Rock mass structural plane recognition method, device and equipment based on point cloud and storage medium

Abstract

The invention provides a rock mass structural surface identification method, device and equipment based on point cloud and a storage medium, and relates to the technical field of data processing, and the method comprises the steps: processing rock mass dense three-dimensional point cloud data to obtain processed three-dimensional point cloud data; performing uniform down-sampling on the processed three-dimensional point cloud data to obtain sampled point cloud data and unsampled point cloud data; training a neural network model according to the sampled point cloud data to obtain a rock mass structural surface recognition model; inputting the unsampled and processed point cloud data and the unsampled and processed point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled and processed point cloud label set; and obtaining the rock mass structural plane occurrence according to the unsampled point cloud label set. According to the invention, the recognition rate is improved while the recognition precision of the rock mass structural surface is ensured.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, equipment and storage medium for identifying rock mass structural surfaces based on point clouds. Background Art

[0002] Identifying rock surface structures is extremely important in fields such as geological engineering. These structures refer to discontinuities in rock, such as joints, fissures, and faults, which have a decisive influence on the physical and mechanical properties of the rock mass. Current methods for automatically identifying rock surface structures from point cloud data are overly complex, often requiring long runtimes to process large amounts of data, and are prone to errors. Summary of the Invention

[0003] The present invention solves the problem of how to improve the recognition rate while ensuring the recognition accuracy of rock mass structural surfaces.

[0004] To solve the above problems, the present invention provides a method, device, equipment and storage medium for identifying rock structure surfaces based on point cloud.

[0005] In a first aspect, the present invention provides a method for identifying rock mass structural surfaces based on point clouds, comprising: Processing the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; uniformly downsampling the processed three-dimensional point cloud data to obtain sampled processed point cloud data and unsampled processed point cloud data; Training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; Inputting the unsampled processed point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled processed point cloud label set; The occurrence of the rock mass structural surface is obtained according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0006] Optionally, the step of training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model includes: Based on the agglomerative hierarchical clustering algorithm, the sampled point cloud data is classified to obtain a corresponding sampled point cloud label set; Obtaining a point cloud data training set and a point cloud data test set according to the sampled processed point cloud data and the corresponding sampled processed point cloud label set; Training the neural network model using the point cloud data training set until the neural network converges to obtain an initial neural network model; The initial neural network model is tested using the point cloud data test set to test the test set accuracy of the initial neural network model. When the test set accuracy does not meet the accuracy requirement, more data is acquired for training until the accuracy requirement is met, thereby obtaining the rock mass structural surface recognition model. The test set accuracy includes: , Among them, AC is the accuracy of the test set, is the number of accurately classified point clouds, is the number of point clouds in the point cloud data test set.

[0007] Optionally, the classifying the sampled processed point cloud data to obtain a corresponding sampled processed point cloud label set includes: Processing the point cloud data by the sampling to obtain a similarity metric; The similarity measurement includes: , in, is the similarity metric, is the normal vector of the i-th point cloud, is the transposed matrix of the normal vector of the j-th point cloud; The sampled and processed point cloud data are grouped into structural surface categories according to the similarity metric to obtain a corresponding sampled and processed point cloud label set.

[0008] Optionally, obtaining the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data includes: A density-based noisy spatial clustering algorithm is used to separate the unsampled point cloud data according to the unsampled point cloud label set to obtain multiple rock mass structural surfaces; Based on the least square method, data fitting is performed according to the rock mass structural surface to obtain the corresponding plane normal vector; Obtaining the inclination angle and the dip of the rock mass structural surface according to the plane normal vector; The rock mass structural surface occurrence includes: , , in, is the inclination angle of the rock mass structural surface, is the inclination of the rock mass structural surface, (A, B, C) is the plane normal vector; The occurrence of the rock mass structural surface is obtained according to the inclination angle of the rock mass structural surface and the inclination of the rock mass structural surface.

[0009] Optionally, obtaining the sampled point cloud data and the unsampled point cloud data by uniformly downsampling the processed three-dimensional point cloud data includes: Dividing the processed three-dimensional point cloud data into a plurality of three-dimensional grids at a fixed size; Obtaining a grid center point according to each of the three-dimensional grids; Using all the grid center points as the sampling point cloud data; The unsampled processed point cloud data is obtained according to data in the processed three-dimensional point cloud data except the sampled processed point cloud data.

[0010] Optionally, the dense three-dimensional point cloud data of the rock mass includes a plurality of point clouds, and the dense three-dimensional point cloud data of the rock mass is processed to obtain processed three-dimensional point cloud data, including: Based on the KD tree algorithm, corresponding neighborhood point sets are obtained according to the point clouds; The processed three-dimensional point cloud data is obtained by performing principal component analysis on the neighborhood point set.

[0011] Optionally, it also includes: The structural surface image is obtained by collecting rock structural surface data; Based on the SfM-MVS algorithm, the structural surface image is converted into point cloud data to obtain dense three-dimensional point cloud data of the rock mass.

[0012] In a second aspect, the present invention provides a rock mass structural surface identification device based on point cloud, comprising: a data processing module for processing dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; a downsampling module, configured to uniformly downsample the processed three-dimensional point cloud data to obtain sampled point cloud data and unsampled point cloud data; A model training module is used to train a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; An unsampled point cloud label set acquisition module is used to input the unsampled point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled point cloud label set; The rock mass structural surface occurrence acquisition module is used to obtain the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0013] In a third aspect, the present invention provides an electronic device comprising a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the point cloud-based rock mass structural surface identification method as described in the first aspect when executing the computer program.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the point cloud-based rock structure surface identification method as described in the first aspect is implemented.

[0015] The beneficial effects of the point cloud-based rock structure surface identification method, device, equipment and storage medium of the present invention are: by processing the dense three-dimensional point cloud data of the rock mass, by processing the point cloud data, the accuracy of the data is improved, and the foundation is laid for subsequent data processing. By uniformly downsampling the processed three-dimensional point cloud data, the impact of redundant data on the training speed is reduced, the data processing time is shortened, and the key geometric features of the rock surface are retained, thereby improving the accuracy of rock structure surface identification. The rock structure surface identification model is obtained by training a neural network model using the sampled processed point cloud data, and the structural understanding ability is improved by the neural network model, so that the model can converge quickly using simplified data. The unsampled processed point cloud data is input into the rock structure surface identification model to obtain the corresponding unsampled processed point cloud label set, and the rock structure surface attitude is further obtained, which can improve the recognition rate while ensuring the accuracy of rock structure surface identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of a flow chart of a method for identifying rock mass structural surfaces based on point clouds according to an embodiment of the present invention; Figure 2 A schematic structural diagram of a neural network model according to an embodiment of the present invention; Figure 3 Schematic diagram of the structure of a rock mass structural surface identification device based on point cloud according to an embodiment of the present invention; Figure 4 The figure is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0018] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0019] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to"; the term "based on" means "based at least in part on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0020] It should be noted that the modifications of "one" and "multiple" mentioned in the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0021] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0022] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying rock mass structural surfaces based on point clouds, comprising: Step 110 : Process the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data.

[0023] Specifically, the rock structure surface data on site is collected through drones or high-definition cameras, and structural surface photos are obtained, which are converted into point cloud data to obtain dense three-dimensional point cloud data of the rock mass.

[0024] Step 120 : uniformly down-sampling the processed three-dimensional point cloud data to obtain sampled processed point cloud data and unsampled processed point cloud data.

[0025] Specifically, uniform sampling is a point cloud simplification method that selects representative points at fixed spatial intervals to remove redundant information. The data obtained from uniform sampling is used as the sampled point cloud data, and the remaining data in the processed 3D point cloud data is used as the unsampled point cloud data.

[0026] Step 130: training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model.

[0027] Specifically, the sampled point cloud data is labeled and classified using an agglomerative hierarchical clustering algorithm to achieve automatic labeling of the point cloud. A neural network model is trained on the labeled data to obtain a rock mass structural surface recognition model.

[0028] Step 140: Input the unsampled point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled point cloud label set. Specifically, the unsampled point cloud data is input into the trained rock mass structural surface recognition model for automatic classification to obtain the classification labels of all point clouds, i.e., structural surface groups, and the point clouds can be automatically assigned to different structural surface groups.

[0029] Step 150 : Obtaining the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0030] Specifically, in engineering geology and rock mechanics, the attitude of a rock mass structural surface refers to its geometric orientation and tilt in three-dimensional space. It is typically uniquely determined by its inclination angle and dip.

[0031] In this embodiment, by processing the dense three-dimensional point cloud data of the rock mass, the accuracy of the data is improved by processing the point cloud data, laying the foundation for subsequent data processing. By uniformly downsampling the processed three-dimensional point cloud data, the impact of redundant data on the training speed is reduced, the data processing time is shortened, and the key geometric features of the rock mass surface are retained, thereby improving the accuracy of rock mass structural surface recognition. The rock mass structural surface recognition model is obtained by training the neural network model using the sampled processed point cloud data, and the structural understanding ability is improved by the neural network model, so that the model can converge quickly using simplified data. The unsampled processed point cloud data is input into the rock mass structural surface recognition model to obtain the corresponding unsampled processed point cloud label set, and the rock mass structural surface attitude is further obtained, which can ensure the accuracy of rock mass structural surface recognition while improving the recognition rate.

[0032] Optionally, the step of training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model includes: Based on the agglomerative hierarchical clustering algorithm, the sampled point cloud data is classified to obtain a corresponding sampled point cloud label set; Obtaining a point cloud data training set and a point cloud data test set according to the sampled processed point cloud data and the corresponding sampled processed point cloud label set; Training the neural network model using the point cloud data training set until the neural network converges to obtain an initial neural network model; The initial neural network model is tested using the point cloud data test set to test the test set accuracy of the initial neural network model. When the test set accuracy does not meet the accuracy requirement, more data is acquired for training until the accuracy requirement is met, thereby obtaining the rock mass structural surface recognition model. The test set accuracy includes: , Among them, AC is the accuracy of the test set, is the number of accurately classified point clouds, is the number of point clouds in the point cloud data test set.

[0033] Specifically, combined Figure 2 As shown in the figure, the neural network model includes a 1×6 input layer, five convolutional layers, two 4096-bit fully connected layers, and an output layer. The 1×6 input layer, Conv1, takes in point cloud coordinates and normals. The five convolutional layers, Conv2, Conv3, Conv4, Conv5, and Conv6, have output channels of 96, 256, 384, 384, and 256, respectively, and a convolution kernel size of 1×6. There are also two 4096-bit fully connected layers and an output layer whose output value is the point cloud's label.

[0034] In some more specific embodiments, the hyperparameters of the neural network model are set, including activation function, batch size, learning rate, epoch, and optimization algorithm. The batch size represents the number of samples used in each iteration; the learning rate is the rate at which the model updates weights in each iteration; the epoch refers to the number of times the network processes the entire training dataset; and the optimization algorithm is a model used to adjust model parameters to minimize the loss function. The model is fine-tuned using a trial-and-error method to optimize the training process until the training accuracy exceeds 90%. The sampled dataset is divided into a point cloud data training set and a point cloud data test set in an 8:2 ratio. The point cloud data training set is input into the neural network model for training, and the model is tested using the point cloud data test set to verify the model's test set accuracy AC. When the test set accuracy exceeds 90%, the trained rock mass structural surface recognition model is saved.

[0035] In this optional embodiment, an agglomerative hierarchical clustering algorithm is used to classify the sampled point cloud data to obtain corresponding point cloud labels. The point cloud data with cluster labels is divided into a training set and a test set for subsequent training and verification of the neural network model.

[0036] Optionally, the classifying the sampled processed point cloud data to obtain a corresponding sampled processed point cloud label set includes: Processing the point cloud data by the sampling to obtain a similarity metric; The similarity measurement includes: , in, is the similarity metric, is the normal vector of the i-th point cloud, is the transposed matrix of the normal vector of the j-th point cloud; The sampled and processed point cloud data are grouped into structural surface categories according to the similarity metric to obtain a corresponding sampled and processed point cloud label set.

[0037] Specifically, the Agglomerative Hierarchical Clustering (AHC) algorithm is a classic unsupervised clustering method, which belongs to the bottom-up hierarchical clustering. It builds a tree structure by continuously merging sample points or clusters with high similarity, thereby revealing the inherent hierarchical structure of the data. The sampled point cloud is classified using the agglomerative hierarchical clustering algorithm to achieve automatic labeling of the point cloud, treating each object as a cluster, and then merging these clusters into larger clusters until a certain termination condition is met. Currently, the square of the cosine value between the unit normal vectors is the most effective way to cluster the rock structure surface. Determine how many categories or structural surface groups you want to divide the point cloud into, and set the number of clusters M. M is an integer representing the number of clusters to be obtained in the end. Calculate the cosine distance of all point clouds, that is, the similarity measure, where: , , in, is the similarity metric, is the normal vector of the i-th point cloud, is the transposed matrix of the normal vector of the j-th point cloud, The sharp angle between the normal vectors of two point clouds is automatically assigned to the same label L (L = 1, 2, ..., M). Points belonging to the same structural surface are initially grouped together to form candidate regions. These labeling results are manually reviewed and corrected by professional technicians. For example, for areas that are mis-clustered due to sparse point clouds or occlusion, operators can manually adjust the point cloud labels to ensure that the point set corresponding to each label accurately represents an independent rock mass structural surface.

[0038] In this optional embodiment, the point cloud is classified based on the agglomerative hierarchical clustering algorithm. The agglomerative hierarchical clustering algorithm has good topological structure preservation ability and is suitable for point cloud data classification tasks with complex rock surface structures and diverse morphologies.

[0039] Optionally, obtaining the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data includes: A density-based noisy spatial clustering algorithm is used to separate the unsampled point cloud data according to the unsampled point cloud label set to obtain multiple rock mass structural surfaces; Based on the least square method, data fitting is performed according to the rock mass structural surface to obtain the corresponding plane normal vector; Obtaining the inclination angle and the dip of the rock mass structural surface according to the plane normal vector; The rock mass structural surface occurrence includes: , , in, is the inclination angle of the rock mass structural surface, is the inclination of the rock mass structural surface, (A, B, C) is the plane normal vector; The occurrence of the rock mass structural surface is obtained according to the inclination angle of the rock mass structural surface and the inclination of the rock mass structural surface.

[0040] Specifically, after obtaining the category label of each point, in order to further distinguish the structural planes belonging to the same category but in different positions (such as multiple parallel joint planes), it is necessary to perform cluster analysis on the points of each category, use the density-based spatial clustering with noise (DBSCAN) algorithm to separate the individual rock structural planes, and then further obtain the structural plane attitude for each structural plane. The normal vector of the rock structural plane is calculated based on the least squares fitting. Assuming that the plane equation is Ax+By+Cz+D=0, any given set of three-dimensional point cloud data , the goal is to find the appropriate coefficients (A, B, C, D) to minimize the sum of the squares of the distances from all points to the plane. Then the plane is a plane that meets the requirements. The plane normal vector is (A, B, C). The attitude of the rock mass structural surface is used to represent the orientation of the structural surface in three-dimensional space, usually described by dip and dip. The dip is used to represent the maximum downward direction perpendicular to the strike line, pointing to the downward-dipping side of the structural surface. The dip is used to represent the maximum inclination angle of the structural surface relative to the horizontal plane, usually expressed in degrees, ranging from 0° (completely horizontal) to 90° (completely vertical). The attitude of the rock mass structural surface can be described by the ratio of the dip of the rock mass structural surface and the dip of the rock mass structural surface.

[0041] Optionally, obtaining the sampled point cloud data and the unsampled point cloud data by uniformly downsampling the processed three-dimensional point cloud data includes: Dividing the processed three-dimensional point cloud data into a plurality of three-dimensional grids at a fixed size; Obtaining a grid center point according to each of the three-dimensional grids; Using all the grid center points as the sampling point cloud data; The unsampled processed point cloud data is obtained according to data in the processed three-dimensional point cloud data except the sampled processed point cloud data.

[0042] Specifically, uniform downsampling is performed on dense point clouds. Uniform downsampling of point clouds is a point cloud data processing technique that aims to reduce the number of points in a point cloud while maintaining its overall geometric shape and distribution characteristics, thereby improving data processing efficiency or reducing storage requirements. The specific implementation is as follows: the point cloud is divided into a fixed-size 3D grid (voxels); in each voxel, only one representative point (such as the center point or mean point) is retained.

[0043] Optionally, the dense three-dimensional point cloud data of the rock mass includes a plurality of point clouds, and the dense three-dimensional point cloud data of the rock mass is processed to obtain processed three-dimensional point cloud data, including: Based on the KD tree algorithm, corresponding neighborhood point sets are obtained according to the point clouds; The processed three-dimensional point cloud data is obtained by performing principal component analysis on the neighborhood point set.

[0044] Specifically, the KD tree method is used to calculate the neighborhood point set of any point cloud. The KD tree is a data structure in which each node is a hyperrectangular region in k-dimensional space. It can be seen as a generalization of the binary search tree (BST) in multidimensional space. The KD tree is constructed based on the point cloud data. This process recursively divides the space into two parts until the stopping condition is met. For a given point, the constructed KD tree is used to quickly obtain the neighborhood point set. Based on the neighborhood point set of the point cloud, its covariance matrix is ​​constructed. . Using PCA calculation The eigenvalue of and eigenvectors . Calculate each element in the eigenvector separately By comparing the corresponding modulus value, the normal vector of the point is determined. ,but This is the normal vector of the point .

[0045] .

[0046] Optionally, it also includes: The structural surface image is obtained by collecting rock structural surface data; Based on the SfM-MVS algorithm, the structural surface image is converted into point cloud data to obtain dense three-dimensional point cloud data of the rock mass.

[0047] Specifically, SfM (Structure from Motion) and MVS (Multi-View Stereo) are two techniques for reconstructing 3D models from 2D image sequences. They are often used in conjunction to generate high-quality 3D point clouds or mesh models from a series of uncalibrated 2D images.

[0048] like Figure 3 As shown, an embodiment of the present invention provides a rock structure surface identification device based on point cloud, comprising: The data processing module 10 is used to process the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; A downsampling module 20 is configured to uniformly downsample the processed three-dimensional point cloud data to obtain sampled point cloud data and unsampled point cloud data; A model training module 30 is used to train a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; An unsampled point cloud label set acquisition module 40 is configured to input the unsampled point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled point cloud label set; The rock mass structural surface occurrence acquisition module 50 is configured to obtain the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0049] The point cloud-based rock structure surface recognition device of this embodiment is used to implement the point cloud-based rock structure surface recognition method described above. Its advantages over the existing technology are the same as the advantages of the point cloud-based rock structure surface recognition method described above over the existing technology, and will not be repeated here.

[0050] Optionally, the model training module 30 is specifically configured to: classify the sampled processed point cloud data based on an agglomerative hierarchical clustering algorithm to obtain a corresponding sampled processed point cloud label set; Obtaining a point cloud data training set and a point cloud data test set according to the sampled processed point cloud data and the corresponding sampled processed point cloud label set; Training the neural network model using the point cloud data training set until the neural network converges to obtain an initial neural network model; The initial neural network model is tested using the point cloud data test set to test the test set accuracy of the initial neural network model. When the test set accuracy does not meet the accuracy requirement, more data is acquired for training until the accuracy requirement is met, thereby obtaining the rock mass structural surface recognition model. The test set accuracy includes: , Among them, AC is the accuracy of the test set, is the number of accurately classified point clouds, is the number of point clouds in the point cloud data test set.

[0051] Optionally, the model training module 30 is specifically configured to: obtain a similarity measure by processing the sampling point cloud data; The similarity measurement includes: , in, is the similarity metric, is the normal vector of the i-th point cloud, is the transposed matrix of the normal vector of the j-th point cloud; The sampled and processed point cloud data are grouped into structural surface categories according to the similarity metric to obtain a corresponding sampled and processed point cloud label set.

[0052] Optionally, the rock mass structural surface occurrence acquisition module 50 is specifically configured to: separate the unsampled processed point cloud data according to the unsampled processed point cloud label set using a density-based noisy spatial clustering algorithm to obtain a plurality of rock mass structural surfaces; Based on the least square method, data fitting is performed according to the rock mass structural surface to obtain the corresponding plane normal vector; Obtaining the inclination angle and the dip of the rock mass structural surface according to the plane normal vector; The rock mass structural surface occurrence includes: , , in, is the inclination angle of the rock mass structural surface, is the inclination of the rock mass structural surface, (A, B, C) is the plane normal vector; The occurrence of the rock mass structural surface is obtained according to the inclination angle of the rock mass structural surface and the inclination of the rock mass structural surface.

[0053] Optionally, the downsampling module 20 is specifically configured to: divide the processed three-dimensional point cloud data into fixed-size segments to obtain a plurality of three-dimensional grids; Obtaining a grid center point according to each of the three-dimensional grids; Using all the grid center points as the sampling point cloud data; The unsampled processed point cloud data is obtained according to data in the processed three-dimensional point cloud data except the sampled processed point cloud data.

[0054] Optionally, the data processing module 10 is specifically configured to: obtain corresponding neighborhood point sets according to the point clouds based on a KD tree algorithm; The processed three-dimensional point cloud data is obtained by performing principal component analysis on the neighborhood point set.

[0055] Optionally, the point cloud-based rock structure surface identification device further includes a data acquisition module, wherein the data acquisition module is configured to: obtain a structure surface image by collecting rock structure surface data; Based on the SfM-MVS algorithm, the structural surface image is converted into point cloud data to obtain dense three-dimensional point cloud data of the rock mass.

[0056] like Figure 4 As shown, an electronic device 400 provided by an embodiment of the present invention includes a memory 410 and a processor 420; the memory 410 is used to store computer programs; the processor 420 is used to implement the above-mentioned point cloud-based rock structure surface identification method when executing the computer program.

[0057] In other words, an electronic device 400 includes a memory 410 and a processor 420 coupled to the memory 410; the memory 410 is configured to store a computer program; and the processor 420 is configured to perform the following operations when executing the computer program: Processing the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; uniformly downsampling the processed three-dimensional point cloud data to obtain sampled processed point cloud data and unsampled processed point cloud data; Training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; Inputting the unsampled processed point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled processed point cloud label set; The occurrence of the rock mass structural surface is obtained according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0058] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the above-mentioned point cloud-based rock mass structural surface identification method is implemented.

[0059] In other words, a non-volatile computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following operations: Processing the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; uniformly downsampling the processed three-dimensional point cloud data to obtain sampled processed point cloud data and unsampled processed point cloud data; Training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; Inputting the unsampled processed point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled processed point cloud label set; The occurrence of the rock mass structural surface is obtained according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

[0060] An electronic device 400 that can serve as a server or client of the present invention will now be described, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device 400 is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 400 can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0061] Electronic device 400 includes a computing unit that can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) or a computer program loaded from a storage unit into a random access memory (RAM). The RAM can also store various programs and data required for device operation. The computing unit, ROM, and RAM are interconnected via a bus. An input / output (I / O) interface is also connected to the bus.

[0062] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM). In this application, the units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network elements. Some or all of these units can be selected based on actual needs to achieve the objectives of the embodiments of the present invention. Furthermore, the functional units in the various embodiments of the present invention can be integrated into a single processing unit, each unit can exist physically separately, or two or more units can be integrated into a single unit. These integrated units can be implemented in either hardware or software functional units.

[0063] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for identifying rock mass structural surfaces based on point cloud, characterized in that: include: Processing the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; uniformly downsampling the processed three-dimensional point cloud data to obtain sampled processed point cloud data and unsampled processed point cloud data; Training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; Inputting the unsampled processed point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled processed point cloud label set; The occurrence of the rock mass structural surface is obtained according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

2. The method for identifying rock mass structural surfaces based on point cloud according to claim 1, characterized in that: The step of training a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model includes: Based on the agglomerative hierarchical clustering algorithm, the sampled point cloud data is classified to obtain a corresponding sampled point cloud label set; Obtaining a point cloud data training set and a point cloud data test set according to the sampled processed point cloud data and the corresponding sampled processed point cloud label set; Training the neural network model using the point cloud data training set until the neural network converges to obtain an initial neural network model; The initial neural network model is tested using the point cloud data test set to test the test set accuracy of the initial neural network model. When the test set accuracy does not meet the accuracy requirement, more data is acquired for training until the accuracy requirement is met, thereby obtaining the rock mass structural surface recognition model. The test set accuracy includes: ， Among them, AC is the accuracy of the test set, is the number of accurately classified point clouds, is the number of point clouds in the point cloud data test set.

3. The method for identifying rock mass structural surfaces based on point cloud according to claim 2, characterized in that: The classifying and processing the sampled and processed point cloud data to obtain a corresponding sampled and processed point cloud label set includes: Processing the point cloud data by the sampling to obtain a similarity metric; The similarity measurement includes: ， in, is the similarity metric, is the normal vector of the i-th point cloud, is the transposed matrix of the normal vector of the j-th point cloud; The sampled and processed point cloud data are grouped into structural surface categories according to the similarity metric to obtain a corresponding sampled and processed point cloud label set.

4. The method for identifying rock mass structural surfaces based on point cloud according to claim 1, characterized in that: Obtaining the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data includes: A density-based noisy spatial clustering algorithm is used to separate the unsampled point cloud data according to the unsampled point cloud label set to obtain multiple rock mass structural surfaces; Based on the least square method, data fitting is performed according to the rock mass structural surface to obtain the corresponding plane normal vector; Obtaining the inclination angle and the dip of the rock mass structural surface according to the plane normal vector; The rock mass structural surface occurrence includes: ， ， in, is the inclination angle of the rock mass structural surface, is the inclination of the rock mass structural surface, (A, B, C) is the plane normal vector; The occurrence of the rock mass structural surface is obtained according to the inclination angle of the rock mass structural surface and the inclination of the rock mass structural surface.

5. The method for identifying rock mass structural surfaces based on point cloud according to claim 1, characterized in that: The step of uniformly downsampling the processed three-dimensional point cloud data to obtain sampled point cloud data and unsampled point cloud data comprises: Dividing the processed three-dimensional point cloud data into a plurality of three-dimensional grids at a fixed size; Obtaining a grid center point according to each of the three-dimensional grids; Using all the grid center points as the sampling point cloud data; The unsampled processed point cloud data is obtained according to data in the processed three-dimensional point cloud data except the sampled processed point cloud data.

6. The method for identifying rock mass structural surfaces based on point cloud according to claim 1, characterized in that: The rock mass dense three-dimensional point cloud data includes a plurality of point clouds, and the rock mass dense three-dimensional point cloud data is processed to obtain processed three-dimensional point cloud data, including: Based on the KD tree algorithm, corresponding neighborhood point sets are obtained according to the point clouds; The processed three-dimensional point cloud data is obtained by performing principal component analysis on the neighborhood point set.

7. The method for identifying rock mass structural surfaces based on point cloud according to claim 6, characterized in that: Also includes: The structural surface image is obtained by collecting rock structural surface data; Based on the SfM-MVS algorithm, the structural surface image is converted into point cloud data to obtain dense three-dimensional point cloud data of the rock mass.

8. A rock mass structural surface identification device based on point cloud, characterized in that: include: A data processing module is used to process the dense three-dimensional point cloud data of the rock mass to obtain processed three-dimensional point cloud data; a downsampling module, configured to uniformly downsample the processed three-dimensional point cloud data to obtain sampled point cloud data and unsampled point cloud data; A model training module is used to train a neural network model based on the sampled and processed point cloud data to obtain a rock mass structural surface recognition model; An unsampled point cloud label set acquisition module is used to input the unsampled point cloud data into the rock mass structural surface recognition model to obtain a corresponding unsampled point cloud label set; The rock mass structural surface occurrence acquisition module is used to obtain the rock mass structural surface occurrence according to the unsampled processed point cloud label set and the unsampled processed point cloud data.

9. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is configured to implement the point cloud-based rock mass structural surface identification method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by the processor, the point cloud-based rock structure surface identification method according to any one of claims 1 to 7 is implemented.

Images