A method and system for automatically identifying linear structural surfaces in complex rock outcrops

Abstract

The present invention provides a method and system for automatically identifying linear structural surfaces in complex rock outcrops, belonging to the field of rock engineering. The method involves using a DeepLabV3+ model to semantically segment the input image of a rock outcrop, outputting a binary mask of structural surface trace pixels and performing topological reconstruction; establishing a mapping relationship between the image coordinate system and a three-dimensional point cloud; calculating the spatial angle between the direction vector of the trace and the average attitude of the structural surface group; classifying the three-dimensional point cloud data by structural surface group based on an angle threshold; and extracting three-dimensional line segments using an improved iterative Hough transform algorithm. Finally, a greedy algorithm is used to implement global multi-segment connectivity and trace length calculation for discrete line segments. This method overcomes the topographical limitations of manual measurement when obtaining rock structural surface data, improving the accuracy and efficiency of structural surface extraction in complex rock masses.

Description

Technical Field

[0001] The present invention belongs to the field of rock mass engineering, and in particular relates to a method and system for automatically identifying linear structural surfaces in complex rock mass outcrops. Background Art

[0002] Based on the spatial intersection between structural surfaces and rock outcrops, the exposure of structural surfaces in outcrops can be divided into two types: planar and linear. Linear structural surfaces are formed by factors such as tectonic activity and unloading, and appear as linear surfaces. Linear structural surfaces often dominate outcrops and have a significant impact on rock mechanical properties, slope stability, and the exploration and mining of mineral resources. Therefore, accurately identifying the spatial distribution and characteristics of linear structural surfaces is of great practical significance to geological engineering, rock engineering, mining engineering, and other fields.

[0003] However, traditional linear structural surface identification relies heavily on manual observation and measurement, which usually requires extensive on-site investigations and data recording. This method is not only time-consuming and labor-intensive, but also susceptible to human error. Especially when dealing with complex rock outcrops, it is difficult to accurately extract all potential structural surface information. As the survey area expands, the limitations of manual methods become increasingly apparent, especially in three-dimensional space, where the analysis of the occurrence, distribution, and connectivity of structural surfaces becomes increasingly complex, and the structured extraction of structural surface trace data is difficult to achieve. In the existing technology, although the structural surface identification method based on image processing can partially replace manual work, it is limited by the ability to analyze two-dimensional data and is difficult to associate with three-dimensional spatial attributes; and although point cloud data processing technology can construct three-dimensional models, it lacks a targeted segmentation and classification mechanism for linear structural surfaces. Summary of the Invention

[0004] In order to solve the problem of difficulty in identifying linear structural surfaces in complex rock masses, the present invention provides a method and system for automatically identifying linear structural surfaces in complex rock mass outcrops.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for automatically identifying linear structural surfaces in complex rock outcrops comprises the following steps:

[0007] Acquire a rock outcrop image and three-dimensional point cloud data of the rock outcrop image; perform semantic segmentation on the rock outcrop image using a DeepLabV3+ model, and output a binary mask image containing structural surface trace pixels;

[0008] Establishing a mapping relationship between the image coordinate system and the three-dimensional point cloud, and converting the binary mask image into three-dimensional trace point cloud data;

[0009] Extracting trace segment direction vectors based on the three-dimensional trace point cloud data, calculating the spatial angle between the trace segment direction vectors and the average occurrence of the structural surface group, and classifying the trace segments according to a preset angle threshold; and performing three-dimensional line segment extraction on the classified traces in each group;

[0010] Based on the greedy algorithm, the extracted three-dimensional line segments are globally connected by polylines to obtain the polyline sets of each structural surface group. The linear structural surfaces of the rock outcrop are identified based on the polyline sets of the structural surface groups, and the three-dimensional trace length distribution of the linear structural surfaces is calculated.

[0011] Preferably, the binary mask image is converted into three-dimensional trace point cloud data, specifically comprising the following steps:

[0012] Fill the pixel edges of the binary mask image;

[0013] Use a 3×3 sliding window to traverse the padded binary mask image. When the center of the sliding window is a trace pixel, trace the path connected to the center.

[0014] Counting the number of connectable paths in each trace pixel, dynamically marking checkpoints based on the number of connectable paths, segmenting intersections and storing line segments;

[0015] Reset the processed pixels to background labels, iterate all traces in the binary mask image, and complete the topological reconstruction of the traces;

[0016] The binary mask image after topological reconstruction is converted into three-dimensional trace point cloud data.

[0017] Preferably, the mapping relationship between the image coordinate system and the three-dimensional point cloud is specifically:

[0018]

[0019] Among them, the three-dimensional point cloud coordinates are P(P X ,P Y ,P Z ), the image coordinate point is p IMG (p x ,p y ), K is the internal parameter matrix of the camera, f x and f y is the focal length in each direction, c x and c y is the focal coordinate where the optical axis and the image plane intersect, [R|t] is the combination of the rotation matrix and the translation matrix, r is the rotation matrix vector, and t is the translation vector.

[0020] Preferably, the three-dimensional line segment extraction is performed on the classified traces in each group, specifically using an improved iterative Hough transform algorithm to perform the three-dimensional line segment extraction, including the following steps:

[0021] Define a three-dimensional line with the minimum parameter in Hough space, the direction vector of the three-dimensional line The parameterization of is:

[0022]

[0023] in, is the direction vector; the three-dimensional space vector is divided into uniform intervals using the subdivision method of Platonic solids. The Platonic solid with the largest number of vertices is the icosahedron. For each interval of the icosahedron, a three-dimensional straight line direction vector is uniformly discretized;

[0024] The discretized 3D line direction vector is combined with the point cloud density to dynamically adjust the Hough space resolution, prioritize the extraction of long line segments and remove noise;

[0025] Set the distance threshold d thres_seg Merge collinear segments to obtain 3D line segments.

[0026] Preferably, the method of performing global polyline connection on the extracted three-dimensional line segments based on a greedy algorithm to obtain a polyline set of each structural surface group, identifying the linear structural surface of the rock outcrop based on the polyline set of the structural surface group, and calculating the three-dimensional trace length distribution of the linear structural surface comprises the following steps:

[0027] A greedy strategy is used to search the space between the endpoints of the three-dimensional line segment in the KD tree to find the global nearest neighbor;

[0028] If the endpoint spacing is less than the endpoint spacing threshold of polyline aggregation, they are merged into polylines, retaining the longest path to obtain the polyline set of each structural face group;

[0029] The linear structural surfaces of the rock outcrop are identified based on the polyline sets of each structural surface group, and the three-dimensional trace length distribution of the linear structural surfaces is calculated.

[0030] Preferably, the trace segment direction vector is extracted through principal component analysis and covariance matrix.

[0031] The present invention also provides a system for automatically identifying linear structural surfaces in complex rock outcrops, which specifically includes:

[0032] The trace detection module is used to obtain rock outcrop images and three-dimensional point cloud data of the rock outcrop images; use the DeepLabV3+ model to perform semantic segmentation on the rock outcrop images, and output a binary mask image containing structural surface trace pixels.

[0033] The spatial processing module is used to establish a mapping relationship between the image coordinate system and the three-dimensional point cloud, and convert the binary mask image into three-dimensional trace point cloud data.

[0034] A line segment acquisition module is used to extract the direction vector of the trace segment based on the three-dimensional trace point cloud data, calculate the spatial angle between the trace segment direction vector and the average occurrence of the structural surface group, classify the trace segments according to a preset angle threshold, and perform three-dimensional line segment extraction on the classified traces in each group.

[0035] The linear structural surface recognition module is used to perform global polyline connection on the extracted three-dimensional line segments based on a greedy algorithm to obtain a polyline set of each structural surface group, identify the linear structural surface of the rock outcrop based on the polyline set of the structural surface group, and calculate the three-dimensional trace length distribution of the linear structural surface.

[0036] The present invention also provides a computer device, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps described in the method for automatically identifying linear structural surfaces in complex rock outcrops.

[0037] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is loaded by a processor, the computer program can execute the steps described in the method for automatically identifying linear structural surfaces in complex rock outcrops.

[0038] The method for automatically identifying linear structural surfaces in complex rock outcrops provided by the present invention has the following beneficial effects:

[0039] The present invention obtains the rock outcrop image input through the DeepLabV3+ model to obtain a binary mask image containing the structural surface trace pixels, enhances the sensitivity of trace recognition in complex backgrounds, and improves the accuracy of subsequent calculations and processing. Based on the binary mask image, the topology of the trace is reconstructed and the mapping relationship between the image coordinate system and the three-dimensional point cloud is established, achieving high-precision conversion of the trace from the image to the point cloud and associating the three-dimensional spatial attributes of the structural surface. Based on the greedy algorithm, three-dimensional line segments are extracted and global multi-line segment connection and trace length calculation are performed to structure the structural surface trace data, while reducing the computational time of three-dimensional line segment segmentation and connection, achieving rapid and accurate identification of linear structural surfaces in complex rock outcrops, breaking through the limitations of the terrain area when manually measuring and obtaining rock structural surface data. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the embodiments of the present invention and its design, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0041] Figure 1 The figure is a flow chart of a method for automatically identifying linear structural surfaces in complex rock outcrops according to an embodiment of the present invention.

[0042] Figure 2 A comparison chart of the segmentation and refinement effects of the DeepLabV3+ model in an embodiment of the present invention

[0043] Figure 3 This is the three-dimensional structuring result of the two-dimensional trace pixels in the embodiment of the present invention.

[0044] Figure 4 Schematic diagram of three-dimensional structured space representation in an embodiment of the present invention.

[0045] Figure 5 This is the point cloud grouping result in an embodiment of the present invention.

[0046] Figure 6 This is the three-dimensional line segmentation effect of the iterative Hough transform in an embodiment of the present invention.

[0047] Figure 7 This is the final structural surface trace distribution after the multi-segment lines are connected in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to enable those skilled in the art to better understand the technical solution of the present invention and to be able to implement it, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention.

[0049] Example

[0050] The present invention provides a method for automatically identifying linear structural surfaces in complex rock outcrops. Figure 1 As shown, the specific steps include:

[0051] Step 1: Fusion input of multi-source heterogeneous data.

[0052] (1) Obtain rock outcrop images.

[0053] (2) The ContextCapture software is used to generate three-dimensional spatial trace reconstruction of the rock outcrop image to obtain three-dimensional point cloud data of the rock outcrop image. The trace is the linear crack trace.

[0054] (3) Set dynamic threshold groups, including:

[0055] Angle threshold (θ thres_set ):The structural face group to which the line segment belongs (recommended value 20°); distance threshold (d thres_seg):neighborhood judgment threshold for 3D segmentation (recommended value 0.05m); connection threshold (d thres_poly ): The endpoint spacing threshold for polyline aggregation (recommended value 0.1m).

[0056] Step 2: Pixel-level detection of two-dimensional traces based on deep learning.

[0057] The pre-trained DeepLabV3+ model is used to perform semantic segmentation on the input rock outcrop image, such as Figure 2 As shown in the figure, the blue dots indicate the image locations for each semantic segmentation. Large-scale images are processed through adaptive cropping (400×400 pixels) and splicing strategies, outputting binary masks of structural surface trace pixels. The model incorporates a channel attention mechanism to enhance sensitivity to low-contrast traces and uses transfer learning to optimize generalization under complex lighting conditions.

[0058] Step 3: Pixel-level trace refinement and topology reconstruction.

[0059] Perform morphological refinement on the 2D mask image to eliminate redundant trace thickness. Perform topological reconstruction on the 2D line segments based on the improved 8-neighborhood tracking algorithm. This topological reconstruction clarifies the trunk and branches, that is, the primary and secondary connection relationships between pixels, and defines their topological properties. This includes the following steps:

[0060] (1) Boundary filling: fill the image with 1 pixel edge to avoid loss of boundary pixels;

[0061] (2) Path search: traverse the image with a 3×3 sliding window and perform path tracking when the center is the trace pixel;

[0062] (3) Branch determination: Dynamically mark checkpoints based on the number of connected paths in the neighborhood to achieve intersection segmentation and independent storage of line segments.

[0063] The number of connectable paths is 0: If the number of pixels connected to any path so far is equal to 1, the current point pixel will be changed to the background label and proceed to the next step;

[0064] The number of connectable paths is 1: If the number of connected pixels so far exceeds 5, and the angle between the path to the current point and the next path is greater than 30°, the current point is designated as a checkpoint and added to the checkpoint list, and the next step is performed. Otherwise, the pixel of the next path is designated as the current point, and the process returns to step (3) to re-check the pixels;

[0065] Number of connectable paths ≥ 2: The current point is designated as a checkpoint and added to the checkpoint list. If the number of pixels connected so far is equal to 1, the next pixel in any path will be assigned to the current point and the process will return to step (3); otherwise, continue to the next step.

[0066] (4) Background reset: After completing the line segment extraction, the processed pixels are reset to the background label, and the process is iterated until the full image trace in the binary mask image is traversed.

[0067] Step 4: 2D-3D space mapping and coordinate inversion.

[0068] Based on the camera parameters and distortion correction models associated with different camera models, the mapping relationship between the image coordinate system and the 3D point cloud is established:

[0069] (1) Projection equation: Define point p IMG (p x ,p y ), that is, from the point P(P X ,P Y ,P Z ) is projected into a point in the image coordinate system, such as Figure 3 As shown, the intrinsic and extrinsic parameters of the camera are expressed as:

[0070]

[0071] Among them, f x and f y is the focal length in each direction, c x and c y is the focal coordinate where the optical axis and the image plane intersect. [R|t] is the combination of a rotation matrix and a translation matrix, which converts 3D coordinates into camera coordinates. The elements of this matrix consist of the camera's extrinsic parameters, including focal length, distortion coefficients, rotation matrix, and internal and external orientation parameters.

[0072] (2) Inverse mapping analysis: combining radial distortion (k1, k2, k3) and tangential distortion (p 1, Using the p2) parameter, a nonlinear system of equations is constructed to invert the 3D spatial coordinates corresponding to the 2D pixels. Before multiplying by the camera's intrinsic parameter matrix K, the normalized coordinates reflecting the distortion are calculated using the following formula.

[0073]

[0074] in, The subscripts d and u denote the distorted and undistorted states, respectively.

[0075] Therefore, point P in the three-dimensional coordinate system can be projected to point p in the image coordinate system IMG,Applying this step in reverse, the 2D pixel coordinates can be converted into 3D trace point cloud data, such as Figure 4 shown.

[0076] Step 5: Point cloud classification driven by structural face group, such as Figure 5 shown.

[0077] Based on principal component analysis (PCA) and covariance matrix, the line segment direction vector is extracted, and the spatial angle between the line segment direction vector and the average occurrence of the structural surface group is calculated. thres_set ) to filter the belonging group. For intersecting traces, a multi-label assignment strategy is adopted to ensure that the same line segment can be associated with multiple intersecting structure face groups.

[0078] Step 6: Iterate the 3D line segmentation of Hough transform, such as Figure 6 shown.

[0079] After classifying the point cloud into structural surface groups, an improved iterative Hough transform algorithm is used to extract 3D line segments within each group to improve the efficiency and accuracy of Hough transform recognition. The following steps are performed:

[0080] (1) Parameter space discretization: The 3D line direction vector is uniformly discretized based on the icosahedron subdivision method to reduce computational complexity. A 3D line with the minimum parameter is defined in Hough space. The definition of the 3D line is as follows:

[0081]

[0082] Among them, the direction vector It is simplified to one variable through vector space discretization method. Defined by the straight line x / and y / , the direction vector is considered as a normal vector passing through the origin of the coordinate system, specifically:

[0083]

[0084] Among them, a x 、a y 、a z Respectively Coordinates on the x, y, and z axes; b x 、b y 、b z Respectively The coordinates on the x, y, and z axes.

[0085] Direction vector Defined by discretizing vector space into uniform intervals. The space of vectors is divided into uniform intervals using the subdivision method of the Platonic solids. The vector space is discretized uniformly by progressively partitioning the icosahedron.

[0086] (2) Adaptive voting mechanism: Dynamically adjust the Hough space resolution based on the point cloud density, prioritize the extraction of long line segments and remove noise.

[0087] (3) Segment aggregation optimization: by distance threshold (d thres_seg ) Merge collinear segments to improve segmentation continuity.

[0088] Step 7: Implement global polyline connection and trace length calculation of discrete line segments based on the greedy algorithm, specifically including the following steps:

[0089] (1) Neighbor search: KD tree is used to accelerate the spatial neighbor search between line segment endpoints;

[0090] (2) Topological connection: If the distance between endpoints is less than the connection threshold d thres_poly , they are merged into polylines, with the longest path being the first to be retained.

[0091] (3) Trace length statistics: Output the polyline set of each structural surface group, and identify the linear structural surface of the rock outcrop based on the polyline set of each structural surface group, such as Figure 7 As shown in the figure, different colors correspond to different structural surface groups, and the three-dimensional trace length distribution of the linear structural surface is calculated.

[0092] The present invention also provides a system for automatically identifying linear structural surfaces in complex rock outcrops, which specifically includes:

[0093] The trace detection module is used to obtain rock outcrop images and three-dimensional point cloud data of rock outcrop images; the DeepLabV3+ model is used to perform semantic segmentation on the rock outcrop images and output a binary mask image containing structural surface trace pixels.

[0094] The spatial processing module is used to establish the mapping relationship between the image coordinate system and the three-dimensional point cloud, and convert the binary mask image into three-dimensional trace point cloud data.

[0095] The line segment acquisition module is used to extract the direction vector of the trace segment based on the three-dimensional trace point cloud data, calculate the spatial angle between the trace segment direction vector and the average occurrence of the structural surface group, classify the trace segments according to the preset angle threshold; and perform three-dimensional line segment extraction on the classified traces in each group.

[0096] The linear structural surface recognition module is used to perform global polyline connection on the extracted three-dimensional line segments based on a greedy algorithm to obtain the polyline set of each structural surface group. The linear structural surface of the rock outcrop is identified based on the polyline set of the structural surface group, and the three-dimensional trace length distribution of the linear structural surface is calculated.

[0097] Each module in the aforementioned system for automatically identifying linear structural surfaces in complex rock outcrops can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0098] The present invention also provides a computer device comprising a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of an embodiment of a method for automatically identifying linear structural surfaces in complex rock outcrops. The specific implementation method can be found in the method embodiments and will not be repeated here.

[0099] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, wherein a computer program is stored on the storage medium. For example, a memory containing instructions, wherein the instructions can be executed by a processor of a computer device to complete the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device. When the computer program is executed by the processor, the steps in an embodiment of a method for automatically identifying linear structural surfaces in complex rock outcrops can be implemented. The specific implementation method can be found in the method embodiment and will not be repeated here.

[0100] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0101] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0102] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0104] It should be pointed out that the specific implementation methods described above can enable those skilled in the art to understand the invention more comprehensively, but do not limit the invention in any way. Therefore, although the present specification and examples have described the invention in detail, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the scope of protection of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved. Any simple change or equivalent replacement of the technical solution that can be obviously obtained by any person familiar with the art within the technical scope disclosed in the present invention falls within the scope of protection of the present invention.

Claims

1. A method for automatically identifying linear structural surfaces in complex rock outcrops, characterized in that: The following steps are involved: Acquire a rock outcrop image and three-dimensional point cloud data of the rock outcrop image; perform semantic segmentation on the rock outcrop image using a DeepLabV3+ model, and output a binary mask image containing structural surface trace pixels; Establishing a mapping relationship between the image coordinate system and the three-dimensional point cloud, and converting the binary mask image into three-dimensional trace point cloud data; Extracting trace segment direction vectors based on the three-dimensional trace point cloud data, calculating the spatial angle between the trace segment direction vectors and the average occurrence of the structural surface group, and classifying the trace segments according to a preset angle threshold; and performing three-dimensional line segment extraction on the classified traces in each group; Based on the greedy algorithm, the extracted three-dimensional line segments are globally connected by polylines to obtain the polyline sets of each structural surface group. The linear structural surfaces of the rock outcrop are identified based on the polyline sets of the structural surface groups, and the three-dimensional trace length distribution of the linear structural surfaces is calculated.

2. The method for automatically identifying linear structural surfaces in complex rock outcrops according to claim 1, characterized in that: The binary mask image is converted into three-dimensional trace point cloud data, specifically comprising the following steps: Fill the pixel edges of the binary mask image; Use a 3×3 sliding window to traverse the padded binary mask image. When the center of the sliding window is a trace pixel, trace the path connected to the center. Counting the number of connectable paths in each trace pixel, dynamically marking checkpoints based on the number of connectable paths, segmenting intersections and storing line segments; Reset the processed pixels to background labels, iterate all traces in the binary mask image, and complete the topological reconstruction of the traces; The binary mask image after topological reconstruction is converted into three-dimensional trace point cloud data.

3. The method for automatically identifying linear structural surfaces in complex rock outcrops according to claim 1, characterized in that: The mapping relationship between the image coordinate system and the three-dimensional point cloud is specifically: Among them, the three-dimensional point cloud coordinates are P(P X ,P Y ,P Z ), the image coordinate point is p IMG (p x ,p y ), K is the internal parameter matrix of the camera, f x and f y is the focal length in each direction, c x and c y is the focal coordinate where the optical axis and the image plane intersect, [R|t] is the combination of the rotation matrix and the translation matrix, r is the rotation matrix vector, and t is the translation vector.

4. The method for automatically identifying linear structural surfaces in complex rock outcrops according to claim 1, characterized in that: The three-dimensional line segment extraction is performed on the classified traces in each group, specifically using an improved iterative Hough transform algorithm to perform the three-dimensional line segment extraction, including the following steps: Define a three-dimensional line with the minimum parameter in Hough space, the direction vector of the three-dimensional line The parameterization of is: in, is the direction vector; the three-dimensional space vector is divided into uniform intervals using the subdivision method of Platonic solids. The Platonic solid with the largest number of vertices is the icosahedron. For each interval of the icosahedron, a three-dimensional straight line direction vector is uniformly discretized; The discretized 3D line direction vector is combined with the point cloud density to dynamically adjust the Hough space resolution, prioritize the extraction of long line segments and remove noise; Set the distance threshold d thres_seg Merge collinear segments to obtain 3D line segments.

5. The method for automatically identifying linear structural surfaces in complex rock outcrops according to claim 1, characterized in that: The method performs global polyline connection on the extracted three-dimensional line segments based on the greedy algorithm to obtain a polyline set of each structural surface group, identifies the linear structural surface of the rock outcrop based on the polyline set of the structural surface group, and calculates the three-dimensional trace length distribution of the linear structural surface, including the following steps: A greedy strategy is used to search the space between the endpoints of the three-dimensional line segment in the KD tree to find the global nearest neighbor; If the endpoint spacing is less than the endpoint spacing threshold of polyline aggregation, they are merged into polylines, retaining the longest path to obtain the polyline set of each structural face group; The linear structural surfaces of the rock outcrop are identified based on the polyline sets of each structural surface group, and the three-dimensional trace length distribution of the linear structural surfaces is calculated.

6. The method for automatically identifying linear structural surfaces in complex rock outcrops according to claim 1, characterized in that: The trajectory segment direction vector is specifically extracted through principal component analysis and covariance matrix.

7. An automatic identification system for linear structural surfaces in complex rock outcrops, characterized by: include: A trace detection module, used to obtain a rock outcrop image and three-dimensional point cloud data of the rock outcrop image; Using the DeepLabV3+ model to perform semantic segmentation on the rock outcrop image, and output a binary mask image containing structural surface trace pixels; A spatial processing module is used to establish a mapping relationship between the image coordinate system and the three-dimensional point cloud, and convert the binary mask image into three-dimensional trace point cloud data; a line segment acquisition module, configured to extract a line segment direction vector based on the three-dimensional line point cloud data, calculate a spatial angle between the line segment direction vector and the average occurrence of the structural surface group, classify the line segments according to a preset angle threshold, and perform three-dimensional line segment extraction on the classified lines within each group; The linear structural surface recognition module is used to perform global polyline connection on the extracted three-dimensional line segments based on a greedy algorithm to obtain a polyline set of each structural surface group, identify the linear structural surface of the rock outcrop based on the polyline set of the structural surface group, and calculate the three-dimensional trace length distribution of the linear structural surface.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is loaded into a processor, it can execute the steps of the method according to any one of claims 1 to 6.

Images