Open-pit mine three-dimensional modeling calibration method based on unmanned aerial vehicle oblique photography

Abstract

The application discloses an open-pit mine three-dimensional modeling calibration method based on unmanned aerial vehicle oblique photography, relates to the mine management field and comprises the following steps: collecting mine basic structure parameters, constructing a mine initial model based on the mine basic structure parameters, and taking the mine initial model as a modeling calibration ontology; selecting a calibration area on the mine initial model, performing oblique photography on the mine entity surface based on the calibration area to collect mine surface images; the initial model is constructed through high-precision ranging, the calibration area images are collected in a targeted manner in combination with the unmanned aerial vehicle oblique photography, the local images are accurately segmented to focus on key areas, the window is flexibly adapted to the calibration accuracy requirement, the image gray scale features are converted into quantified convex and concave values through shape recognition, the initial model surface is accurately stretched and adjusted, and the detail restoration degree of the mine three-dimensional model is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of mine management technology, specifically to a method for 3D modeling and calibration of open-pit mines based on UAV oblique photography. Background Technology

[0002] Mine safety management is the core of ensuring safe production in mines. Through system construction, risk assessment, equipment testing, personnel training, and emergency drills, it prevents accidents such as collapses and gas explosions, standardizes operating procedures, implements responsibilities, protects the lives of miners and the safety of mine property, and ensures compliant and orderly production.

[0003] Patent application No. 202311142209.7 discloses a method for evaluating the effectiveness of open-pit mine ecological restoration. The method includes: determining the ecological restoration area of ​​the mine; acquiring measurement data within the ecological restoration area using UAV oblique photography; constructing a three-dimensional solid model and a mine surface model based on preset UAV parameter data and the measurement data; extracting ecological restoration evaluation parameters based on the three-dimensional solid model and the mine surface model; obtaining quantity growth indicators and quality improvement indicators based on the ecological restoration evaluation parameters; calculating the total evaluation value of the open-pit mine ecological restoration effectiveness using the quantity growth indicators and the quality improvement indicators; and determining the open-pit mine ecological restoration level. This application aims to address the problems of "unscientific evaluation system, unreasonable indicator settings, and imprecise data collection methods in the evaluation of open-pit mine ecological restoration effectiveness: the methods for obtaining open-pit mine ecological restoration data are mostly manual field exploration and satellite remote sensing, both of which suffer from low efficiency, high cost, and poor data real-time performance. Various provinces and cities in my country have implemented and completed a large amount of open-pit mine ecological restoration work. Currently, the effectiveness of open-pit mine ecological restoration management suffers from unclear baseline data, unclear situation, and weak constraints."

[0004] However, both mine restoration work and daily mine safety management require accurate 3D modeling of the mine. While existing technologies offer relatively reliable modeling accuracy, the mine structure may change due to human or natural factors. Therefore, existing technologies can only obtain relatively reliable mine models by remodeling, which is a cumbersome process and not suitable for application scenarios that require frequent and high-precision updates to the mine model.

[0005] To address this, we propose a method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a method for three-dimensional modeling and calibration of open-pit mines based on UAV oblique photography, which can effectively solve the problems of the existing technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0008] This invention discloses a method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry, including:

[0009] The process involves collecting basic structural parameters of the mine, constructing an initial mine model based on these parameters, and using this initial model as the modeling calibration ontology. A calibration area is selected on the initial mine model, and oblique photography is performed on the mine surface based on this area to acquire surface images. These acquired surface images are then segmented to obtain local images corresponding to the calibration areas. These local images are received, and a verification window is set. The verification window is then slid across the local images to perform morphological recognition on each sub-local image. The morphological recognition results for each sub-local image are obtained and converted into convexity and concavity values. Based on these values, the corresponding point on the surface of the initial mine model is used as the stretching point, and convexity stretching or concavity stretching is performed along the acquisition direction of the local images. The stretched initial mine model is then recorded as the calibrated 3D mine model.

[0010] Furthermore, the basic structural parameters of the mine are collected using a ranging device, with a collection accuracy of no less than one collection operation per square meter from the mine's top-down view.

[0011] The ranging device is equipped with a drone, which flies over the area occupied by the mine and performs ranging operations from above. The ranging angle is downward from the plane above the mine. The ranging frequency is one data collection operation for each square meter of the area above the mine. Each data collection position is the center point of the one square meter area defined for each data collection operation.

[0012] In 3D space, draw the bounded plane corresponding to the area of ​​the mine as seen from above. Determine that the coordinates of any edge of the bounded plane are consistent with the coordinates of the actual boundary of the area of ​​the mine as seen from above. Mark the distance measurement points on the surface of the bounded plane according to the relative positions measured by the distance measurement device. Draw line segments vertically upward at each distance measurement point according to the corresponding distance measurement results. Then connect the endpoints of each drawn line segment that are far away from the bounded plane, and extend the endpoints of the edge positions to the nearest endpoint of the edge of the bounded plane to complete the construction of the initial model of the mine.

[0013] Furthermore, when selecting calibration areas on the initial mine model, options include: manual selection and automatic selection.

[0014] Manual selection: Select no fewer than three 3D coordinates on the surface of the initial mine model, and connect the selected 3D coordinates adjacent to each other to form a closed area; fly a drone equipped with a camera module to the two ends of the infinitely extending vertical line where the center point of the closed area is located, and perform the acquisition of mine surface images with the image acquisition angle perpendicular to the surface of the closed area.

[0015] Automatic selection: The image acquisition area of ​​the mine surface is each sub-face that makes up the top surface of the initial mine model. The operation of acquiring the mine surface image is the same as manual selection.

[0016] Each time the camera module acquires an image of the mine surface, it detects the straight-line distance between the ranging end and the mine surface, and controls the drone to move on a vertical line that extends infinitely at both ends of the center point of the closed area. After moving to a preset distance, the acquisition of the mine surface image is performed, so that the distance between the ranging end of the camera module and the mine surface is equal each time an image of the mine surface is acquired, and the area corresponding to the acquired mine surface image at least includes the closed area.

[0017] Furthermore, when segmenting the mine surface image, the three-dimensional coordinates of the boundaries of the corresponding mine surface regions in any two mine surface images are obtained simultaneously. The image scale is determined based on the straight-line distance between the two boundary coordinates. According to the scale and the three-dimensional coordinates used to determine the closed region, the corresponding region of the closed region is determined in the mine surface image. The segmentation operation is then performed, with the region corresponding to the closed region in the mine surface image as the retention target. Finally, the segmentation operation of the mine surface image is completed, and the segmented result is the local image corresponding to the calibration region in the mine surface image.

[0018] Furthermore, the size of the inspection window is customized by the user. When setting the inspection window, the aspect ratio of the inspection window is equal to the aspect ratio of the maximum inscribed rectangle of the local image. The higher the calibration accuracy requirement of the initial mine model, the smaller the inspection window, and vice versa.

[0019] The length and width of the inspection window and the length and width of the maximum inscribed rectangle of the local image are both in pixels.

[0020] The morphological recognition content of the sub-local image under each window is: protrusion, depression, degree of protrusion, and degree of depression.

[0021] Furthermore, the morphological recognition logic of the sub-local image is as follows:

[0022]

[0023] In the formula: S(x,y,k) represents the degree of protrusion and depression in the sub-local image; k is the neighborhood extension scale; G h(i) represents the gray value of the i-th pixel in the horizontal neighborhood of the target pixel; G(x,y) represents the gray value of the pixel with coordinates (x,y) in the image; G v (j) represents the gray value of the j-th pixel in the vertical neighborhood of the target pixel; α is the modulation index; H is the gray value distribution entropy of the neighborhood; ε is a very small positive number;

[0024] Where G(x,y) and G h (i), G v (j) all point to the center pixel of the sub-local image within the test window. A positive value of S(x,y,k) indicates a depression, and a negative value indicates a convexity. The larger the absolute value, the more significant the depression or convexity.

[0025] Furthermore, the modulation index α is calculated using the following formula:

[0026]

[0027] The degree of convexity and concavity S(x,y,k) based on the sub-local image representation is converted into convexity and concavity values:

[0028]

[0029] Where: H up (x,y),H down (x,y) represents the convexity and concaveness values; K is the calibration coefficient; L is the vertical distance from the camera to the object surface; p is the physical size of the image sensor pixel; d is the focal length of the camera lens; θ is the angle between the image acquisition angle and the normal to the object surface.

[0030] Among them, θ is obtained through the attitude sensor of the camera module, and L is the preset distance used by the UAV to move on the vertical line that extends infinitely at both ends of the center point of the closed area. In the preparation stage of this method, the calibration coefficient K is obtained by standard height sample calibration to reflect the mapping relationship between gray scale difference and physical height.

[0031] Furthermore, when the surface of the initial mine model is stretched based on the stretching point, the stretching range of the stretching point is the corresponding area of ​​the local image of the calibration area in the mine surface image on the surface of the initial mine model.

[0032] Furthermore, the calibrated 3D mine model is continuously updated based on a preset cycle. When updating the 3D mine model, the 3D mine model obtained from the previous stretching process is used as the initial mine model.

[0033] In the process of continuously updating the 3D model of the mine based on a preset cycle, the 3D model of the mine will be updated once when rainfall exceeding the preset rainfall threshold occurs in the area where the mine is located, crosswinds exceeding the preset wind level occur, snow melts, or the mine structure changes due to the resumption of mine work.

[0034] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0035] This invention provides a method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry. During execution, the method constructs an initial model through high-precision ranging, combines this with targeted acquisition of calibration area images via UAV oblique photogrammetry, accurately segments local images to focus on key areas, and flexibly adapts the verification window to calibration accuracy requirements. Through morphological recognition, image grayscale features are converted into quantified convex and concave values, enabling precise stretching and adjustment of the initial model surface, effectively improving the detail reproduction of the 3D mine model. Simultaneously, the calibration area supports both manual and automatic selection, and consistent acquisition distance settings ensure image scale stability, enhancing the accuracy of mapping grayscale differences to physical height. Furthermore, a periodic model iteration mechanism triggered by environmental changes and mine structure updates dynamically captures the real-time mine status, ensuring the 3D model continuously conforms to actual working conditions and providing more reliable spatial data support for mine management and safety monitoring. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0037] Figure 1 This is a flowchart illustrating a method for 3D modeling and calibration of open-pit mines based on UAV oblique photography.

[0038] Figure 2 This is a schematic diagram illustrating the construction process of the initial mine model in this invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0040] The present invention will be further described below with reference to embodiments.

[0041] Example:

[0042] This embodiment presents a method for 3D modeling and calibration of open-pit mines based on UAV oblique photography, such as... Figure 1 As shown, it includes:

[0043] Collect basic structural parameters of the mine, construct an initial model of the mine based on the basic structural parameters, and use the initial model of the mine as the modeling calibration ontology;

[0044] The basic structural parameters of the mine are collected using a ranging device, with a collection accuracy of no less than one collection operation per square meter from the mine's top-down view.

[0045] The ranging device is equipped with a drone, which flies over the mining area and performs ranging operations from above. The ranging angle is downward from the plane above the mine. The ranging frequency is one data collection operation for each square meter of the area above the mine. Each data collection position is the center point of the one square meter area defined for each data collection operation.

[0046] In 3D space, draw the bounded plane corresponding to the area of ​​the mine as seen from above. Determine that the coordinates of any edge of the bounded plane are consistent with the coordinates of the actual boundary of the area of ​​the mine as seen from above. Mark the distance measurement points on the surface of the bounded plane according to the relative positions of the distance measurement devices. Draw line segments vertically upward at each distance measurement point according to the corresponding distance measurement results. Then connect the endpoints of each drawn line segment that are far away from the bounded plane to each other, and extend the endpoints of the edge positions to the nearest endpoint of the edge of the bounded plane to complete the construction of the initial model of the mine.

[0047] A calibration area is selected on the initial model of the mine, and oblique photography is performed on the surface of the mine entity based on the calibration area to acquire images of the mine surface;

[0048] Selecting a calibration area on the initial mine model includes: manual selection and automatic selection;

[0049] Manual selection: Select no fewer than three 3D coordinates on the surface of the initial mine model, and connect the selected 3D coordinates adjacent to each other to form a closed area; fly a drone equipped with a camera module to the two ends of the infinitely extending vertical line where the center point of the closed area is located, and perform the acquisition of mine surface images with the image acquisition angle perpendicular to the surface of the closed area.

[0050] Automatic selection: The image acquisition area of ​​the mine surface is each sub-face that makes up the top surface of the initial mine model. The operation of acquiring the mine surface image is the same as manual selection.

[0051] Each time the camera module acquires an image of the mine surface, it detects the straight-line distance between the ranging end and the mine surface, and controls the drone to move on the vertical lines that extend infinitely at both ends of the center point of the closed area. After moving to the preset distance, the acquisition of the mine surface image is performed, so that the distance between the ranging end of the camera module and the mine surface is equal each time an image of the mine surface is acquired, and the area corresponding to the acquired mine surface image at least includes the closed area.

[0052] When segmenting a mine surface image, the three-dimensional coordinates of the boundaries of the corresponding mine surface regions in any two mine surface images are acquired simultaneously. The image scale is determined based on the straight-line distance between the two boundary coordinates. According to the scale and the three-dimensional coordinates used to determine the closed region, the corresponding region of the closed region is determined in the mine surface image. The segmentation operation is then performed, with the region corresponding to the closed region in the mine surface image as the retention target. Finally, the segmentation operation of the mine surface image is completed, and the segmented result is the local image corresponding to the calibration region in the mine surface image.

[0053] Acquire the collected images of the mine surface, segment the mine surface images, and obtain the local images corresponding to the calibration areas in the mine surface images;

[0054] Receive local images, set inspection windows, slide the inspection windows in the local images, and perform morphological recognition on the sub-local images under each window;

[0055] The size of the inspection window is customized by the user. When setting the inspection window, the aspect ratio of the inspection window is equal to the aspect ratio of the largest inscribed rectangle of the local image. The higher the calibration accuracy requirement of the initial mine model, the smaller the inspection window, and vice versa.

[0056] The dimensions of the inspection window and the maximum inscribed rectangle of the local image are both in pixels.

[0057] The morphological recognition content of the sub-local image under each window is: convexity, concavity, degree of convexity, and degree of concavity;

[0058] The morphological recognition logic for sub-local images is as follows:

[0059]

[0060] In the formula: S(x,y,k) represents the degree of protrusion and depression in the sub-local image; k is the neighborhood extension scale; G h (i) represents the gray value of the i-th pixel in the horizontal neighborhood of the target pixel; G(x,y) represents the gray value of the pixel with coordinates (x,y) in the image; G v (j) represents the gray value of the j-th pixel in the vertical neighborhood of the target pixel; α is the modulation index; H is the gray value distribution entropy of the neighborhood; ε is a very small positive number;

[0061] Where G(x,y) and G h (i), G v (j) all point to the center pixel of the sub-local image within the test window. A positive value of S(x,y,k) indicates a depression, and a negative value indicates a convexity. The larger the absolute value, the more significant the depression or convexity.

[0062] The above formula integrates the gray value distribution of the horizontal and vertical neighborhoods of the target pixel, and introduces the modulation index and neighborhood gray value distribution entropy to quantify the local morphological features of the image. The gray value contrast of the horizontal and vertical neighborhoods can capture the surface undulation trend in different directions. The modulation index can adjust the weight of gray value difference, and the neighborhood entropy reflects the degree of disorder of gray value distribution. The combination of the three enables the calculation results to accurately distinguish between convex and concave, and the absolute value reflects the degree of morphological significance. It effectively transforms the gray value changes of the image into quantifiable surface morphological differences, and improves the recognition accuracy of subtle undulations on the mine surface.

[0063] The formula for calculating the modulation index α is:

[0064]

[0065] The degree of convexity and concavity S(x,y,k) based on the sub-local image representation is converted into convexity and concavity values:

[0066]

[0067] Where: H up (x,y),H down (x,y) represents the convexity and concaveness values; K is the calibration coefficient; L is the vertical distance from the camera to the object surface; p is the physical size of the image sensor pixel; d is the focal length of the camera lens; θ is the angle between the image acquisition angle and the normal to the object surface.

[0068] Among them, θ is obtained through the attitude sensor built into the camera module, and L is the preset distance used by the UAV to move on the vertical line that extends infinitely at both ends of the center point of the closed area. In the preparation stage of this method, the calibration coefficient K is obtained by standard height sample calibration to reflect the mapping relationship between gray scale difference and physical height.

[0069] The above formula is based on quantified morphological differences in the image, incorporating hardware parameters such as the vertical distance from the camera to the mine surface, the physical size of the image sensor pixels, the lens focal length, and the angle between the acquisition angle and the surface normal. Then, through calibration coefficients pre-calibrated with standard height samples, a mapping relationship between grayscale differences and physical height is established. The introduction of these parameters allows visual features in the image to be directly converted into actual height changes in three-dimensional space. The calibration coefficients solve the problem of inconsistent mapping between grayscale and physical quantities under different devices and environments, achieving a precise conversion from image information to the actual height of protrusions and depressions on the mine surface, providing a reliable physical scale basis for the stretching calibration of the initial model.

[0070] Obtain the shape recognition results of the sub-local images under each window, convert the shape recognition results into convex and concave values, and use the corresponding point on the surface of the initial mine model with the center pixel position of the sub-local image under each window in the local image as the stretching point, and perform convex stretching or concave stretching along the acquisition direction of the local image.

[0071] When the surface of the initial mine model is stretched based on the stretching point, the stretching range of the stretching point is the corresponding area of ​​the local image of the calibration area in the mine surface image to which it points on the surface of the initial mine model.

[0072] The initial mine model after stretching is recorded as the calibrated three-dimensional mine model.

[0073] The calibrated 3D mine model is continuously updated based on a preset cycle. When updating the 3D mine model, the 3D mine model obtained from the previous stretching process is used as the initial mine model.

[0074] In the process of continuously updating the 3D model of the mine based on a preset cycle, the 3D model of the mine will be updated once when rainfall exceeding the preset rainfall threshold occurs in the area where the mine is located, crosswinds exceeding the preset wind level occur, snow melts, or the mine structure changes due to the resumption of mine work.

[0075] In this embodiment, the above method improves the accuracy of 3D modeling of open-pit mines through UAV oblique photography and precise calibration, resulting in more realistic details of local protrusions and depressions. Relying on a dynamic update mechanism, it can promptly capture changes in mine structure caused by rainfall, resumption of work, etc., ensuring the timeliness of the model. The accurate 3D model provides reliable data support for mine safety monitoring and production planning, optimizing management efficiency.

[0076] See Figure 2 As shown in the figure, the process of constructing the initial model of the mine is demonstrated based on the arrows in the figure.

[0077] The following example illustrates an application of the method described in the above embodiments:

[0078] To achieve precise management of the mining area, an open-pit coal mine adopted a 3D modeling and calibration method based on UAV oblique photography to construct a 3D model of the mine. The specific implementation process is as follows:

[0079] I. Constructing the Initial Mine Model

[0080] The coal mine uses drones equipped with high-precision ranging devices to collect basic structural parameters of the mine. The drones fly over the mine's area and perform ranging operations from a top-down perspective. The ranging frequency is strictly controlled to collect data once per square meter of the mine's overhead area, with each collection point being the center point of the corresponding one-square-meter area. The data collection accuracy meets the requirement of once per square meter.

[0081] In three-dimensional space, technicians draw a bounded plane that matches the actual boundary coordinates of the mine's overhead area. Based on the distance measurement points marked by the distance measuring equipment, they draw line segments vertically upwards at each distance measurement point according to the distance measurement results. Then, they connect the endpoints of each line segment that are far from the bounded plane to each other, and extend the endpoints at the edge positions to the nearest endpoint at the edge of the bounded plane, thus constructing the initial three-dimensional model of the open-pit coal mine.

[0082] II. Select the calibration area and acquire surface images

[0083] Based on the actual conditions of the mine, a combination of manual and automatic methods was used to select the calibration area. During manual selection, technicians selected three key three-dimensional coordinates on the surface of the initial model, such as the mining bench and the intersection of the transportation road, and connected them to form a closed area. During automatic selection, each sub-face of the top surface of the initial model was set as the acquisition area.

[0084] The drone, equipped with a camera module, flies to the vertical line of the center point of each enclosed area. Using a ranging function, it detects the straight-line distance to the mine surface and controls the drone to move to a preset distance of 20 meters before stopping, ensuring that the distance between the camera module and the mine surface is consistent each time data is collected. The acquisition angle is perpendicular to the surface of the enclosed area, and the acquired image of the mine surface completely encompasses the enclosed area, thus completing the image acquisition.

[0085] III. Image Segmentation for Local Image Acquisition

[0086] After data acquisition, technicians simultaneously obtain the 3D coordinates of the boundaries of the corresponding mining areas from any two mine surface images. The image scale is then calculated based on the straight-line distance between the two boundary coordinates. Combined with the manually selected 3D coordinates of a closed region, the corresponding area of ​​the closed region is precisely located in the mine surface image. This area is then used as the target for segmentation to obtain a local image corresponding to the calibration region.

[0087] IV. Sub-local image morphology recognition

[0088] Based on the accuracy requirements for mine modeling calibration (high-precision modeling is required in this case), the inspection window size is set, with the aspect ratio of the window matching the aspect ratio of the largest inscribed rectangle in the local image, and the window size is relatively small. The inspection window is then slid across the local image, and morphological recognition is performed on the sub-local images under each window.

[0089] During the recognition process, the center pixel of the sub-local image is taken as the target pixel. The gray values ​​of each pixel in its horizontal and vertical neighborhoods are analyzed, and the degree of protrusion or depression is calculated by combining parameters such as modulation index and neighborhood gray-level distribution entropy. If the calculation result is positive, it indicates that there is a depression in the corresponding area of ​​the sub-local image; if it is negative, it indicates that there is a protrusion. The larger the absolute value of the result, the more significant the degree of depression or protrusion. In this recognition, the calculation result of the sub-local image of the edge of a mining step is -1.8 (negative value), indicating that there is a significant protrusion; the result of the sub-local image of a low-lying area is 2.1 (positive value), indicating that there is a significant depression.

[0090] 5. Convert the convexity and concaveness values ​​and stretch the initial model.

[0091] The convexity and concavity values ​​are converted based on the shape recognition results. During conversion, the values ​​are calculated using calibration coefficients (reflecting the mapping relationship between grayscale differences and physical height) obtained beforehand through standard height sample calibration, as well as parameters such as the vertical distance from the camera to the object surface (i.e., the preset 20 meters), the physical size of the image sensor pixels, the camera lens focal length, and the angle between the image acquisition angle and the normal to the object surface (obtained through the camera module's attitude sensor). The calculated convexity value for the convex area is 0.6 meters, and the concavity value for the concave area is 0.7 meters.

[0092] Using the corresponding point of the center pixel of each window's local image on the surface of the initial model as the stretching point, the initial model is stretched along the image acquisition direction: the raised area is stretched outward by 0.6 meters along the acquisition direction, and the concave area is stretched inward by 0.7 meters along the acquisition direction. The stretching range is limited to the corresponding area of ​​the initial model corresponding to the local image, and finally the calibrated 3D model of the mine is obtained.

[0093] VI. Continuous Model Updates

[0094] The coal mine sets a monthly model update cycle, using the previously calibrated model as the new initial model to repeat the above process. After a heavy rainfall (rainfall exceeding a preset threshold), technicians immediately initiate the model update process; when the mine resumes operations and the structure of the mining area changes, updates are also performed promptly to ensure that the 3D model always reflects the latest state of the mine.

[0095] In summary, the method described in the above embodiments constructs an initial model through high-precision ranging, combines UAV oblique photography to selectively acquire calibration area images, accurately segments local images to focus on key areas, and flexibly adapts the verification window to the calibration accuracy requirements. Through morphological recognition, image grayscale features are converted into quantified convex and concave values, enabling precise stretching and adjustment of the initial model surface, effectively improving the detail reproduction of the 3D mine model. Furthermore, the calibration area supports both manual and automatic selection, and the consistent acquisition distance ensures image scale stability, enhancing the accuracy of mapping grayscale differences to physical height. In addition, the periodic model iteration mechanism triggered by environmental changes and mine structure updates dynamically captures the real-time mine status, ensuring the 3D model continuously conforms to actual working conditions and providing more reliable spatial data support for mine management and safety monitoring.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry, characterized in that, include: Collect basic structural parameters of the mine, construct an initial model of the mine based on the basic structural parameters, and use the initial model of the mine as the modeling calibration ontology; A calibration area is selected on the initial model of the mine, and oblique photography is performed on the surface of the mine entity based on the calibration area to acquire images of the mine surface; Acquire the collected images of the mine surface, segment the mine surface images, and obtain the local images corresponding to the calibration areas in the mine surface images; Receive local images, set inspection windows, slide the inspection windows in the local images, and perform morphological recognition on the sub-local images under each window; The morphological recognition logic of the sub-local image is as follows: ； In the formula: The degree of protrusion and depression shown in the sub-local image; Extend the scale to the neighborhood; The gray value of the i-th pixel in the horizontal neighborhood of the target pixel; The gray value of the pixel with coordinates (x, y) in the image; Let be the gray value of the j-th pixel in the vertical neighborhood of the target pixel; The modulation index; The entropy of the neighborhood grayscale distribution; It is a very small positive number; in, , , Both point to the center pixel of the sub-local image within the inspection window. A positive value indicates a depression, and a negative value indicates a convexity. The larger the absolute value, the more pronounced the depression or convexity. Obtain the shape recognition results of the sub-local images under each window, convert the shape recognition results into convex and concave values, and use the corresponding point on the surface of the initial mine model with the center pixel position of the sub-local image under each window in the local image as the stretching point, and perform convex stretching or concave stretching along the acquisition direction of the local image. The initial model of the mine after stretching is denoted as the calibrated three-dimensional model of the mine.

2. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, The basic structural parameters of the mine are collected by a ranging device, and the collection accuracy is no less than one collection operation per square meter from the mine's top-down view. The ranging device is equipped with a drone, which flies over the area occupied by the mine and performs ranging operations from above. The ranging angle is downward from the plane above the mine. The ranging frequency is one data collection operation for each square meter of the area above the mine. Each data collection position is the center point of the one square meter area defined for each data collection operation. In 3D space, draw the bounded plane corresponding to the area of ​​the mine as seen from above. Determine that the coordinates of any edge of the bounded plane are consistent with the coordinates of the actual boundary of the area of ​​the mine as seen from above. Mark the distance measurement points on the surface of the bounded plane according to the relative positions measured by the distance measurement device. Draw line segments vertically upward at each distance measurement point according to the corresponding distance measurement results. Then connect the endpoints of each drawn line segment that are far away from the bounded plane, and extend the endpoints of the edge positions to the nearest endpoint of the edge of the bounded plane to complete the construction of the initial model of the mine.

3. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, Selecting a calibration area on the initial mine model includes: manual selection and automatic selection; Manual selection: Select no fewer than three 3D coordinates on the surface of the initial mine model, and connect the selected 3D coordinates adjacent to each other to form a closed area; fly a drone equipped with a camera module to the two ends of the infinitely extending vertical line where the center point of the closed area is located, and perform the acquisition of mine surface images with the image acquisition angle perpendicular to the surface of the closed area. Automatic selection: The image acquisition area of ​​the mine surface is each sub-face that makes up the top surface of the initial mine model. The operation of acquiring the mine surface image is the same as manual selection. Each time the camera module acquires an image of the mine surface, it detects the straight-line distance between the ranging end and the mine surface, and controls the drone to move on a vertical line that extends infinitely at both ends of the center point of the closed area. After moving to a preset distance, the acquisition of the mine surface image is performed, so that the distance between the ranging end of the camera module and the mine surface is equal each time an image of the mine surface is acquired, and the area corresponding to the acquired mine surface image at least includes the closed area.

4. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 3, characterized in that, When segmenting a mine surface image, the three-dimensional coordinates of the boundaries of the corresponding mine surface regions in any two mine surface images are acquired simultaneously. The image scale is determined based on the straight-line distance between the two boundary coordinates. According to the scale and the three-dimensional coordinates used to determine the closed region, the corresponding region of the closed region is determined in the mine surface image. The segmentation operation is then performed, with the region corresponding to the closed region in the mine surface image as the retention target. Finally, the segmentation operation of the mine surface image is completed, and the segmented result is the local image corresponding to the calibration region in the mine surface image.

5. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, The size of the inspection window is customized by the user. When setting the inspection window, the aspect ratio of the inspection window is equal to the aspect ratio of the maximum inscribed rectangle of the local image. The higher the calibration accuracy requirement of the initial mine model, the smaller the inspection window, and vice versa. The length and width of the inspection window and the length and width of the maximum inscribed rectangle of the local image are both in pixels. The morphological recognition content of the sub-local image under each window is: protrusion, depression, degree of protrusion, and degree of depression.

6. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, The modulation index The calculation formula is: ； Based on the degree of protrusion and depression represented by sub-local images Convert to bump / depression values: ； In the formula: Values ​​for raised and recessed areas; These are calibration coefficients; The vertical distance from the camera to the object's surface; The physical size of the image sensor pixel; The focal length of the camera lens; The angle between the image acquisition viewpoint and the normal to the object surface; in, The image is captured by the camera module's built-in attitude sensor. This refers to the preset distance that the UAV moves along an infinitely extending vertical line between the two ends of the center point of the enclosed area. In the preparation phase of this method, calibration coefficients are obtained through standard altitude sample calibration. This is used to reflect the mapping relationship between grayscale differences and physical height.

7. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, When the surface of the initial mine model is stretched based on the stretching point, the stretching range of the stretching point is the corresponding area of ​​the local image of the calibration area in the mine surface image to which it points on the surface of the initial mine model.

8. The method for 3D modeling and calibration of open-pit mines based on UAV oblique photogrammetry according to claim 1, characterized in that, The calibrated 3D mine model is continuously updated based on a preset cycle. When updating the 3D mine model, the 3D mine model obtained from the previous stretching process is used as the initial mine model. In the process of continuously updating the 3D model of the mine based on a preset cycle, the 3D model of the mine will be updated once when rainfall exceeding the preset rainfall threshold occurs in the area where the mine is located, crosswinds exceeding the preset wind level occur, snow melts, or the mine structure changes due to the resumption of mine work.

Images