Mining area surface subsidence information extraction method based on visible light image of unmanned aerial vehicle

Abstract

The invention discloses a mining area surface subsidence information extraction method based on an unmanned aerial vehicle visible light image, and relates to the technical field of mining area information detection, and the method comprises the steps: data collection and data processing, and the data processing mainly comprises the steps: obtaining a digital orthoimage DOM and a digital surface model DSM based on the unmanned aerial vehicle visible light image; an unmanned aerial vehicle subsidence model capable of reflecting surface subsidence information is generated on the basis; generating an InSAR subsidence model based on the SAR data; and obtaining an improved unmanned aerial vehicle subsidence model based on the unmanned aerial vehicle subsidence model and the InSAR subsidence model in combination with measured data. According to the method provided by the invention, system error correction is carried out on the unmanned aerial vehicle subsidence model and the InSAR subsidence model by using the leveling data, so that the obtained subsidence model has higher precision, and meanwhile, the ground subsidence model is obtained based on the visible light image of the unmanned aerial vehicle, so that the ground subsidence monitoring cost is greatly reduced; and the advantages of high spatial resolution and high flexibility of the unmanned aerial vehicle platform can be fully utilized.

Description

Technical Field

[0001] The present invention belongs to the technical field of mining area information detection, and in particular relates to a method for extracting mining area surface subsidence information based on unmanned aerial vehicle visible light images. Background Art

[0002] UAV remote sensing has the advantages of high resolution, strong timeliness and flexible operation. It has been widely used in periodic and refined monitoring of small-scale areas. Due to the distinctive characteristics of rich bands, multispectral and hyperspectral sensors based on UAV platforms can detect slight differences in the reflective spectral characteristics of different objects, thereby obtaining more refined spectral curves and more accurate object recognition effects. However, compared with traditional RGB cameras, professional airborne multispectral and hyperspectral sensors are often more expensive, and there are prominent problems such as large data volume, high redundancy, strong correlation between different bands, and complex processing.

[0003] In the field of mining area subsidence monitoring, the commonly used technical means currently include traditional leveling observation and time-series remote sensing technology, such as obtaining surface cumulative subsidence based on time-series InSAR, and obtaining multi-period DEM and surface subsidence models based on UAV LiDAR or multi-view oblique photogrammetry cameras. However, in practical applications, although traditional leveling observations have high accuracy, they often cannot obtain spatially continuous subsidence data and have low efficiency; although time-series InSAR technology can obtain small surface deformations in the edge area of ​​the subsidence basin with high accuracy, it is difficult to accurately obtain large gradient subsidence on the surface of the central area of ​​the subsidence basin; although UAV LiDAR technology and oblique photogrammetry technology can obtain relatively accurate DEM data, especially in the monitoring of surface subsidence in the central area of ​​the subsidence basin, they show high accuracy, but airborne LiDAR and multi-view oblique photogrammetry cameras are relatively expensive and have high barriers to entry.

[0004] Therefore, using the RGB camera carried by the UAV to obtain the surface subsidence information of the mining area is a novel and compromise method that can simultaneously meet the requirements of low cost, high resolution and relatively reliable accuracy. Summary of the invention

[0005] In view of the problems mentioned in the background technology, the purpose of the present invention is to provide a method for extracting surface subsidence information in mining areas based on visible light images of unmanned aerial vehicles, so as to solve the problems mentioned in the above background technology.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] A method for extracting surface subsidence information in mining areas based on UAV visible light images includes data collection and data processing. The data processing mainly includes obtaining digital orthophotos DOM and digital surface models DSM based on UAV visible light images, and generating a UAV subsidence model that can reflect surface subsidence information based on these; generating an InSAR subsidence model based on SAR data; and obtaining an improved UAV subsidence model based on the UAV subsidence model and the InSAR subsidence model combined with measured data.

[0008] Specifically, data collection includes obtaining working face mining information, UAV visible light images, leveling data, SAR data, image control points and checkpoint coordinates, and auxiliary data required for InSAR data processing.

[0009] Specifically, obtain the working face mining information: obtain specific working face mining information from the preliminary preparation of the mining area, which should include the working face mining cycle, strike length, dip width, average mining depth and coal seam inclination, and determine and set the mining drone flight time, route planning, flight altitude and flight speed based on the mining information.

[0010] Specifically, the drone obtains visible light images: drone data collection is carried out before and after mining, and the drone visible light images before and after mining are obtained. It is worth noting that the visible light images collected by the drone before and after mining should have a high heading overlap and lateral overlap, and completely cover the subsidence basin and an area of ​​a certain width around it.

[0011] Specifically, leveling data: multiple groups of leveling observation lines are arranged along the edge of the working face and through the center of the mining area to measure and record relevant point data.

[0012] Specifically, the generation of the UAV subsidence model includes the following steps:

[0013] 1.1 Initial data processing: The visible light images of the UAV before and after mining are imported into the UAV data processing software. After initialization processing, image control points, re-matching and optimization, and generation of dense point clouds, the digital orthophoto DOM and digital surface model DSM are finally obtained;

[0014] 1.2 Obtaining the initial subsidence model: Import the digital surface models (DSM) of the two periods before and after mining into the geographic information system platform respectively, and use the raster calculator to subtract the digital surface models (DSM) of the two periods to obtain the initial subsidence model;

[0015] 1.3 Obtain vegetation mask: Import the digital orthophotos DOM before and after mining into the image processing platform respectively, use the band operation tool to calculate the visible light vegetation index before and after mining, select the appropriate threshold value to segment the DOM image in combination with the visible light vegetation index to extract vegetation pixels, the sum of vegetation pixels is the vegetation coverage boundary, and take the union of the vegetation coverage boundaries before and after mining as the final vegetation mask;

[0016] 1.4 Remove the mask and obtain the UAV subsidence model: Apply the vegetation mask to the initial subsidence model to remove the vegetation coverage area, and then use the spatial interpolation method to fill in the surface subsidence information of the original vegetation coverage area to obtain the UAV surface subsidence model.

[0017] Specifically, the InSAR subsidence model generation includes the following steps:

[0018] 2.1 Based on multi-period SAR data covering the entire study area in space and the entire mining cycle of the working face in time, the cumulative surface settlement value of each pixel position is calculated using time series InSAR technology;

[0019] 2.2 The InSAR surface subsidence model is obtained by processing the accumulated surface settlement values.

[0020] Preferably, the SAR data adopts data obtained based on SAR.

[0021] Specifically, improving the generation of the UAV subsidence model includes the following steps:

[0022] 3.1 Error calibration: Use leveling observation data to correct the systematic errors of the UAV subsidence model and the InSAR subsidence model;

[0023] 3.2 Determine the fusion boundary: Combine relevant literature and leveling observation data to select an appropriate subsidence threshold to determine the fusion boundary, which is located at the edge of the subsidence basin;

[0024] 3.3 Improved UAV surface subsidence model: The surface within the fusion boundary is a large gradient subsidence, so the calibrated UAV subsidence model with higher accuracy is used. The surface outside the fusion boundary is a small subsidence area, so the calibrated InSAR subsidence model with higher accuracy is used.

[0025] Specifically, it also includes accuracy verification, including the following steps:

[0026] (1) Obtaining the three-dimensional coordinates of the checkpoints in the data collection part: evenly distribute and record the three-dimensional coordinates of the checkpoints in the UAV flight area;

[0027] (2) Combine the leveling observation data before and after mining to obtain the difference in the settlement value curve along the strike and dip of the working face, and combine the checkpoint data before and after mining to obtain the difference in the settlement value of the checkpoint;

[0028] (3) The initial subsidence model, InSAR subsidence model and improved UAV subsidence model were tested for their accuracy by combining the differences in subsidence value curves and the differences in checkpoint subsidence values.

[0029] Beneficial Effects

[0030] The present invention obtains a surface subsidence model based on visible light images of unmanned aerial vehicles, greatly reduces the cost of surface subsidence monitoring, can fully utilize the advantages of high spatial resolution and high flexibility of unmanned aerial vehicle platforms, and has high accuracy; the present invention uses visible light vegetation index to extract vegetation information and further generates a vegetation mask, and on this basis, removes the vegetation coverage area in the unmanned aerial vehicle subsidence model, thereby eliminating the influence of vegetation height changes on the accuracy of the subsidence model and improving the model accuracy; the method proposed in the present invention uses leveling data to correct the systematic errors of the unmanned aerial vehicle subsidence model and the InSAR subsidence model, so that the obtained subsidence model has higher accuracy; the method proposed in the present invention determines a fusion boundary by determining a suitable subsidence threshold, and a UAV subsidence model with higher accuracy is used within the fusion boundary, and an InSAR subsidence model with higher accuracy is used outside the fusion boundary. The obtained improved subsidence model can reflect the surface subsidence information of the entire subsidence basin with higher accuracy.

[0031] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The present invention provides a flowchart of a method for extracting surface subsidence information in mining areas based on UAV visible light images. DETAILED DESCRIPTION

[0033] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0034] Example 1

[0035] refer to Figure 1 The method for extracting surface subsidence information in a mining area based on visible light images of unmanned aerial vehicles described in this embodiment includes two parts: data collection and data processing.

[0036] Specifically, data collection includes obtaining working face mining information, UAV visible light images, leveling data, image control points and checkpoint coordinates, and auxiliary data required for InSAR data processing.

[0037] Obtaining working face mining information: Obtaining specific working face mining information from the preliminary preparation of the mining area, which should include the working face mining cycle, strike length, dip width, average mining depth and coal seam dip, and determining and setting the mining drone flight time, route planning, flight altitude, flight speed and other parameters based on the mining information;

[0038] Obtaining drone visible light images: drone data collection is carried out before and after mining to obtain drone visible light images before and after mining. It is worth noting that the visible light images collected by drones before and after mining should have a high degree of heading overlap and lateral overlap, and completely cover the subsidence basin and an area of ​​a certain width around it.

[0039] Leveling data: Multiple groups of leveling observation lines are arranged along the edge of the working face and through the center of the mining area to measure and record relevant point data.

[0040] Image control points and checkpoint coordinates: Evenly distribute and record the three-dimensional coordinates of the checkpoints within the drone’s flight area.

[0041] Auxiliary data required for InSAR data processing: Search and obtain the auxiliary data required for InSAR data processing on the geographic information system platform.

[0042] Data processing mainly includes obtaining digital orthophotos DOM and digital surface models DSM based on UAV visible light images, and on this basis generating a UAV subsidence model that can reflect surface subsidence information; generating an InSAR subsidence model based on SAR data; and obtaining an improved UAV subsidence model based on the UAV subsidence model and the InSAR subsidence model combined with measured data.

[0043] The improved UAV subsidence model obtained based on the UAV subsidence model and InSAR subsidence model combined with the measured data is used to correct the systematic errors of the UAV subsidence model and the InSAR subsidence model, so that the obtained subsidence model has higher accuracy.

[0044] The generation of the UAV subsidence model includes the following steps:

[0045] 1.1 Initial data processing: The visible light images of the UAV before and after mining are imported into the UAV data processing software respectively. After initialization processing, image control points, re-matching and optimization, and generation of dense point clouds, digital orthophotos DOM (image data sets generated by digital differential correction and mosaicking of aerial photos or satellite remote sensing images, and cropping according to a certain map range) and digital surface models DSM (three-dimensional geometric model of the earth's surface, which records the elevation information of the surface of objects in discrete digital form) presented in the form of raster images are finally obtained.

[0046] Preferably, the drone data processing software uses Pix4Dmapper.

[0047] 1.2 Obtain the initial subsidence model: import the digital surface models (DSM) of the two periods before and after mining into the geographic information system platform respectively, and use the raster calculator to subtract the digital surface models (DSM) of the two periods, that is, subtract the corresponding pixel values ​​of the two sets of digital surface models (DSM) to obtain the initial subsidence model in the form of a raster image.

[0048] Preferably, the geographic information system platform uses ArcGIS.

[0049] Compared with other commonly used surface subsidence monitoring technologies, such as traditional leveling, UAV LiDAR, UAV oblique photogrammetry, and time-series InSAR technology, obtaining surface subsidence models based on UAV visible light images greatly reduces the cost of surface subsidence monitoring, and can fully utilize the advantages of high spatial resolution and high flexibility of the UAV platform while having higher accuracy.

[0050] 1.3 Obtain vegetation mask: The digital orthophoto DOM of the two periods before and after mining are imported into the image processing platform respectively. Since each pixel of the digital orthophoto DOM has three pixel values ​​corresponding to three bands, the image processing platform uses band operation tools to calculate (relatively mature calculation methods have been available in related fields) to obtain the visible light vegetation index before and after mining in the form of raster images. Combined with the visible light vegetation index, a suitable threshold is selected to segment the DOM image to extract vegetation pixels. The sum of the vegetation pixels is the vegetation coverage boundary, and the union of the vegetation coverage boundaries before and after mining is taken as the final vegetation mask.

[0051] Each vegetation index is a formula that combines operations of different bands, such as the visible light band difference vegetation index (VDVI). When a vegetation index needs to be calculated, the corresponding band value is entered into the formula, and the calculation result is a vegetation index layer, such as the VDVI layer, in which each pixel corresponds to a value.

[0052] It is worth noting that the pixel values ​​corresponding to vegetation and water surface in the visible light vegetation index calculation result layer have their own rules and correspond to different threshold ranges. As common knowledge among people working in related fields, when selecting, just select the appropriate one within the threshold range based on the actual visible light vegetation index.

[0053] Preferably, the image processing platform uses ENVI.

[0054] 1.4 Remove the mask and obtain the UAV subsidence model: Apply the vegetation mask to the initial subsidence model to remove the vegetation coverage area, and then use the spatial interpolation method (there are relatively mature and clear spatial interpolation methods in the relevant field) to fill in the surface subsidence information of the original vegetation coverage area, and then obtain the UAV surface subsidence model.

[0055] It is worth noting that the UAV cannot obtain the surface subsidence data of the corresponding location through the vegetation. The change of vegetation height will affect the accuracy of the subsidence model. The visible light vegetation index is used to extract vegetation information and further generate a vegetation mask. On this basis, the vegetation coverage area in the UAV subsidence model is eliminated, thereby eliminating the impact of vegetation height changes on the accuracy of the subsidence model and improving the model accuracy.

[0056] The InSAR subsidence model generation specifically includes the following steps:

[0057] 2.1 Based on multi-period SAR data covering the entire study area in space and the entire mining cycle of the working face in time, the cumulative surface settlement value of each pixel position is calculated using time series InSAR technology;

[0058] 2.2 The InSAR surface subsidence model is obtained by processing the accumulated surface settlement values.

[0059] The data processing methods include traditional modeling methods, such as linear regression model, polynomial regression model; time series analysis methods, such as ARIMA model, Kalman filter model, etc.

[0060] Improving the generation of UAV subsidence model specifically includes the following steps:

[0061] 3.1 Error calibration: Use leveling observation data to correct the systematic errors of the UAV subsidence model and the InSAR subsidence model;

[0062] 3.2 Determine the fusion boundary: Combine relevant literature and leveling observation data to select an appropriate subsidence threshold to determine the fusion boundary, which is located at the edge of the subsidence basin;

[0063] 3.3 Improved UAV surface subsidence model: The surface within the fusion boundary is a large-gradient subsidence, so the UAV subsidence model with higher accuracy (after calibration) is used. The surface outside the fusion boundary is a small subsidence area, so the InSAR subsidence model with higher accuracy (after calibration) is used.

[0064] The fusion boundary is determined by determining an appropriate subsidence threshold. The UAV subsidence model with higher accuracy is used within the fusion boundary, and the InSAR subsidence model with higher accuracy is used outside the fusion boundary. The obtained improved subsidence model can reflect the surface subsidence information of the entire subsidence basin with higher accuracy.

[0065] Example 2

[0066] refer to Figure 1 , the method for extracting surface subsidence information in mining areas based on visible light images from drones described in this embodiment can also be used for accuracy verification to verify the accuracy of the above extraction method.

[0067] Specifically, the following steps are included:

[0068] (1) Combine the level observation data before and after mining to obtain the difference in the settlement value curve along the strike and dip of the working face, and combine the checkpoint data before and after mining to obtain the difference in the settlement value of the checkpoint;

[0069] (2) The initial subsidence model, InSAR subsidence model and improved UAV subsidence model were tested for their accuracy based on the differences in subsidence value curves and the differences in subsidence values ​​at the checkpoints.

[0070] This will determine the rationality and scientificity of this method.

[0071] It should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0072] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0073] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral one; it can be a mechanical connection, an electrical connection, or communication with each other; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0074] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0075] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for extracting surface subsidence information in mining areas based on visible light images from unmanned aerial vehicles, characterized in that: It includes data collection and data processing. Data processing mainly includes obtaining digital orthophotos DOM and digital surface models DSM based on UAV visible light images, and on this basis generating a UAV subsidence model that can reflect surface subsidence information; generating an InSAR subsidence model based on SAR data; and obtaining an improved UAV subsidence model based on the UAV subsidence model and the InSAR subsidence model combined with measured data.

2. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 1 is characterized by: Data collection includes obtaining working face mining information, UAV visible light images, leveling data, SAR data, image control points and checkpoint coordinates, and auxiliary data required for InSAR data processing.

3. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 2 is characterized by: Obtain working face mining information: Obtain specific working face mining information from the preliminary preparation of the mining area, which should include the working face mining cycle, strike length, dip width, average mining depth and coal seam inclination, and determine and set the mining drone flight time, route planning, flight altitude and flight speed based on the mining information.

4. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 3 is characterized by: Obtaining visible light images using drones: drone data collection is carried out before and after mining to obtain visible light images of the drones before and after mining. It is worth noting that the visible light images collected by drones before and after mining should have a high degree of heading overlap and lateral overlap, and completely cover the subsidence basin and an area of ​​a certain width around it.

5. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 4 is characterized in that: Leveling data: Multiple groups of leveling observation lines are arranged along the edge of the working face and through the center of the mining area to measure and record relevant point data.

6. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 5 is characterized by: Generation of UAV subsidence model The following steps are involved: 1.1 Initial data processing: The visible light images of the UAV before and after mining are imported into the UAV data processing software. After initialization processing, image control points, re-matching and optimization, and generation of dense point clouds, the digital orthophoto DOM and digital surface model DSM are finally obtained; 1.2 Obtaining the initial subsidence model: Import the digital surface models (DSM) of the two periods before and after mining into the geographic information system platform respectively, and use the raster calculator to subtract the digital surface models (DSM) of the two periods to obtain the initial subsidence model; 1.3 Obtain vegetation mask: Import the digital orthophotos DOM before and after mining into the image processing platform respectively, use the band operation tool to calculate the visible light vegetation index before and after mining, select the appropriate threshold value to segment the DOM image in combination with the visible light vegetation index to extract vegetation pixels, the sum of vegetation pixels is the vegetation coverage boundary, and take the union of the vegetation coverage boundaries before and after mining as the final vegetation mask; 1.4 Remove the mask and obtain the UAV subsidence model: Apply the vegetation mask to the initial subsidence model to remove the vegetation coverage area, and then use the spatial interpolation method to fill in the surface subsidence information of the original vegetation coverage area to obtain the UAV surface subsidence model.

7. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 6 is characterized by: The InSAR subsidence model generation specifically includes the following steps: 2.1 Based on multi-period SAR data covering the entire study area in space and the entire mining cycle of the working face in time, the cumulative surface settlement value of each pixel position is calculated using time series InSAR technology; 2.2 The InSAR surface subsidence model is obtained by processing the accumulated surface settlement values.

8. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 7 is characterized by: Improving the generation of UAV subsidence model specifically includes the following steps: 3.1 Error calibration: Use leveling observation data to correct the systematic errors of the UAV subsidence model and the InSAR subsidence model; 3.2 Determine the fusion boundary: Combine relevant literature and leveling observation data to select an appropriate subsidence threshold to determine the fusion boundary, which is located at the edge of the subsidence basin; 3.3 Improved UAV surface subsidence model: The surface within the fusion boundary is a large gradient subsidence, so the calibrated UAV subsidence model with higher accuracy is used. The surface outside the fusion boundary is a small subsidence area, so the calibrated InSAR subsidence model with higher accuracy is used.

9. The method for extracting mining area surface subsidence information based on drone visible light images according to claim 1 is characterized by: It also includes accuracy verification, which includes the following steps: (1) Obtaining the three-dimensional coordinates of the checkpoints in the data collection part: evenly distribute and record the three-dimensional coordinates of the checkpoints in the UAV flight area; (2) Combine the leveling observation data before and after mining to obtain the difference in the settlement value curve along the strike and dip of the working face, and combine the checkpoint data before and after mining to obtain the difference in the settlement value of the checkpoint; (3) The initial subsidence model, InSAR subsidence model and improved UAV subsidence model were tested for their accuracy by combining the differences in subsidence value curves and the differences in checkpoint subsidence values.

Images