanning imaging device, ensuring clear imaging of the measured object.

[0069] Laser line plane calibration is performed, a laser plane equation is constructed, and a light ray tracing model and a multi-medium refraction model are used to fuse the multi-medium refraction model and the camera imaging model, to obtain a normalized model for underwater imaging.

[0070] The obtained underwater concrete apparent laser line image is converted into a corresponding laser line image in air by using a normalization model and a laser plane equation. This step realizes three-dimensional scanning measurement of the underwater concrete apparent laser line.

[0071] By splicing the three-dimensional coordinates of continuous multiple frames of laser lines, the three-dimensional point cloud coordinates of the underwater concrete in the target area can be obtained, thereby completing the acquisition of scanning data.

[0072] S12, using a DBSCAN algorithm, classifying noise based on water medium characteristics , wherein is the energy value of a point , the Euclidean distance between points and , the water medium attenuation coefficient , the distance weight factor , and the number of neighborhood points.

[0073] The DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm is a density-based clustering algorithm that uses the concepts of core points, boundary points, and noise points to find a sufficient number of neighboring points within a specified radius to construct clusters. When processing underwater concrete scanning data, the characteristics of the water medium, such as energy attenuation, propagation speed variation, etc., have a significant impact on noise.

[0074] To achieve this process, the parameters of the DBSCAN algorithm are expanded, including the energy value of a point, the Euclidean distance between two points, the water medium attenuation coefficient, and the distance weight factor. Among them, the energy value of a point reflects the intensity of the signal during propagation, the water medium attenuation coefficient reflects the attenuation characteristics of the signal during underwater propagation, and the distance weight factor is used to adjust the influence of points at different distances on the clustering result.

[0075] By introducing these parameters, the DBSCAN algorithm can more accurately identify noise points in underwater data and distinguish them from valid concrete structure data. This not only improves the accuracy of data processing, but also helps subsequent data analysis and reconstruction work.

[0076] By step S12, using the DBSCAN algorithm and considering the characteristics of the water medium to classify noise, the efficiency and accuracy of data processing can be significantly improved when processing underwater concrete scanning data, providing a solid foundation for subsequent structure reconstruction and analysis.

[0077] ​​S13, combining wavelet transform and statistical filtering, eliminates noise at different scales. ,in These are wavelet coefficients. The original signal, For wavelet basis functions, For scale parameters, For displacement parameters, This is a time parameter.

[0078] Wavelet transform is a mathematical tool used to decompose a signal into components of different frequencies. This is achieved by adjusting the scaling parameter. and displacement parameters It can transform signals at different scales and with different displacements, thereby separating signal components of different frequencies.

[0079] When processing underwater concrete scanning data, wavelet transform can accurately separate useful and noisy components from the signal. Subsequently, statistical filtering further analyzes the separated signal to identify and remove noise points and outliers. This combined approach not only eliminates high-frequency noise but also preserves the signal's detailed features, thereby improving data clarity and accuracy.

[0080] S14 employs bilateral filtering to protect geometric features. ,in Points after filtering The strength, As the normalization factor, For the spatial domain Gaussian kernel, For the range Gaussian kernel, For the neighborhood point set, and For points in a point cloud, and Points and The intensity value.

[0081] Normalization factor Ensure that the sum of all weights is 1. Spatial domain Gaussian kernel. Weights are assigned based on the Euclidean distance between pixels; the closer the pixels, the greater the weight. The value range uses a Gaussian kernel. Weights are assigned based on the differences in pixel grayscale values ​​(or intensity values), with greater weights given to pixels whose grayscale values ​​are closer to each other.

[0082] The role of bilateral filtering is mainly reflected in two aspects: first, it removes noise through smoothing, improving the accuracy of the data; second, it preserves edge features while smoothing, ensuring that the geometric features of the reconstructed underwater concrete structure are complete and clear. This property gives bilateral filtering a significant advantage in underwater concrete scanning reconstruction, providing reliable data support for subsequent numerical model reconstruction and surface morphology analysis.

[0083] Step S2: Perform point cloud density equalization on the scan data after adaptive noise filtering.

[0084] In this embodiment, step S2, which involves performing point cloud density equalization on the scan data after adaptive noise filtering, specifically includes the following steps:

[0085] S21, Establish a density field model based on kernel density estimation: ,in For position Density estimation at [location] The number of sample points. For bandwidth parameters, For data dimensions, For kernel function, For the first 1 sample point.

[0086] Kernel density estimation (KDE) is a nonparametric estimation method used to estimate the probability density function of a random variable. In point cloud processing, KDE can be used to estimate the density at various locations within a point cloud.

[0087] In practice, the kernel function is first determined. Kernel functions include Gaussian kernels, uniform kernels, etc. The kernel type depends on the number of sample points. Bandwidth parameters and data dimensions Calculate each position using the KDE formula Density estimation at [location] The summation operation in the formula iterates through all sample points. Through kernel function Mapping sample points onto the density field, bandwidth parameter By controlling the influence range of the kernel function, the density estimation results can be smoothed.

[0088] The purpose of establishing a density field model is to quantify the density distribution of a point cloud. Using the KDE (Knowledge-Defined Density Model), a continuous density field can be obtained, reflecting the density variations in different regions of the point cloud. This provides a basis for subsequent non-uniform resampling, ensuring that the resampled point cloud is more density-uniform.

[0089] S22, Non-uniform resampling based on density field: ,in is a newly generated point, is a weight coefficient, is a number of neighborhood points, is a density estimation value at the point .

[0090] The purpose of non-uniform resampling based on the density field is to adjust the density distribution of the point cloud to make it more uniform. In implementation, the weight coefficient of each point is calculated according to the density field model . The weight coefficient is inversely proportional to the density estimation value of the point, that is, the weight of the point with higher density is smaller, and the weight of the point with lower density is larger.

[0091] For each newly generated point , its position is calculated by weighted averaging. The weight of weighted averaging is the previously calculated weight coefficient , and the object of weighted averaging is the point in the neighborhood. The selection of neighborhood points can be determined according to the spatial distance or density similarity.

[0092] The role of non-uniform resampling is to balance the density distribution of the point cloud. By adjusting the weight coefficient and weighted averaging, the number of points in the high-density area can be reduced and the number of points in the low-density area can be increased while maintaining the overall shape and features of the point cloud. In this way, the resampled point cloud is more uniform in density, which is beneficial for subsequent surface reconstruction and geometric feature extraction.

[0093] S23, maintain geometric features by Laplacian operator: , where is the Laplacian coordinate of the point , is the neighborhood point set of , and is the number of neighborhood points.

[0094] Maintaining geometric features by Laplacian operator is the last step of point cloud density equalization. Laplacian operator is a differential operator used to describe the curvature or rate of change of a function at a certain point. In point cloud processing, the Laplacian operator can be used to calculate the Laplacian coordinate of each point, reflecting the geometric relationship between the point and the neighborhood points.

[0095] In implementation, the neighborhood point set of each point is determined. According to the definition of the Laplacian operator, the Laplacian coordinate of each point is calculated. The calculation of the Laplacian coordinate involves the difference between the weighted average position of the neighborhood points and the original position of the point, and the weight of the weighted average can be determined according to the distance or similarity between the neighborhood points and the point.

[0096] The action of Laplace operator is to maintain the geometric characteristics of point cloud. In the process of resampling, although the density distribution of point cloud is adjusted, some geometric distortion or smoothing may be introduced. By calculating and adjusting the Laplace coordinates, the geometric characteristics of point cloud such as edges, corners, etc. can be restored or enhanced. In this way, the resampled point cloud can better retain the geometric information of the original scanning data while maintaining the uniformity of density.

[0097] In addition, the Laplace operator also has isometry invariance, that is, when the surface is isometrically transformed, the value of the Laplace operator remains unchanged.

[0098] Step S3, performing geometric feature enhancement on the scanning data after the density equalization of the point cloud.

[0099] In the embodiment, step S3, performing geometric feature enhancement on the scanning data after the density equalization of the point cloud specifically includes the steps of:

[0100] S31, fusing curvature, normal and texture features: wherein is the comprehensive feature value of point , is the curvature feature, is the normal variation feature, is the texture feature, is the weight coefficient.

[0101] First, the curvature feature , the normal variation feature and the texture feature of each point in the scanning data need to be calculated. The curvature feature reflects the bending degree of the surface around the point , which can be estimated by calculating the normal vector angle of the point and its neighborhood points or by using local fitting surface. The normal variation feature is obtained by comparing the normal vector difference between the point and its neighborhood points, which reflects the change of surface normal vector. The texture feature is obtained by extracting the image texture information of the region where the point is located through gray level co-occurrence matrix, local binary pattern, etc.

[0102] After obtaining these features, the comprehensive feature value of each point is calculated using the formula . The weight coefficient , to balance the influence of different features on the comprehensive feature value. These weight coefficients can be determined through experiments or experience to ensure that the comprehensive feature value can accurately reflect the geometric and texture characteristics of the point .

[0103] The effect of fusing curvature, normal and texture features is that it can integrate information in multiple dimensions to enhance the geometric features of the scanning data. The curvature feature and the normal change feature provide information about the surface shape and the change of the normal vector, which is crucial for subsequent three-dimensional reconstruction and disease detection. The texture feature provides information about the surface material and texture, which helps to improve the realism and detail performance of three-dimensional reconstruction. By fusing these features, the scanning data can be more accurate and reliable in subsequent processing.

[0104] S32, feature enhancement using an improved U-Net network: wherein is the total loss function, is the reconstruction loss, is the edge preservation loss, is the structural similarity loss, is the loss weight.

[0105] Before the step of feature enhancement using the improved U-Net network, the scanning data can be preprocessed, including denoising, normalization and other operations, to ensure the quality and consistency of the input data. Then, the preprocessed data is input into the improved U-Net network.

[0106] The improved U-Net network is optimized and enhanced based on the original U-Net network to better adapt to the characteristics of underwater concrete scanning data. The network uses multi-scale feature extraction and attention mechanism technology to improve the ability of feature extraction and representation. In the training process, the total loss function is used to guide the optimization of the network. The reconstruction loss is used to measure the difference between the network output and the real data; the edge preservation loss is used to preserve the edge information in the scanning data; the structural similarity loss is used to ensure that the network output is consistent with the real data in structure. The loss weight is used to balance the influence of different loss terms.

[0107] The role of the improved U-Net network for feature enhancement is to further enhance the feature representation capability of the scanning data. Through multi-scale feature extraction and attention mechanism, the network can capture more detailed information and context information, thereby generating more accurate and rich feature representation. This helps the subsequent three-dimensional reconstruction and disease detection tasks to be more accurate and efficient. At the same time, under the guidance of the total loss function, the network can achieve better reconstruction effect while maintaining edge information and structural consistency.

[0108] S33, surface optimization based on feature constraints: wherein is the optimized vertex set, is the original vertex, is the feature constraint operator, is the balance parameter.

[0109] In the step of surface optimization based on feature constraints, first, an initial surface model is constructed according to the scanning data. Then, the formula is used for surface optimization. The feature constraint operator is used to describe the constraint conditions of the surface in terms of features (such as curvature, normal variation, etc.), and the balance parameter is used to balance the weight between vertex displacement and feature constraints.

[0110] In the optimization process, the vertex position is continuously adjusted to meet the feature constraint conditions, and the vertex displacement is as small as possible. This can be achieved by iterative solution methods such as gradient descent method, conjugate gradient method, etc. In each iteration, the gradient or update direction is calculated according to the current vertex position and feature constraint conditions, and the vertex position is updated until the convergence condition is met or the maximum number of iterations is reached.

[0111] The role of surface optimization based on feature constraints is to further improve the accuracy and realism of the three-dimensional reconstruction result. By introducing feature constraint conditions, the reconstructed surface can maintain geometric feature consistency while being more consistent with actual conditions. This helps to improve the visualization effect and practical application value of the three-dimensional reconstruction result. At the same time, by continuously optimizing the vertex position through iterative solution methods, the reconstruction result can be more accurate and stable.

[0112] Step S4, data completion and repair of scanning data after geometric feature enhancement.

[0113] In this embodiment, step S4, data completion and repair of scanning data after geometric feature enhancement specifically includes the following steps:

[0114] S41, predict missing data using underwater acoustic propagation model: wherein Distance The sound intensity at that location, The initial sound intensity, The absorption coefficient is... For reference distance.

[0115] Underwater acoustic propagation models are used to predict missing data due to environmental factors or equipment limitations. The model parameters, including the initial sound intensity, are determined based on the characteristics of the underwater environment. absorption coefficient and reference distance These parameters can be obtained through experimental measurement.

[0116] Once the parameters are determined, the formula can be used. Calculate the sound intensity at different distances By comparing the actual scan data with the predicted sound intensity, regions with missing data can be identified.

[0117] For regions with missing data, interpolation methods such as linear interpolation and spline interpolation can be used to estimate the value of the missing data points based on the values ​​of adjacent known data points.

[0118] Predicting missing data using underwater acoustic propagation models can significantly improve data completeness and accuracy. Through prediction and interpolation, gaps in scan data caused by equipment limitations or environmental factors can be filled, providing a reliable foundation for subsequent data processing and reconstruction.

[0119] S42 employs non-rigid registration based on a bio-inspired algorithm: ,in For transformation function, For regularization terms, For smoothing parameters.

[0120] Non-rigid registration is used to align images from different viewpoints or at different time points. In underwater concrete scanning and reconstruction, non-rigid registration can help solve image misalignment problems caused by changes in the underwater environment and movement of the scanning equipment. Non-rigid registration methods based on bio-inspired algorithms, such as particle swarm optimization and ant colony optimization, have strong global search capabilities and fast convergence speeds.

[0121] First, an initial transformation function needs to be estimated. This function is used to initially align the image to be registered with the reference image. The initial transformation estimate can be obtained through rigid registration or a simple affine transformation.

[0122] Next, construct the objective function. , which is used to measure the difference between the transformed image and the reference image. The objective function includes both data terms and regularization terms. Data terms are used to measure the difference between images, while regularization terms are used to control the smoothness of the transformation.

[0123] The objective function is optimized using a bio-inspired algorithm to find the transformation function that minimizes the objective function value During optimization, the bio-inspired algorithm adjusts the parameters of the transformation function until the convergence condition is met.

[0124] Non-rigid registration based on bio-inspired algorithms can significantly improve the alignment accuracy of underwater concrete scanning data. Through optimization, a smooth and accurate transformation function can be obtained, which can accurately align scanning data from different angles or different time points, providing strong support for subsequent three-dimensional reconstruction and analysis work.

[0125] S43, hole filling algorithm combined with structural features: , where is the hole region, is the normal vector, is the geometric distance.

[0126] Hole filling is used to fill in the blank or missing areas in the image. In underwater concrete scanning reconstruction, due to the limitations of scanning equipment or interference from the underwater environment, scanning data often has holes. The hole filling algorithm combined with structural features can estimate and fill in the data of the hole region according to the known structural features.

[0127] By comparing the geometric features (such as normal vectors, curvatures, etc.) of adjacent data points, the hole region in the scanning data can be identified. When the geometric features of adjacent data points differ by more than a certain threshold, it can be considered that there is a hole in that region.

[0128] Extract the feature information of the known structural region, including normal vectors, geometric distances, etc. These information is used to guide the filling process of the hole region.

[0129] According to the extracted structural feature information, a filling algorithm is used to estimate and fill in the data of the hole region. The filling algorithm can be implemented based on surface reconstruction, voxel filling or interpolation, etc. In the filling process, the consistency of the filled region with the known structural region needs to be maintained.

[0130] The hole filling algorithm combined with structural features can significantly improve the integrity and quality of underwater concrete scanning data. By identifying and filling the hole region, the blank and missing parts in the scanning data can be eliminated, providing more accurate and reliable data support for subsequent three-dimensional reconstruction and analysis work. At the same time, the algorithm can also retain the structural feature information in the original scanning data, making the reconstruction result more consistent with the actual situation.

[0131] Step S5, multi-view registration is performed on the scanned data after completion and repair.

[0132] In this embodiment, step S5, multi-view registration is performed on the scanned data after completion and repair specifically includes steps of:

[0133] S51, initial registration is performed using FPFH features: Wherein is the FPFH feature of point , is the angle feature, is the distance feature, is the scale parameter.

[0134] FPFH (Fast Point Feature Histograms) feature is a high-efficiency and robust point cloud feature description method, which encodes the local geometric relationship of each point in the point cloud, thereby realizing the initial registration of point clouds of different views.

[0135] First, FPFH feature extraction is performed on the source point cloud and the target point cloud respectively. The FPFH feature vector of each point describes the local geometric structure of the point , including the relative angle and distance information with the points in the neighborhood.

[0136] After extracting the FPFH feature, a feature matching algorithm such as matching based on feature vector distance is used to find similar feature vectors in the source point cloud and the target point cloud. By comparing the FPFH feature vector of each point, the corresponding point pair can be determined.

[0137] After finding an initial set of feature matching pairs, the RANSAC (Random Sample Consensus) algorithm is used to estimate the initial transformation matrix. The RANSAC algorithm finds the optimal transformation matrix through random sampling and iterative optimization, so that more point pairs become inliers (i.e. point pairs that meet the registration conditions).

[0138] The role of initial registration using FPFH features mainly reflects in the following aspects:

[0139] Improve registration accuracy. FPFH features can comprehensively describe the local geometric structure of points, thereby improving the accuracy of feature matching and laying a foundation for subsequent accurate registration.

[0140] Enhance robustness. FPFH features consider the information of points in the neighborhood during calculation, enhancing the robustness of the features, making them perform well under noise and incomplete data.

[0141] ​The FPFH feature vector is less computationally intensive than other higher-dimensional feature vectors such as the PFH feature vector, while still providing an adequate description of the environment. This makes it easier to implement initial registration quickly.

[0142] S52, an underwater adaptive ICP algorithm is used: wherein is a rotation matrix, is a translation vector, is a confidence weight.

[0143] The ICP (Iterative Closest Point) algorithm finds the best point cloud registration result by iteratively optimizing the least squares error between two point clouds. However, in underwater environments, traditional ICP algorithms often struggle to achieve ideal registration results due to factors such as light refraction and noise interference. Therefore, using an underwater adaptive ICP algorithm is an effective solution.

[0144] Before performing the ICP algorithm, the input point cloud data is preprocessed, including removing outliers and downsampling, to improve the stability and efficiency of the algorithm.

[0145] Nearest neighbor search: For each point in the reference point cloud, find the closest point in the target point cloud to establish a correspondence. This step is one of the core steps of the ICP algorithm, as it determines the accuracy of the subsequent transformation matrix calculation.

[0146] Confidence weight assignment: In underwater environments, due to noise and interference, the registration reliability of different point pairs varies. Therefore, a confidence weight is introduced in the ICP algorithm to reflect the registration reliability of each point pair. The confidence weight can be calculated based on the distance and angle between the points.

[0147] Transformation matrix calculation: Based on the correspondence between the points, the rotation matrix and translation vector are calculated using the least squares method to align the target point cloud to the reference point cloud coordinate system. During the calculation process, the influence of the confidence weight is considered to improve the accuracy of the transformation matrix.

[0148] Iterative optimization: Apply the transformation matrix to the target point cloud to obtain an updated target point cloud. Then calculate the mean square error (MSE) or other similarity measurement indicators to determine whether the preset convergence condition is met. If not, return to the nearest neighbor search step and continue iterative optimization.

[0149] To implement the underwater adaptive ICP algorithm, the following improvements can be made:

[0150] ​Considering the characteristics of underwater environment, the ICP algorithm is adapted to the characteristics of underwater light refraction and noise interference. For example, a more robust search method can be used in the nearest neighbor search process; in the transformation matrix calculation process, the statistical characteristics of underwater noise can be considered for weighted processing.

[0151] An adaptive mechanism is introduced to dynamically adjust the parameters and strategies of the ICP algorithm according to the error change, convergence speed and other indicators in the iteration process. For example, a larger step size can be used for fast convergence in the early stage of iteration; a smaller step size can be used for fine adjustment in the later stage of iteration.

[0152] The role of the underwater adaptive ICP algorithm mainly reflects in the following aspects:

[0153] Improve registration accuracy: by introducing confidence weight and adaptive mechanism, the underwater adaptive ICP algorithm can more accurately estimate the transformation matrix, thereby improving the registration accuracy;

[0154] Enhance robustness: adaptive adjustment is made according to the characteristics of underwater environment, so that the algorithm can still work stably in the presence of noise and interference;

[0155] Speed up the convergence: by dynamically adjusting the algorithm parameters and strategies, the underwater adaptive ICP algorithm can speed up the convergence while ensuring the accuracy, improving the calculation efficiency.

[0156] S53, global optimization based on pose graph: , where is the node pose, is the relative pose measurement, is the covariance matrix.

[0157] Global optimization based on pose graph solves the point cloud registration problem by transforming it into a graph optimization problem.

[0158] According to the results of initial registration and accurate registration, a pose graph is constructed. The pose graph is composed of nodes and edges, where the nodes represent the pose estimation of point clouds from different perspectives, and the edges represent the constraint relationship between nodes (such as relative pose measurement).

[0159] Define the optimization objective: the goal of global optimization based on pose graph is to minimize the error function , which represents the weighted sum of squares of all edge constraints. The node pose , the relative pose measurement and the covariance matrix are used to reflect the statistical characteristics of measurement error.

[0160] The error function is solved by using a nonlinear least squares method (such as the Gauss-Newton method, the Levenberg-Marquardt algorithm, etc.). In the optimization process, the node pose is continuously adjusted to meet the edge constraint condition and minimize the error function.

[0161] According to the error change, the number of iterations, and other indicators in the optimization process, it is determined whether the convergence condition is reached. If the convergence condition is reached, the optimized node pose is output as the final registration result; otherwise, the iteration optimization is continued.

[0162] In some optional implementations of the embodiment, in order to realize global optimization based on the pose graph, the following improvements can also be made:

[0163] Loop closure detection is introduced. By detecting the loop closure in the pose graph and correcting the error, the consistency of the pose estimation can be improved. Loop closure detection can be realized by using feature matching, loop detection algorithm, etc.

[0164] Sensor error is considered. In actual application, due to the existence of sensor error, there is a certain uncertainty in relative pose measurement. Therefore, when constructing the pose graph and defining the optimization target, the statistical characteristics of the sensor error need to be considered for weighted processing.

[0165] Parallel computing is adopted. For large-scale point cloud data, the global optimization based on the pose graph has a large amount of calculation. Therefore, parallel computing technology can be used to speed up the optimization process and improve the calculation efficiency.

[0166] The role of global optimization based on the pose graph mainly reflects in the following aspects:

[0167] Consistency is improved. Through global optimization, the consistency of the entire point cloud model between different perspectives can be ensured, and problems such as misalignment or overlap can be avoided.

[0168] Accuracy is enhanced. Considering factors such as sensor error and loop closure detection for global optimization can further improve the accuracy of registration.

[0169] Optimization calculation efficiency is improved. The use of parallel computing and other technologies can speed up the optimization process and improve the calculation efficiency, making the global optimization method based on the pose graph more feasible and efficient in actual application.

[0170] Step S6, surface reconstruction is performed on the scanned data after multi-view registration.

[0171] In the embodiment, step S6, surface reconstruction is performed on the scanned data after multi-view registration, specifically comprising the steps of:

[0172] S61, spatial division based on point cloud density: , wherein For node decision-making, The number of points within a node. The threshold value is used.

[0173] Spatial partitioning based on point cloud density is a crucial step in the preprocessing of scanned data. It partitions the space according to the density of each node in the point cloud data, thereby optimizing the efficiency and accuracy of subsequent surface reconstruction.

[0174] The scanned data after multi-view registration is preprocessed, including noise removal and missing data filling, to ensure the integrity and accuracy of the point cloud data.

[0175] The point cloud density of each node is calculated by dividing the three-dimensional point cloud space into fixed-size blocks (such as cubes or spheres) and counting the number of points in each block. This is the preset density threshold.

[0176] Spatial partitioning decisions are made based on node density. If If so, the nodes are split to reduce density; if If so, adjacent low-density nodes will be merged to increase density; if ≤ ≤ If so, then the node remains unchanged.

[0177] Spatial partitioning based on point cloud density plays the following roles in surface reconstruction:

[0178] Optimizing data distribution by segmenting and merging nodes makes point cloud data more evenly distributed in space, which is beneficial for subsequent surface reconstruction algorithms to better capture the geometric features of the data.

[0179] To improve reconstruction efficiency, the computational complexity of the surface reconstruction algorithm is reduced by decreasing the number of points in high-density regions and increasing the number of points in low-density regions, thus improving reconstruction efficiency.

[0180] The optimized point cloud data more accurately reflects the geometry of underwater concrete, thereby improving the accuracy of surface reconstruction.

[0181] S62 employs an improved Poisson reconstruction algorithm: ,in For indicator functions, It is a vector field.

[0182] It is an operator; it is not a value in itself, but an operation command.

[0183] It is a divergence operation, which, for a vector field, represents the intensity of the divergence.

[0184] is the Laplace operator, which measures the degree of curvature for a scalar field.

[0185] The Poisson reconstruction algorithm is a function-based triangular mesh reconstruction algorithm that is suitable for reconstructing continuous surfaces from discrete point cloud data. In the improved Poisson reconstruction algorithm, the concept of vector field is introduced to better capture the geometric features of the point cloud data.

[0186] Calculate the normal vector of each point in the point cloud. These normal vectors help determine the direction of each point on the surface and are the basis of the Poisson reconstruction algorithm.

[0187] Construct a vector field based on the point cloud data and normal vectors Each vector in the vector field points in a certain direction on the surface of the point cloud, reflecting the geometric features of the surface.

[0188] Estimate the indicator function by solving the Poisson equation . The indicator function is a scalar field that is positive near the surface of the point cloud and negative or zero in areas far from the surface. By solving the Poisson equation, an approximate value of the indicator function can be obtained.

[0189] Extract isosurfaces from the indicator function using the MC (Marching Cubes) algorithm. These isosurfaces are the reconstructed surface model.

[0190] The improved Poisson reconstruction algorithm plays the following roles in surface reconstruction:

[0191] Capture geometric features: by introducing the concept of vector field, the algorithm can more accurately capture the geometric features of the point cloud data, such as curvature, edges, etc.

[0192] Improve reconstruction quality: the Poisson reconstruction algorithm is based on implicit function for surface reconstruction, which can generate smooth and continuous surface model, avoiding the grid distortion or holes that may occur in traditional reconstruction algorithm.

[0193] Enhance robustness: the improved Poisson reconstruction algorithm has strong robustness to noise and missing data, and can tolerate the incompleteness and errors of data to a certain extent.

[0194] S63, perform hierarchical mesh optimization: where is the layer mesh, is the point cloud data, is the Laplace operator.

[0195] The hierarchical mesh optimization is a step of further optimizing the reconstructed surface model. It mainly improves the quality of the mesh by adjusting the positions of the mesh vertices, splitting or merging the mesh elements, etc.

[0196] An initial mesh is generated based on the reconstructed surface model. The initial mesh may contain some irregular or poor-quality elements.

[0197] The initial mesh is hierarchically divided by dividing it into multiple levels (or levels), each level containing different numbers of mesh elements. The mesh elements of higher levels are larger, and the mesh elements of lower levels are smaller.

[0198] Mesh optimization, at each level, adjusts the positions of the mesh vertices using optimization algorithms. The optimization goal is to minimize the error between the mesh and the point cloud data while maintaining the smoothness and continuity of the mesh. This can be achieved by solving an optimization problem that includes a data fidelity term and a smoothing term, such as .

[0199] According to the optimization results, the mesh is refined or merged. The refinement operation can increase the resolution of the mesh and improve the detail performance of the model; the merging operation can reduce the complexity of the mesh and reduce the computational overhead.

[0200] The hierarchical mesh optimization plays the following roles in surface reconstruction:

[0201] Improving mesh quality, hierarchical mesh optimization can significantly improve the quality of the mesh by adjusting the positions of the mesh vertices and splitting or merging the mesh elements, making it smoother, more continuous, and more consistent with the actual geometric shape;

[0202] Improving model accuracy, the optimized mesh model can more accurately reflect the geometric shape and detail features of the underwater concrete, thereby improving the accuracy and reliability of the reconstructed model;

[0203] Reducing computational overhead, through reasonable mesh refinement and merging operations, the computational overhead can be reduced while maintaining the model accuracy, improving the reconstruction efficiency.

[0204] The implementation of this embodiment has the following benefits:

[0205] By using high-precision equipment to obtain scanning data of underwater concrete, and then using adaptive noise filtering technology, random noise in the data is effectively removed, and data quality is improved;

[0206] Implementing point cloud density equalization processing ensures consistent density distribution of scanning data in different areas, laying a solid foundation for subsequent processing;

[0207] Through geometric feature enhancement, the edges and key shapes of the concrete structure can be highlighted, providing more accurate references for data completion and repair. Data completion and repair techniques precisely restore missing or damaged parts to ensure the integrity and accuracy of the scanned data.

[0208] Through multi-view registration, the scanned data from different angles is accurately aligned to form a globally consistent three-dimensional model, which is crucial for constructing a three-dimensional model of complex structures.

[0209] Through surface reconstruction technology, the registered scanned data is converted into continuous and smooth three-dimensional surfaces, realizing the accurate digital reconstruction of underwater concrete structures.

[0210] Not only does it improve data processing efficiency and accuracy, but it also significantly enhances the realism and practicality of the reconstructed model, providing strong technical support for underwater engineering detection, maintenance and design.

[0211] The present application can be used in a variety of general-purpose or special-purpose computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in a distributed computing environment, in which tasks are performed by remote processing devices connected by a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0212] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by computer-readable instructions instructing relevant hardware, which can be stored in a computer-readable storage medium. The program, when executed, can include the processes of the above-mentioned embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0213] It should be understood that although each step in the flowchart of the accompanying drawings is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless explicitly stated herein, the execution of these steps is not strictly limited in sequence, and they can be executed in other sequences. Moreover, at least part of the steps in the flowchart of the accompanying drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be alternately executed with at least part of other steps or sub-steps or stages of other steps.

[0214] Embodiment Two:

[0215] Further referring to Figure 2 , as an implementation of the method shown in the above Figure 1 , the present application provides an embodiment of a data processing device based on underwater concrete scanning reconstruction. The device embodiment corresponds to the method embodiment shown in Figure 1 , and the device can be applied to various electronic devices.

[0216] As shown in Figure 2 , the data processing device 70 based on underwater concrete scanning reconstruction in this embodiment includes a filtering module 71, an equalization module 72, an enhancement module 73, a repair module 74, a registration module 75, and a reconstruction module 76. Among them:

[0217] The filtering module 71 is configured to obtain scanning data based on underwater concrete, and perform adaptive noise filtering on the scanning data.

[0218] The equalization module 72 is configured to perform point cloud density equalization on the scanning data after adaptive noise filtering.

[0219] The enhancement module 73 is configured to perform geometric feature enhancement on the scanning data after point cloud density equalization.

[0220] The repair module 74 is configured to perform data completion and repair on the scanning data after geometric feature enhancement.

[0221] The registration module 75 is configured to perform multi-view registration on the scanning data after completion and repair.

[0222] The reconstruction module 76 is configured to perform surface reconstruction on the scanning data after multi-view registration.

[0223] By implementing this embodiment, the beneficial effects are:

[0224] The scanning data of underwater concrete is obtained by high-precision equipment, and then adaptive noise filtering technology is used to effectively remove random noise in the data and improve data quality.

[0225] The point cloud density equalization processing is implemented to ensure consistent density distribution of the scanning data in different areas, laying a solid foundation for subsequent processing.

[0226] Through geometric feature enhancement, the edges and key shapes of the concrete structure can be highlighted, providing more accurate reference for data completion and repair. Data completion and repair technology accurately restores the missing or damaged parts to ensure the integrity and accuracy of the scanning data.

[0227] Through multi-view registration, the scanning data from different angles is accurately aligned to form a globally consistent three-dimensional model, which is crucial for constructing a three-dimensional model of complex structures.

[0228] Through surface reconstruction technology, the registered scanning data is converted into continuous and smooth three-dimensional surfaces, realizing accurate digital reconstruction of underwater concrete structures.

[0229] Not only does it improve data processing efficiency and accuracy, but it also significantly enhances the realism and practicality of the reconstructed model, providing strong technical support for underwater engineering detection, maintenance and design.

[0230] Embodiment Three

[0231] To solve the above technical problems, the embodiments of the present application also provide a device. For details, please refer to Figure 3 , Figure 3 The basic structure diagram of the device of the present embodiment is shown in the figure.

[0232] The device 8 includes a memory 81, a processor 82, and a network interface 83 connected to each other through a system bus. It should be noted that only the device 8 with components memory 81, processor 82 and network interface 83 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the device here is a device that can automatically perform numerical calculation and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), field programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0233] The device can be a desktop computer, a notebook computer, a palm computer, a cloud server, or the like. The device can interact with a user through a keyboard, a mouse, a remote controller, a touchpad, a voice control device, or the like.

[0234] The memory 81 can include at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, or the like), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, or the like. In some embodiments, the memory 81 can be an internal storage unit of the device 8, such as a hard disk or a memory of the device 8. In other embodiments, the memory 81 can also be an external storage device of the device 8, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, or the like. Of course, the memory 81 can include both an internal storage unit and an external storage device of the device 8. In this embodiment, the memory 81 is generally used to store an operating system and various application software installed in the device 8, such as computer readable instructions of the data processing method based on underwater concrete scanning and reconstruction, or the like. In addition, the memory 81 can also be used to temporarily store various data that has been output or will be output.

[0235] The processor 82 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 82 is generally used to control the overall operation of the device 8. In this embodiment, the processor 82 is used to run computer readable instructions or process data stored in the memory 81, such as computer readable instructions of the data processing method based on underwater concrete scanning and reconstruction.

[0236] The network interface 83 can include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the device 8 and other electronic devices.

[0237] By implementing this embodiment, the following beneficial effects can be achieved:

[0238] The scanning data of underwater concrete is obtained by a high-precision device, and then the adaptive noise filtering technology is used to effectively remove random noise in the data and improve the data quality.

[0239] The point cloud density equalization processing is implemented, so that the scanning data has consistent density distribution in different regions, and a solid foundation is laid for subsequent processing;

[0240] Through geometric feature enhancement, the edges and key shapes of the concrete structure can be highlighted, and more accurate references are provided for data completion and repair, and the data completion and repair technology accurately restores the missing or damaged parts, so that the integrity and accuracy of the scanning data are ensured.

[0241] Through multi-angle registration, the scanning data from different angles is accurately aligned to form a globally consistent three-dimensional model, which is crucial for constructing a three-dimensional model of a complex structure.

[0242] Through surface reconstruction technology, the registered scanning data is converted into continuous and smooth three-dimensional surfaces, so that the accurate digital reconstruction of the underwater concrete structure is realized.

[0243] Not only the data processing efficiency and accuracy are improved, but also the realism and practicality of the reconstructed model are significantly enhanced, which provides strong technical support for underwater engineering detection, maintenance and design.

[0244] Embodiment four:

[0245] The application also provides another embodiment, that is, a computer readable storage medium storing computer readable instructions, which can be executed by at least one processor to make the at least one processor execute the steps of the data processing method based on underwater concrete scanning reconstruction as described above.

[0246] The implementation of the embodiment has the following beneficial effects:

[0247] The scanning data of the underwater concrete is obtained by high-precision equipment, and then the adaptive noise filtering technology is used to effectively remove random noise in the data and improve the data quality.

[0248] The point cloud density equalization processing is implemented, so that the scanning data has consistent density distribution in different regions, and a solid foundation is laid for subsequent processing;

[0249] Through geometric feature enhancement, the edges and key shapes of the concrete structure can be highlighted, and more accurate references are provided for data completion and repair, and the data completion and repair technology accurately restores the missing or damaged parts, so that the integrity and accuracy of the scanning data are ensured.

[0250] Through multi-angle registration, the scanning data from different angles is accurately aligned to form a globally consistent three-dimensional model, which is crucial for constructing a three-dimensional model of a complex structure.

[0251] The registered scanning data is converted into continuous and smooth three-dimensional surface by surface reconstruction technology, so as to realize accurate digital reconstruction of the underwater concrete structure.

[0252] The data processing efficiency and precision are improved, the reality and practicability of the reconstructed model are significantly enhanced, and powerful technical support is provided for underwater engineering detection, maintenance and design.

[0253] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and the necessary general hardware platform, of course, they can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) execute the various embodiment methods of the present application.

[0254] Obviously, the above-described embodiments are only some of the embodiments of the present application, not all the embodiments, and the preferred embodiments of the present application are given in the drawings, but do not limit the patent scope of the present application. The present application can be realized in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing specific embodiments, or equivalently replace some technical features. Any equivalent structure made by using the content of the present application specification and drawings, directly or indirectly applied to other related technical fields, is also within the patent protection scope of the present application.

Claims

1. A data processing method based on underwater concrete scanning and reconstruction, characterized in that, Includes the following steps: Acquire scanning data based on underwater concrete, and perform adaptive noise filtering on the scanning data; Point cloud density equalization is performed on the scan data after adaptive noise filtering; Geometric feature enhancement is performed on the scanned data after point cloud density equalization; Data completion and repair are performed on the scanned data after geometric feature enhancement. Multi-view registration is performed on the completed and repaired scan data; Surface reconstruction is performed on the scan data after multi-view registration; The step of acquiring underwater concrete-based scanning data and performing adaptive noise filtering on the scanning data specifically includes: Acquire scanning data based on underwater concrete; The DBSCAN algorithm is used to classify noise based on the characteristics of the water medium. ,in For point Energy value, For point and Euclidean distance, The attenuation coefficient of the water medium. Distance weighting factor; By combining wavelet transform and statistical filtering, noise is eliminated at different scales. ,in These are wavelet coefficients. The original signal, For wavelet basis functions, For scale parameters, These are displacement parameters; Bilateral filtering is used to protect geometric features. ,in The intensity of the filtered point. As the normalization factor, For the spatial domain Gaussian kernel, For the range Gaussian kernel, For the neighborhood point set; The step of performing geometric feature enhancement on the scanned data after point cloud density equalization specifically includes: Integrating curvature, normal, and texture features: ,in For comprehensive eigenvalues, Curvature characteristics Characteristics of normal changes For texture features, These are the weighting coefficients; Feature enhancement is performed using an improved U-Net network: ,in For the total loss function, To rebuild the losses, To preserve loss at the edge, For structural similarity loss, For loss weights; Feature-constrained surface optimization: ,in For the optimized vertex set, For the original vertex, For feature constraint operators, For balance parameters; The steps for data completion and repair of the scanned data after geometric feature enhancement specifically include: Predicting missing data using underwater acoustic propagation models: ,in Distance The sound intensity at that location, The initial sound intensity, The absorption coefficient is... For reference distance; Non-rigid registration based on bio-inspired algorithms is employed: ,in For transformation function, For regularization terms, For smoothing parameters; Hole-filling algorithm combining structural features: ,in For the area with holes, It is the normal vector. For geometric distance.

2. The data processing method based on underwater concrete scanning and reconstruction according to claim 1, characterized in that, The step of performing point cloud density equalization on the scanned data after adaptive noise filtering specifically includes: Establish a density field model based on kernel density estimation: ,in For position Density estimation at [location] The number of sample points. For bandwidth parameters, For data dimensions, For kernel functions; Non-uniform resampling based on density field: ,in For the newly generated points, These are the weighting coefficients. The number of neighboring points; Preserving geometric characteristics using the Laplacian operator: ,in For point Laplace coordinates, for The neighborhood point set.

3. The data processing method based on underwater concrete scanning and reconstruction according to claim 1, characterized in that, The step of performing multi-view registration on the completed and repaired scan data specifically includes: Initial registration using FPFH features: ,in For point FPFH characteristics, As an angular feature, For distance features, For scale parameters; Underwater adaptive ICP algorithm is used: ,in For rotation matrix, It is a translation vector. Confidence weights; Global optimization based on pose graph: ,in For the node pose, For relative pose measurement, Let be the covariance matrix.

4. The data processing method based on underwater concrete scanning and reconstruction according to any one of claims 1 to 3, characterized in that, The step of performing surface reconstruction on the scan data after multi-view registration specifically includes: Spatial partitioning based on point cloud density: ,in For node decision-making, The number of points within a node. For the threshold; An improved Poisson reconstruction algorithm is used: ,in For indicator functions, It is a vector field; Perform hierarchical mesh optimization: ,in For the first Layered mesh, For point cloud data, For the Laplace operator.

5. A data processing device based on underwater concrete scanning and reconstruction, characterized in that, include: A filtering module is used to acquire scanning data based on underwater concrete and to perform adaptive noise filtering on the scanning data; An equalization module is used to equalize the point cloud density of the scanned data after adaptive noise filtering. The enhancement module is used to perform geometric feature enhancement on the scanned data after point cloud density equalization; The repair module is used to complete and repair the scanned data after geometric feature enhancement; The registration module is used to perform multi-view registration on the completed and repaired scan data; The reconstruction module is used to perform surface reconstruction on the scan data after multi-view registration; The filtering module is further used for: Acquire scanning data based on underwater concrete; The DBSCAN algorithm is used to classify noise based on the characteristics of the water medium. ,in For point Energy value, For point and Euclidean distance, The attenuation coefficient of the water medium. Distance weighting factor; By combining wavelet transform and statistical filtering, noise is eliminated at different scales. ,in These are wavelet coefficients. The original signal, For wavelet basis functions, For scale parameters, These are displacement parameters; Bilateral filtering is used to protect geometric features. ,in The intensity of the filtered point. As the normalization factor, For the spatial domain Gaussian kernel, For the range Gaussian kernel, For the neighborhood point set; The enhancement module is further used for: Integrating curvature, normal, and texture features: ,in For comprehensive eigenvalues, Curvature characteristics Characteristics of normal changes For texture features, These are the weighting coefficients; Feature enhancement is performed using an improved U-Net network: ,in For the total loss function, To rebuild the losses, To preserve loss at the edge, For structural similarity loss, For loss weights; Feature-constrained surface optimization: ,in For the optimized vertex set, For the original vertex, For feature constraint operators, For balance parameters; The repair module is further used for: Predicting missing data using underwater acoustic propagation models: ,in Distance The sound intensity at that location, The initial sound intensity, The absorption coefficient is... For reference distance; Non-rigid registration based on bio-inspired algorithms is employed: ,in For transformation function, For regularization terms, For smoothing parameters; Hole-filling algorithm combining structural features: ,in For the area with holes, It is the normal vector. For geometric distance.

6. An electronic device comprising a memory and a processor, characterized in that, The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, it implements the steps of the data processing method based on underwater concrete scanning and reconstruction as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the data processing method based on underwater concrete scanning and reconstruction as described in any one of claims 1 to 4.

Images