A method, device, equipment and readable storage medium for identifying slope structural planes

Abstract

The present invention provides a method, device, equipment and readable storage medium for identifying slope structural planes, relating to the technical field of three-dimensional point cloud structural plane identification, including obtaining a three-dimensional point cloud data set; obtaining a plurality of initial point cloud sets; obtaining a filtered target point cloud set; obtaining the normal vector of each point in the target point cloud set; obtaining the structural plane grouping category corresponding to each point in the target point cloud set; and calculating according to the structural plane grouping category to obtain a target calculation result. The present invention establishes a three-dimensional point cloud set of the slope by transforming and reconstructing the slope image, and screens and eliminates boundary points through a curvature threshold to improve the calculation accuracy of the normal vector. The remaining center points are input into a preset neural network for identification and classification to obtain the corresponding structural plane grouping category, so as to calculate the dip direction and dip angle of any structural plane. The present invention combines a neural network model to achieve accurate identification of rock mass structural planes, effectively improving the identification efficiency of rock mass structural planes.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional point cloud structural plane recognition. Specifically, it relates to a method, device, equipment and readable storage medium for slope structural plane recognition. Background Art

[0002] In the existing research on slope structural planes, the research object is usually a natural outcrop slope, which is characterized by being undamaged, having a complete original structure, many outcropping structures, obvious structural features, large surface curvature changes, etc. The existing technical means can be used to extract the features of such body structures. However, for a large number of high-steep fractured rock slopes, which are characterized by non-obvious surface structural features and non-obvious surface curvature, the existing technical means have low accuracy, poor effect and slow speed in the application of high-steep slope feature recognition, and it is very difficult to play a role. Therefore, it is urgent to propose an intelligent analysis and recognition method for three-dimensional point cloud structural planes of complex rock masses. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, device, equipment and readable storage medium for slope structural plane recognition to improve the above problems. To achieve the above purpose, the technical solutions adopted by the present invention are as follows:

[0004] In a first aspect, the present application provides a method for slope structural plane recognition, including:

[0005] Obtain a slope image, and perform transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud data set;

[0006] Classify the three-dimensional point cloud data set based on the nearest neighbor search algorithm to obtain a plurality of initial point cloud sets, and each initial point cloud set is composed of target points and neighbor points;

[0007] Calculate the curvature value of each target point based on the plurality of initial point cloud sets, and perform coplanar screening of points based on the curvature value to obtain a screened target point cloud set;

[0008] Calculate the normal vector of each point in the target point cloud set according to the target point cloud set;

[0009] Input the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set;

[0010] Calculate according to the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result, and the target calculation result includes the dip angle and dip direction of each structural plane in the structural plane grouping category.

[0011] In a second aspect, the present application also provides a device for slope structural plane recognition, including:

[0012] A first acquisition unit for acquiring a slope image and performing transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud dataset;

[0013] A classification unit for classifying the three-dimensional point cloud dataset based on a nearest neighbor search algorithm to obtain a plurality of initial point cloud sets, where each initial point cloud set consists of target points and neighbor points;

[0014] A screening unit for calculating the curvature value of each target point based on a plurality of initial point cloud sets and performing coplanarity screening of points based on the curvature value to obtain a screened target point cloud set;

[0015] A first obtaining unit for calculating based on the target point cloud set to obtain the normal vector of each point in the target point cloud set;

[0016] An input unit for inputting the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set;

[0017] A first calculation unit for calculating based on the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result, where the target calculation result includes the dip angle and dip direction of each structural plane in the structural plane grouping category.

[0018] In a third aspect, the present application further provides a slope structural plane recognition device, including:

[0019] A memory for storing a computer program;

[0020] A processor for implementing the steps of the slope structural plane recognition method when executing the computer program.

[0021] In a fourth aspect, the present application further provides a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned slope structural plane recognition method are implemented.

[0022] The beneficial effects of the present invention are:

[0023] The present invention establishes a slope three-dimensional point cloud set by performing transformation and reconstruction on a slope image, and screens and eliminates boundary points through a curvature threshold to improve the calculation accuracy of the normal vector. The remaining center points are input into a preset neural network for recognition and classification to obtain the corresponding structural plane grouping category, so as to calculate the dip direction and dip angle of any structural plane. The present invention combines a neural network model to achieve accurate recognition of rock mass structural planes and effectively improves the recognition efficiency of rock mass structural planes.

[0024] Other features and advantages of the present invention will be described in the subsequent specification, and in part will be obvious from the specification, or can be understood by implementing the embodiments of the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the written specification, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and thus should not be regarded as a limitation of the scope. For those of ordinary skill in the art, without creative efforts, other relevant drawings can also be obtained based on these drawings.

[0026] Figure 1 Schematic flow chart of the slope structural plane identification method described in the embodiments of the present invention;

[0027] Figure 2 Schematic diagram of the initial point cloud set described in the embodiments of the present invention;

[0028] Figure 3 Schematic diagram of the geometric coordinates of the normal vector described in the embodiments of the present invention;

[0029] Figure 4 Schematic structural diagram of the slope structural plane identification device described in the embodiments of the present invention;

[0030] Figure 5 Schematic structural diagram of the slope structural plane identification device described in the embodiments of the present invention.

[0031] Reference numerals in the drawings:

[0032] 10. First acquisition unit; 20. Classification unit; 30. Screening unit; 40. First obtaining unit; 50. Substitution unit; 60. First calculation unit; 800. Slope structural plane identification device; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0034] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0035] Example 1:

[0036] This embodiment provides a slope structural surface identification method.

[0037] See also Figure 1 , the figure shows that the method includes step S10, step S20, step S30, step S40, step S50 and step S60.

[0038] Step S10: Obtain a slope image, and perform transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud dataset;

[0039] Specifically, considering the complex terrain at the location of the slope, drones are usually used to capture slope images. First, it is necessary to determine the key research area based on the on-site environment of the slope and formulate a flight plan for the drone. The flight plan needs to include the drone's take-off and landing locations, flight time, flight trajectory, detection control points, and key detection areas. It is necessary to ensure the flight safety of the drone and to ensure that comprehensive slope images can be collected. All the collected slope image data are converted into a three-dimensional point cloud dataset through the Structure From Motion (SFM) method.

[0040] Step S20: classifying the three-dimensional point cloud dataset based on a nearest neighbor search algorithm to obtain a plurality of initial point cloud sets, where the initial point cloud sets are composed of target points and nearest neighbor points;

[0041] Specifically, the k-nearest neighbor search algorithm is used to perform neighbor search on the 3D point cloud dataset, such as Figure 2As shown, for any point in the three-dimensional point cloud, multiple adjacent points will be found to jointly form a point cloud set. The area within the dashed line is an initial point cloud set, the black dot is the target point, the white dots are adjacent points, and the gray dots are the remaining points.

[0042] Step S30. Calculate the curvature value of each target point based on multiple initial point cloud sets, and perform coplanar screening of points based on the curvature value to obtain the screened target point cloud set;

[0043] Specifically, calculate the curvature value of each point in the initial point cloud set through the principal component analysis algorithm and perform the detection of whether they are coplanar.

[0044] Specifically, step S30 specifically includes step S31, step S32, step S33, step S34, and step S35:

[0045] Step S31. Calculate based on the initial point cloud set to obtain the covariance matrix;

[0046] Step S32. Perform eigenvalue decomposition based on the covariance matrix to obtain multiple eigenvalues;

[0047] Specifically, the principal component analysis algorithm automatically calculates 3 eigenvalues of each point, namely λ1, λ2, and λ3, where λ1 ≥ λ2 ≥ λ3.

[0048] Step S33. Calculate the curvature based on multiple eigenvalues to obtain the curvature value of the target point in the initial point cloud set;

[0049] Step S34. Compare the curvature value of each target point with the first set threshold, and screen out multiple basic points. The basic points correspond to curvature values less than the first set threshold;

[0050] Step S35. Form the target point cloud set based on all basic points;

[0051] Specifically, the calculation formula for the curvature value is:

[0052]

[0053] where φ is the curvature value; λ1, λ2, and λ3 are eigenvalues, and λ1 ≥ λ2 ≥ λ3;

[0054] Compare the calculated curvature value with the preset curvature threshold, and consider the points with curvature values greater than the curvature threshold as boundary points and eliminate them.

[0055] Step S40. Calculate based on the target point cloud set to obtain the normal vector of each point in the target point cloud set;

[0056] Specifically, use the least squares method to calculate the normal vector of the coplanar points. The expression of the fitted plane is:

[0057] A i x + B i y + C i z + D i =0

[0058] Wherein, A i 、B i and C i respectively represent the three - dimensional coordinates of the normal vector of the fitting plane where the i - th point is located; x, y, and z respectively represent the three - dimensional coordinates of the i - th point, and D i is the distance from the i - th point to the center point of the fitting plane.

[0059] Step S50. Input the normal vector into the pre - trained neural network model to obtain the grouped structural surface categories corresponding to each point in the target point cloud set;

[0060] Specifically, in this application, a self - organizing map neural network model is used as the recognition model. Different from the general neural network that is trained based on the back - propagation of the loss function, the self - organizing map neural network model uses a competitive learning strategy, relies on the mutual competition between neurons to gradually optimize the network, and uses a neighborhood relationship function to maintain the topological structure of the input space. Therefore, compared with the neural network that requires supervised learning, the training stage of the self - organizing map neural network model does not require sample labels, can cluster data without knowing the categories, greatly improving the efficiency of data processing. The recognition stage of the self - organizing map neural network model also has high recognition efficiency and recognition accuracy.

[0061] Specifically, step S50 specifically includes step S51, step S52, step S53, step S54, step S55, step S56, step S57, step S58, step S59, and step S510:

[0062] Step S51. Obtain the initial learning rate, initial neighborhood radius, initial weight vector, maximum number of iterations, and set the number of nodes of the neural network model;

[0063] Step S52. Calculate the distance between each point in the target point cloud set and each neuron in the neural network model, and determine the target neuron corresponding to the minimum distance;

[0064] Specifically, randomly select the normal vector of a point in the input target point cloud set as the input sample, calculate the distances between the input sample and multiple neurons, and find the neuron with the minimum distance as the winning neuron.

[0065] In this application, the square of the improved cosine distance is used to calculate the distance between the input sample and the neuron, and the corresponding calculation formula is:

[0066]

[0067] Among them, P i is the i-th point in the target point cloud set; P j is the j-th neuron; is the normal vector of the i-th point in the target point cloud set; is the transpose of the normal vector of the j-th neuron.

[0068] Step S53. Calculation operation: Calculate the winning neighborhood based on the initial neighborhood radius and the minimum distance corresponding to the target neuron;

[0069] Specifically, the calculation formula for the winning neighborhood is:

[0070]

[0071] Among them, h(i) is the winning neighborhood of the i-th point in the target point cloud set; d min (i) is the distance between the i-th point in the target point cloud set and the winning neuron; δ(t - 1) is the initial neighborhood radius after the (t - 1)-th iteration update.

[0072] Step S54. Acquisition operation: Obtain the calculation times of the winning neighborhood as the number of iterations;

[0073] Step S55. First update operation: Update the initial learning rate based on the initial learning rate, the number of iterations, and the maximum number of iterations;

[0074] Specifically, the update formula for the initial learning rate is:

[0075]

[0076] Among them, α(t) is the initial learning rate after the t-th iteration update; α0 is the initial learning rate; t is the number of iterations; T is the maximum number of iterations.

[0077] Step S56. Second update operation: Update the initial neighborhood radius based on the initial neighborhood radius, the number of iterations, and the maximum number of iterations;

[0078] Specifically, the update formula for the initial neighborhood radius is:

[0079]

[0080] Among them, δ(t) is the initial neighborhood radius after the t-th iteration update; k is the set number of nodes; t is the number of iterations; T is the maximum number of iterations.

[0081] Step S57. Third update operation: Update the initial weight vector based on the initial weight vector, the winning neighborhood, the initial weight vector, the initial learning rate, the normal vector, and the initial neighborhood radius;

[0082] Specifically, the update formula for the initial weight vector is as follows:

[0083]

[0084] Among them, W j (t + 1) is the weight vector after the (t + 1)-th iteration update of the j-th neuron; W j (t) is the weight vector after the t-th iteration update of the j-th neuron; α(t) is the initial learning rate after the i-th iteration update; h(t) is the winning neighborhood of the i-th point in the target point cloud set; is the normal vector of the i-th point in the target point cloud set.

[0085] Step S58. Repeat the calculation operation, acquisition operation, first update operation, second update operation, and third update operation until the number of iterations is greater than the maximum number of iterations. Based on the updated initial weight vector, all points in the target point cloud set are divided into structural plane classification clusters, and the structural plane classification clusters contain multiple structural plane clusters;

[0086] Specifically, repeat the above iterative operation until the maximum number of iterations is reached.

[0087] Step S59. Adjust the set number of nodes to obtain multiple corresponding structural plane classification clusters, and the number of structural plane clusters in each structural plane classification cluster is different;

[0088] Specifically, by changing the set number of nodes k in the output layer, different numbers of structural plane classification clusters can be obtained. Generally, the value of the set number of nodes is an integer between 2 and 8, and the number of nodes can be adjusted according to the application scenario, and no special restrictions are made here.

[0089] Step S510. Determine the optimal structural plane classification cluster from multiple structural plane classification clusters as the structural plane grouping category;

[0090] Specifically, step S510 specifically includes step S5101, step S5102, step S5103, step S5104, step S5105, step S5106, step S5107, step S5108, and step S5109:

[0091] Step S5101. Calculate the trace of the covariance matrix of the structural plane clusters based on the structural plane classification clusters to obtain the first value;

[0092] Step S5102. Calculate the trace of the covariance matrix between the structural plane clusters based on the structural plane classification clusters to obtain the second value;

[0093] Step S5103. Calculate the difference between the set number of nodes and the second set threshold to obtain the third value;

[0094] Step S5104. Calculate the difference between the number of points in the target point cloud set and the number of structural plane clusters in the structural plane classification clusters to obtain a fourth value;

[0095] Step S5105. Calculate the ratio of the second value to the first value to obtain a fifth value;

[0096] Step S5106. Calculate the ratio of the fourth value to the third value to obtain a correction factor;

[0097] Step S5107. Calculate the product of the fifth value and the correction factor to obtain an exponential value;

[0098] Step S5108. Determine the maximum value from the multiple exponential values corresponding to the multiple structural plane classification clusters as the target exponential value;

[0099] Step S5109. Take the structural plane classification cluster corresponding to the target exponential value as the optimal classification cluster, and take the optimal classification cluster as the structural plane grouping category;

[0100] Specifically, according to the Calinski-Harabasz index as the evaluation index of clustering validity, determine the optimal classification cluster. The calculation formula of the Calinski-Harabasz index is:

[0101]

[0102] where CH is the Calinski-Harabasz index; tr(B K ) is the trace of the covariance matrix B K of the structural plane clusters; tr(W K ) is the trace of the covariance matrix W K of the structural plane clusters; N is the number of points in the target point cloud set; k is the set number of nodes.

[0103] Take the structural plane classification cluster corresponding to the maximum Calinski-Harabasz index as the structural plane classification group.

[0104] Step S60. Calculate according to the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result. The target calculation result includes the dip angle and dip direction of each structural plane in the structural plane grouping category;

[0105] Specifically, Step S60 specifically includes Step S61, Step S62, Step S63, Step S64, Step S65, Step S66 and Step S67:

[0106] Step S61. Extract the structural plane grouping category based on the spatial clustering algorithm to obtain multiple single structural planes;

[0107] Step S62. Calculate the normal vector of each single structural plane to obtain multiple target normal vectors. The target normal vector consists of the abscissa, ordinate and vertical coordinate;

[0108] Specifically, a spatial clustering algorithm is used to extract a single structural plane from the grouped categories of structural planes, and the normal vector of each structural plane is calculated. This normal vector is a three-dimensional coordinate, including the abscissa, ordinate, and vertical coordinate. The specific calculation of the structural plane normal vector is the same as the calculation principle of the normal vector of each point, which will not be elaborated here.

[0109] Step S63. Calculate the arccosine function value of the vertical coordinate to obtain the dip angle;

[0110] Specifically, according to the definition of the structural plane attitude and Figure 3 from the schematic diagram of the geometric coordinates of the normal vector in [reference], the dip angle is the angle between the structural plane and the horizontal plane, that is, the angle between the component C of the normal vector n = (A, B, C) on the Z-axis and the Z-axis. The calculation formula is:

[0111] α = arccos(C)

[0112] where α is the dip angle of the structural plane; C is the vertical coordinate of the structural plane normal vector.

[0113] Step S64. Calculate based on the abscissa and ordinate to obtain the initial trend;

[0114] Specifically, Step S64 specifically includes Step S641, Step S642, Step S643, and Step S644:

[0115] Step S641. Calculate the sum of the squares of the abscissa and the ordinate to obtain the first result;

[0116] Step S642. Perform a square root calculation on the first result to obtain the second result;

[0117] Step S643. Calculate the ratio of the ordinate to the second result to obtain the third result;

[0118] Step S644. Calculate the arccosine function value of the third result to obtain the initial trend;

[0119] Specifically, the initial trend calculation formula is:

[0120]

[0121] where β0 is the initial trend; B is the ordinate of the structural plane normal vector; A is the abscissa of the structural plane normal vector.

[0122] Step S65. Compare the abscissa with the third set threshold to obtain a comparison result;

[0123] Step S66. When the comparison result indicates that the abscissa is less than the third set threshold, calculate the difference between the fourth set threshold and the initial trend to obtain the trend;

[0124] Step S67. When the comparison result indicates that the abscissa is not less than the third set threshold, use the initial inclination as the inclination;

[0125] Specifically, as Figure 3 shown, the inclination is the angle between the projection of the unit normal vector n = (A, B, C) on the XOY plane and the positive direction of the Y axis. According to the different quadrants of the normal vector projection on the XOY plane, there are multiple cases, and its calculation formula can be simplified as:

[0126]

[0127] where β is the inclination; β0 is the initial inclination; A is the abscissa of the structural plane normal vector.

[0128] Use the average value of the dip angles and inclinations of all single structural planes in each category of structural planes as the dip angle and inclination values of the structural planes in this category, and realize the identification and classification of slope structural planes.

[0129] Embodiment 2:

[0130] As Figure 4 shown, this embodiment provides a slope structural plane identification device, and the device includes:

[0131] The first acquisition unit 10 is used to acquire a slope image and perform transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud data set;

[0132] The classification unit 20 is used to classify the three-dimensional point cloud data set based on the nearest neighbor search algorithm to obtain a plurality of initial point cloud sets, and the initial point cloud sets are composed of target points and neighboring points;

[0133] The screening unit 30 is used to calculate the curvature value of each target point based on a plurality of initial point cloud sets, and perform coplanar screening of points based on the curvature value to obtain a screened target point cloud set;

[0134] The first obtaining unit 40 is used to calculate according to the target point cloud set to obtain the normal vector of each point in the target point cloud set;

[0135] The input unit 50 is used to input the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set;

[0136] The first calculation unit 60 is used to calculate according to the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result, and the target calculation result includes the dip angle and inclination of each structural plane in the structural plane grouping category.

[0137] In a specific implementation manner disclosed in this application, the screening unit 30 includes:

[0138] A second calculation unit, configured to perform calculations based on the initial point cloud set to obtain a covariance matrix;

[0139] A decomposition unit, configured to perform eigenvalue decomposition based on the covariance matrix to obtain a plurality of eigenvalues;

[0140] A second obtaining unit, configured to calculate a curvature based on the plurality of eigenvalues to obtain a curvature value of a target point in the initial point cloud set;

[0141] A first comparison unit, configured to compare the curvature value of each target point with a first set threshold, and filter out a plurality of basic points, where the basic points correspond to curvature values less than the first set threshold;

[0142] A composition unit, configured to form a target point cloud set based on all the basic points.

[0143] In a specific implementation manner disclosed in the present application, the bringing-in unit 50 includes:

[0144] A second obtaining unit, configured to obtain an initial learning rate, an initial neighborhood radius, an initial weight vector, a maximum number of iterations, and a set number of nodes of a neural network model;

[0145] A third calculation unit, configured to calculate the distance between each point in the target point cloud set and each neuron in the neural network model, and determine a target neuron corresponding to the minimum distance;

[0146] A fourth calculation unit, configured to perform an operation: calculate a winning neighborhood based on the initial neighborhood radius and the minimum distance corresponding to the target neuron;

[0147] A third obtaining unit, configured to perform an operation: obtain the number of calculation times of the winning neighborhood as the number of iterations;

[0148] A first updating unit, configured to perform a first updating operation: update the initial learning rate based on the initial learning rate, the number of iterations, and the maximum number of iterations;

[0149] A second updating unit, configured to perform a second updating operation: update the initial neighborhood radius based on the initial neighborhood radius, the number of iterations, and the maximum number of iterations;

[0150] A third updating unit, configured to perform a third updating operation: update the initial weight vector based on the initial weight vector, the winning neighborhood, the initial weight vector, the initial learning rate, the normal vector, and the initial neighborhood radius;

[0151] A repeating unit, configured to repeat the calculation operation, the obtaining operation, the first updating operation, the second updating operation, and the third updating operation until the number of iterations is greater than the maximum number of iterations, and divide all the points in the target point cloud set into structural plane classification clusters based on the updated initial weight vector, where the structural plane classification clusters include a plurality of structural plane clusters;

[0152] An adjustment unit for adjusting the set number of nodes to obtain multiple corresponding structural plane classification clusters, where the number of structural plane clusters in each structural plane classification cluster is different;

[0153] A first determination unit for determining the optimal structural plane classification cluster from multiple structural plane classification clusters as the structural plane grouping category.

[0154] In a specific implementation manner disclosed in the present application, the first determination unit includes:

[0155] A third obtaining unit for calculating the trace of the covariance matrix of the structural plane clusters based on the structural plane classification clusters to obtain a first value;

[0156] A fourth obtaining unit for calculating the trace of the covariance matrix between the structural plane clusters based on the structural plane classification clusters to obtain a second value;

[0157] A fifth calculation unit for calculating the difference between the set number of nodes and a second set threshold to obtain a third value;

[0158] A sixth calculation unit for calculating the difference between the number of points in the target point cloud set and the number of structural plane clusters in the structural plane classification clusters to obtain a fourth value;

[0159] A seventh calculation unit for calculating the ratio of the second value to the first value to obtain a fifth value;

[0160] An eighth calculation unit for calculating the ratio of the fourth value to the third value to obtain a correction factor;

[0161] A ninth calculation unit for calculating the product of the fifth value and the correction factor to obtain an exponential value;

[0162] A second determination unit for determining the maximum value from multiple exponential values corresponding to multiple structural plane classification clusters as the target exponential value;

[0163] A first acting unit for taking the structural plane classification cluster corresponding to the target exponential value as the optimal classification cluster and taking the optimal classification cluster as the structural plane grouping category.

[0164] In a specific implementation manner disclosed in the present application, the first calculation unit 60 includes:

[0165] An extraction unit for extracting the structural plane grouping category based on a spatial clustering algorithm to obtain multiple single structural planes;

[0166] A tenth calculation unit for calculating the normal vector of each single structural plane to obtain multiple target normal vectors, where the target normal vector consists of an abscissa, an ordinate, and a vertical coordinate;

[0167] An eleventh calculation unit for calculating the arccosine function value of the vertical coordinate to obtain the dip angle;

[0168] The twelfth calculation unit is configured to perform calculations based on the abscissa and ordinate to obtain an initial inclination.

[0169] The second comparison unit is configured to compare the abscissa with a third set threshold to obtain a comparison result.

[0170] The fifth obtaining unit is configured to calculate the difference between a fourth set threshold and the initial inclination when the comparison result indicates that the abscissa is less than the third set threshold, so as to obtain an inclination.

[0171] The second acting unit is configured to use the initial inclination as the inclination when the comparison result indicates that the abscissa is not less than the third set threshold.

[0172] In a specific implementation manner disclosed in the present application, the fifth obtaining unit includes:

[0173] The thirteenth calculation unit is configured to calculate the sum of the squares of the abscissa and the ordinate to obtain a first result.

[0174] The square root calculation unit is configured to perform a square root calculation on the first result to obtain a second result.

[0175] The fourteenth calculation unit is configured to calculate the ratio of the ordinate to the second result to obtain a third result.

[0176] The fifteenth calculation unit is configured to calculate the arccosine function value of the third result to obtain the initial inclination.

[0177] It should be noted that regarding the device in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0178] Embodiment 3:

[0179] Corresponding to the above method embodiment, a slope structural plane recognition device is further provided in this embodiment. A slope structural plane recognition device described below can be correspondingly referred to the slope structural plane recognition method described above.

[0180] Figure 5 It is a block diagram of a slope structural plane recognition device 800 shown according to an exemplary embodiment. As Figure 5 shown, the slope structural plane recognition device 800 may include: a processor 801, a memory 802. The slope structural plane recognition device 800 may further include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0181] Among them, the processor 801 is used to control the overall operation of the slope structural plane recognition device 800 to complete all or part of the steps in the above-mentioned slope structural plane recognition method. The memory 802 is used to store various types of data to support the operation of the slope structural plane recognition device 800. These data may include, for example, instructions for any application or method operating on the slope structural plane recognition device 800, as well as application-related data, such as contact data, sent and received messages, pictures, audio, video, and so on. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc. The multimedia component 803 may include a screen and an audio component. Among them, the screen may be a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone, and the microphone is used to receive external audio signals. The received audio signals may be further stored in the memory 802 or sent through the communication component 805. The audio component further includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the slope structural plane recognition device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination of one or more of them. Therefore, the corresponding communication component 805 may include: a Wi-Fi module, a Bluetooth module, and an NFC module.

[0182] In an exemplary embodiment, the slope structural plane recognition device 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components, and is used to execute the above-mentioned slope structural plane recognition method.

[0183] In another exemplary embodiment, there is also provided a computer-readable storage medium including program instructions. When the program instructions are executed by a processor, the steps of the above-mentioned slope structural plane recognition method are implemented. For example, the computer-readable storage medium may be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions may be executed by the processor 801 of the slope structural plane recognition device 800 to complete the above-mentioned slope structural plane recognition method.

[0184] Embodiment 4:

[0185] Corresponding to the above method embodiment, in this embodiment, there is also provided a readable storage medium. A readable storage medium described below and a slope structural plane recognition method described above can be referred to each other correspondingly.

[0186] A readable storage medium has a computer program stored thereon. When the computer program is executed by a processor, the steps of the slope structural plane recognition method in the above method embodiment are implemented.

[0187] Specifically, the readable storage medium may be various readable storage media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

[0188] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention may have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0189] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for identifying slope structural planes, characterized in that, Including: Obtain a slope image, and perform transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud dataset; Classify the three-dimensional point cloud dataset based on the nearest neighbor search algorithm to obtain multiple initial point cloud sets, and each initial point cloud set is composed of target points and neighboring points; Calculate the curvature value of each target point based on multiple initial point cloud sets, and perform coplanar screening of points based on the curvature value to obtain a screened target point cloud set; Calculate the normal vector of each point in the target point cloud set based on the target point cloud set; Input the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set; Calculate based on the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result, and the target calculation result includes the dip angle and dip direction of each structural plane in the structural plane grouping category; Among them, inputting the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set includes: Obtain the initial learning rate, initial neighborhood radius, initial weight vector, maximum number of iterations, and set number of nodes of the neural network model; Calculate the distance between each point in the target point cloud set and each neuron in the neural network model, and determine the target neuron corresponding to the minimum distance; Calculation operation: Calculate the winning neighborhood based on the initial neighborhood radius and the minimum distance corresponding to the target neuron; Obtaining operation: Obtain the number of calculations of the winning neighborhood as the number of iterations; First update operation: Update the initial learning rate based on the initial learning rate, the number of iterations, and the maximum number of iterations; Second update operation: Update the initial neighborhood radius based on the initial neighborhood radius, the number of iterations, and the maximum number of iterations; Third update operation: Update the initial weight vector based on the initial weight vector, the winning neighborhood, the initial weight vector, the initial learning rate, the normal vector, and the initial neighborhood radius; Repeat the calculation operation, the obtaining operation, the first update operation, the second update operation, and the third update operation until the number of iterations is greater than the maximum number of iterations, and divide all points in the target point cloud set into structural plane classification clusters based on the updated initial weight vector, and the structural plane classification clusters contain multiple structural plane clusters; Adjust the set number of nodes to obtain multiple corresponding structural plane classification clusters, and the number of structural plane clusters in each structural plane classification cluster is different; Determine the optimal structural plane classification cluster from multiple structural plane classification clusters as the structural plane grouping category.

2. The slope structural plane identification method according to claim 1, wherein , calculating the curvature value of each target point based on multiple initial point cloud sets, and performing coplanar screening of points based on the curvature value to obtain a screened target point cloud set, including: Calculate based on the initial point cloud set to obtain a covariance matrix; Perform eigenvalue decomposition based on the covariance matrix to obtain multiple eigenvalues; Calculate the curvature based on multiple eigenvalues to obtain the curvature value of the target point in the initial point cloud set; Compare the curvature value of each target point with a first set threshold, and filter out a plurality of basic points, where the curvature value corresponding to the basic points is less than the first set threshold; Construct the target point cloud set based on all the basic points.

3. The slope structural plane identification method according to claim 1, wherein , Determine the optimal structural plane classification cluster from a plurality of structural plane classification clusters as the structural plane grouping category, including: Calculate the trace of the covariance matrix of the structural plane cluster based on the structural plane classification cluster to obtain a first value; Calculate the trace of the covariance matrix between the structural plane clusters based on the structural plane classification cluster to obtain a second value; Calculate the difference between the set number of nodes and a second set threshold to obtain a third value; Calculate the difference between the number of points in the target point cloud set and the number of structural plane clusters in the structural plane classification cluster to obtain a fourth value; Calculate the ratio of the second value to the first value to obtain a fifth value; Calculate the ratio of the fourth value to the third value to obtain a correction factor; Calculate the product of the fifth value and the correction factor to obtain an exponential value; Determine the maximum value from a plurality of exponential values corresponding to a plurality of structural plane classification clusters as the target exponential value; Use the structural plane classification cluster corresponding to the target exponential value as the optimal structural plane classification cluster, and use the optimal structural plane classification cluster as the structural plane grouping category.

4. A slope structural plane recognition device, characterized in that, Including: A first acquisition unit for acquiring a slope image and performing transformation and reconstruction based on the slope image to obtain a three-dimensional point cloud data set; A classification unit for classifying the three-dimensional point cloud data set based on a nearest neighbor search algorithm to obtain a plurality of initial point cloud sets, where the initial point cloud sets are composed of target points and neighboring points; A screening unit for calculating the curvature value of each target point based on a plurality of initial point cloud sets and performing coplanar screening of points based on the curvature value to obtain a screened target point cloud set; A first obtaining unit for calculating according to the target point cloud set to obtain the normal vector of each point in the target point cloud set; An input unit for inputting the normal vector into a pre-trained neural network model to obtain the structural plane grouping category corresponding to each point in the target point cloud set; A first calculation unit for calculating according to the structural plane grouping category corresponding to each point in the target point cloud set to obtain a target calculation result, where the target calculation result includes the dip angle and dip direction of each structural plane in the structural plane grouping category; Among them, the input unit includes: A second acquisition unit for acquiring the initial learning rate, initial neighborhood radius, initial weight vector, maximum number of iterations, and set number of nodes of the neural network model; A third calculation unit for calculating the distance between each point in the target point cloud set and each neuron in the neural network model and determining the target neuron corresponding to the minimum distance; A fourth calculation unit for calculating an operation: calculating a winning neighborhood based on the initial neighborhood radius and the minimum distance corresponding to the target neuron; A third acquisition unit for acquiring an operation: acquiring the number of calculations of the winning neighborhood as the number of iterations; A first update unit for a first update operation: updating the initial learning rate based on the initial learning rate, the number of iterations, and the maximum number of iterations; A second update unit for a second update operation: updating the initial neighborhood radius based on the initial neighborhood radius, the number of iterations, and the maximum number of iterations; A third update unit for a third update operation: updating the initial weight vector based on the initial weight vector, the winning neighborhood, the initial weight vector, the initial learning rate, the normal vector, and the initial neighborhood radius; A repetition unit for repeating the calculation operation, the acquisition operation, the first update operation, the second update operation, and the third update operation until the number of iterations is greater than the maximum number of iterations, and partitioning all points in the target point cloud set into structural plane classification clusters based on the updated initial weight vector, where the structural plane classification clusters include multiple structural plane clusters; An adjustment unit for adjusting the set number of nodes to obtain multiple corresponding structural plane classification clusters, and the number of structural plane clusters in each structural plane classification cluster is different; A first determination unit for determining an optimal structural plane classification cluster from multiple structural plane classification clusters as the structural plane grouping category.

5. The slope structural plane recognition device according to claim 4, characterized in that, The screening unit includes: A second calculation unit for calculating a covariance matrix based on the initial point cloud set; A decomposition unit for performing eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues; A second obtaining unit for calculating the curvature based on multiple eigenvalues to obtain the curvature value of the target point in the initial point cloud set; A first comparison unit for comparing the curvature value of each target point with a first set threshold to screen out multiple basic points, where the basic points correspond to curvature values less than the first set threshold; A composition unit for forming the target point cloud set based on all basic points.

6. The slope structural plane recognition device according to claim 4, characterized in that The determination unit includes: A third obtaining unit for calculating the trace of the covariance matrix of the structural plane cluster based on the structural plane classification cluster to obtain a first value; A fourth obtaining unit for calculating the trace of the covariance matrix between structural plane clusters based on the structural plane classification cluster to obtain a second value; A fifth calculation unit for calculating the difference between the set number of nodes and a second set threshold to obtain a third value; A sixth calculation unit for calculating the difference between the number of points in the target point cloud set and the number of structural plane clusters in the structural plane classification cluster to obtain a fourth value; A seventh calculation unit for calculating the ratio of the second value to the first value to obtain a fifth value; An eighth calculation unit for calculating the ratio of the fourth value to the third value to obtain a correction factor; A ninth calculation unit for calculating the product of the fifth value and the correction factor to obtain an exponential value; A second determination unit for determining the maximum value from multiple exponential values corresponding to multiple structural plane classification clusters as the target exponential value; A first acting unit for using the structural plane classification cluster corresponding to the target exponential value as the optimal structural plane classification cluster and using the optimal structural plane classification cluster as the structural plane grouping category.

7. An equipment for identifying slope structural planes, characterized in that, It includes: A memory for storing a computer program; A processor for implementing the steps of the slope structural plane recognition method according to any one of claims 1 to 3 when executing the computer program.

8. A readable storage medium, characterized in that, A computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the steps of the slope structural plane recognition method according to any one of claims 1 to 3 are implemented.

Images