A method and system for automatically identifying open-pit mine areas
Through mathematical morphological operations and deep learning technology, the morphological feature index of open-pit mining areas is constructed based on multi-scale and multi-directional differential morphological profiles. Combined with threshold segmentation, the problem of open-pit mining areas is solved, automatic identification and fast and accurate information support is achieved, and the improvement of the improvement of the mining and ecological restoration efficiency.
Patent Information
- Application Number
- CN202410460474.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-17
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-04-17
AI Technical Summary
It is difficult to identify open-pit mining areas, especially in the absence of sample collection, and the existing technology is difficult to quickly and accurately identify the scope of the mining area, resulting in complex and changeable environmental problems, ecological system fragmentation and biodiversity loss.
Mathematical morphological operations are used to construct the morphological feature index of the mine area based on multi-scale and multi-directional differential morphological profiles. Combined with deep learning and text analysis, contour features are extracted through satellite images and threshold segmentation is performed to automatically identify the scope of the open-pit mine area.
It realizes automatic identification of open-pit mining areas without sample collection, provides fast and accurate information support, and improves the efficiency of mining and ecological restoration.
Smart Images

Figure CN118155077B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of open-pit mine identification, and in particular to an open-pit mine automatic identification method and system. Background Art
[0002] Open-pit mining refers to the place where ore is mined from the ore body through open-pit mining. Open-pit mining is a technology that mines directly from the ground. Compared with underground mining, open-pit mining has higher productivity, lower production costs and safer and more hygienic working conditions. In open-pit mining areas, the area above the surface closed circle is usually called a hillside open-pit mine, and the area below the surface closed circle is called a depression open-pit mine, depending on the topographic conditions of the buried ore deposit. The components of open-pit mining areas include steps, step slopes and platforms (flat plates).
[0003] Open-pit mining usually requires the removal of overburden from the ground, so it faces certain environmental problems. In addition, due to historical factors and my country's special geological environment, the environmental problems of open-pit coal mining in my country are complex and changeable, and there is great pressure. In the past few decades, open-pit mining in China has grown at an unprecedented rate, causing a series of negative environmental impacts, including the fragmentation and degradation of ecosystems and the loss of biodiversity. Satellite images are an important source of mining scope information to supplement surveys and statistics, while the identification of open-pit mines currently requires a large amount of sample collection, and because open-pit mines are usually large in scope, the identification of open-pit mines is still difficult at present. Summary of the invention
[0004] In view of the above-mentioned problems, the present invention aims to provide a method for automatic identification of open-pit mines, which constructs a mining area morphological feature index based on multi-scale and multi-directional differential morphological contours through mathematical morphological operations, and combines threshold segmentation to realize automatic extraction of the open-pit mine area without sample collection.
[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows: a method for automatic identification of open-pit mines, characterized in that it comprises the following steps:
[0006] S1. Open-pit mine location
[0007] Based on deep learning and text analysis of keywords related to open-pit mines, the basic geographic information of open-pit mines is screened and obtained, and the geographic location of the geographic information is calibrated in the satellite data source;
[0008] S2. Constructing the morphological characteristics of the open-pit mine area
[0009] Based on the calculation of typical ground categories of satellite images, the contour features of the open-pit mine area are extracted, and the morphological characteristic index of the open-pit mine area is calculated based on the contour features;
[0010] S3, automatic identification of open-pit mine area
[0011] Based on the numerical classification of the morphological characteristic index of the open-pit mine area, it is used to realize the automatic identification of the mining area in the area to be detected.
[0012] Furthermore, the open-pit mine area positioning described in step S1 includes the following sub-steps:
[0013] S11, selecting a vocabulary containing administrative division units of the open-pit mine area to be detected as a training data set;
[0014] S12, using a bidirectional long short-term memory network and introducing an attention mechanism to preserve the semantic relationship between words;
[0015] S13. Capture local features in the text, including but not limited to direction, longitude and latitude, administrative division name and open-pit mine type, to achieve accurate capture and calibration of geographic location information.
[0016] Furthermore, in step S12, the attention score is defined as:
[0017]
[0018] In the formula, [h i ;e i ] represents the connection of the output and embedding vectors in the bidirectional long short-term memory network by column, tanh is the hyperbolic tangent activation function, and w a and W a are learnable parameters.
[0019] Furthermore, the geographic location information is calibrated by combining the publicly available global mining database to supplement and update the geographic location of open-pit mining activities, and calibrating in multiple satellite data sources.
[0020] Furthermore, the step S2 of constructing the morphological characteristics of the open-pit mine area includes the following sub-steps:
[0021] S21. In the satellite image, randomly select typical ground categories, draw the average spectral curve, and calculate the band with the largest difference in maximum, average and minimum values in the spectral curve;
[0022] S22. Morphological operators are introduced to extract differential morphological contour features of open-pit mines at multiple scales and directions. Specifically:
[0023] WTH(d,s)=Br B (se)
[0024] In the formula, r B(se) is the image opening after reconstruction of the image in the band with the largest difference, WTH is the white top hat transformation after reconstruction of the structure element se with scale s and direction d as parameters; Se is the shape and size designed as the filtering operator;
[0025] The residuals of two consecutive scales in the same direction are calculated as:
[0026] DMP(d,s)=|WTH(d,(s+Δs))|-WTH(d,s)
[0027] Where Δs represents the scale parameter, whose interval is smin≤Δs≤smin, and DPM is the multi-scale and multi-directional differential morphological profile;
[0028] S23. Add the band ratio parameter to generate the open-pit mine morphological characteristic index:
[0029]
[0030] Among them, Ratio is the ratio of the band with the largest difference to the near-infrared band, its n=d×s represents the total number of DMPs, and λ is the adjustment factor.
[0031] Furthermore, the open-pit mine area range is automatically identified in step S3 by selecting the histogram entropy method to segment the open-pit mine area morphological feature index, and the optimal threshold is obtained by maximizing the prior entropy of the target and background classes:
[0032]
[0033] Where E is the histogram entropy, L is the number of gray levels of the image, and p i is the probability of each gray level.
[0034] An open-pit mine area automatic identification system, characterized by comprising:
[0035] The open-pit mine positioning module is used to conduct deep learning and text analysis of keywords related to the open-pit mine, screen and obtain basic geographic information of the open-pit mine, and calibrate the geographic location in the geographic information in the satellite data source;
[0036] The open-pit mine area morphological feature construction module is used to calculate the typical ground categories of satellite images, extract the contour features of the open-pit mine area, and calculate and generate the open-pit mine area morphological feature index based on the contour features;
[0037] The open-pit mine area automatic identification module is used to divide the numerical values of the open-pit mine area morphological characteristic index and use it for automatic identification of the mining area scope in the area to be detected.
[0038] The beneficial effects of the present invention are as follows: the automatic identification method disclosed in the present invention constructs a mining area morphological feature index based on multi-scale and multi-directional differential morphological contours through mathematical morphological operations, and combines threshold segmentation to automatically extract the scope of the open-pit mining area without sample collection, thereby realizing automatic identification of the open-pit mining area and providing fast and accurate information support for the mining and ecological restoration of the open-pit mining area. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 This is a flow chart of automatic identification of open-pit mine areas according to the present invention.
[0040] Figure 2 The invention provides a specific example of implementing the identification method disclosed in the present invention. DETAILED DESCRIPTION
[0041] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0042] See attached Figures 1-2 The method for automatically identifying an open-pit mine area shown in the figure comprises the following steps:
[0043] S1. Open-pit mine location
[0044] The text analysis method based on deep learning, such as using "Erdos City", "coal mine", "located in", "region", "banner", "town", "open-pit mining area", "open-pit mineral mining" or "open-pit mining" and other keywords as search terms in similar search engines such as Baidu, obtains basic geographic information of open-pit mining areas. Specifically, it includes the following steps:
[0045] S11, selecting a vocabulary containing administrative division units of the open-pit mine area to be detected as a training data set;
[0046] S12. Use a bidirectional long short-term memory network and introduce an attention mechanism to preserve the semantic relationship between words:
[0047]
[0048] In the formula, [h i ;e i ] represents the connection of the output and embedding vectors in the bidirectional long short-term memory network by column, tanh is the hyperbolic tangent activation function, and w a and W a are learnable parameters.
[0049] S13. Use convolutional neural networks to capture local features in the text, including but not limited to direction (east, south, west, north), longitude and latitude, administrative division names (provinces, cities, counties, districts, banners, states, leagues, townships, towns, streets, etc.) and open-pit mines (coal mines, iron mines, copper mines, etc.), to form a comprehensive text to achieve accurate capture of geographic location information.
[0050] S14. In combination with the publicly available global mining database (https: / / www.spglobal.com), the approximate geographic locations of open-pit mining activities are supplemented and updated, and calibrated in multiple satellite data sources (including Landsat satellites, Google satellites, Microsoft Bing images, and Sentinel-2 cloudless).
[0051] S2. Constructing the morphological characteristics of the open-pit mine area
[0052] Based on the calculation of typical ground categories in satellite images, the contour features of the open-pit mine area are extracted, and the morphological feature index of the open-pit mine area is calculated based on the contour features. The specific steps include:
[0053] S21. In satellite images, randomly select typical ground categories (such as roads, water bodies, vegetation, bare land, settlements and open-pit mining areas), draw the average spectral curve, and calculate the band with the largest difference in maximum, average and minimum values in the spectral curve (based on experience and historical literature analysis, the short-wave infrared band is usually obtained).
[0054] S22. Morphological operators are introduced to extract differential morphological contour features of open-pit mines at multiple scales and directions. Since open-pit mines are usually displayed as irregular polygons with isotropic shapes, a top-hat morphological transformation is used to better preserve the shape by taking the residual of the open, closed, and original images:
[0055] WTH(d,s)=Br B (se)
[0056] In the formula, r B (se) is the image opening after reconstruction of the image in the band with the largest difference, WTH is the white top hat transform after reconstruction of the structure element se with scale s and direction d as parameters; Se is the shape and size designed as a filter operator, which is used as a probe to extract or suppress specific image structures. Subsequently, the residual of two consecutive scales in the same direction is calculated as:
[0057] DMP(d, s) = |WTH(d, (s+Δs))| - WTH(d, s) where Δs represents the scale parameter, which is in the range of smin≤Δs≤smin, usually determined by the spatial resolution and image size; DPM is a multi-scale, multi-directional differential morphological profile. Therefore, by changing d and Δs, a series of DMP features can be generated.
[0058] S23. Add the band ratio parameter to highlight the spectral difference between the open-pit mine area and other land covers, and generate the open-pit mine area morphological characteristic index:
[0059]
[0060] Among them, Ratio is the ratio of the band with the largest difference to the near-infrared band, its n=d×s represents the total number of DMPs, and λ is the adjustment factor.
[0061] S3, automatic identification of open-pit mine area
[0062] The morphological characteristic index of the open-pit mine area is segmented based on the histogram entropy method, and the optimal threshold is obtained by maximizing the prior entropy of the target and background classes:
[0063]
[0064] Where E is the histogram entropy, L is the number of gray levels of the image, and p i is the probability of each gray level.
[0065] The identification method of the present application can be used in different regions to determine differentiated optimal thresholds to achieve automatic identification of mining area ranges.
[0066] The present application also provides an open-pit mine area automatic identification system, comprising:
[0067] The open-pit mine positioning module is used to conduct deep learning and text analysis of keywords related to the open-pit mine, screen and obtain basic geographic information of the open-pit mine, and calibrate the geographic location in the geographic information in the satellite data source;
[0068] The open-pit mine area morphological feature construction module is used to calculate the typical ground categories of satellite images, extract the contour features of the open-pit mine area, and calculate and generate the open-pit mine area morphological feature index based on the contour features;
[0069] The open-pit mine area automatic identification module is used to divide the numerical values of the open-pit mine area morphological characteristic index and use it for automatic identification of the mining area scope in the area to be detected.
[0070] The recognition system is used to execute the above recognition method to realize the detection of open-pit mine areas. In addition, by implementing modular programming of the system in Google Earth Engine, remote sensing satellite images of any time and any place can be called at any time to realize all-weather recognition of open-pit mine areas.
[0071] Accuracy test: Use existing open-pit mine datasets (such as the publicly available global mine dataset: https: / / doi.org / 10.1038 / s41597-020-00624-w) to verify the identification method of this application ( Figure 2 It is shown that the identification method of the present application can identify more than 90% of the open-pit mine area.
[0072] The principle of the present invention is: a fully automatic and efficient method for extracting open-pit mines from satellite images is proposed. First, a deep learning text analysis method is used with the assistance of an existing mine database to locate the geographical location of the open-pit mine. Then, through mathematical morphological operations, a mine morphological feature index is constructed based on multi-scale and multi-directional differential morphological contours. Combined with threshold segmentation, the open-pit mine range can be automatically extracted without sample collection, so as to realize automatic identification of the open-pit mine.
[0073] The above shows and describes the basic principle, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention claimed.
Claims
1. A method for automatic identification of open-pit mines, characterized in that: The following steps are involved: S1. Open-pit mine location Based on deep learning and text analysis of keywords related to open-pit mines, the basic geographic information of open-pit mines is screened and obtained, and the geographic location of the geographic information is calibrated in the satellite data source; S2. Constructing the morphological characteristics of the open-pit mine area Based on the calculation of typical ground categories of satellite images, the contour features of the open-pit mine area are extracted, and the morphological characteristic index of the open-pit mine area is calculated based on the contour features; The step S2 of constructing the morphological features of the open-pit mine area includes the following sub-steps: S21. In the satellite image, randomly select typical ground categories, draw the average spectral curve, and calculate the band with the largest difference in the spectral curve; S22. Morphological operators are introduced to extract differential morphological contour features of open-pit mines at multiple scales and directions. Specifically: ; In the formula, (se) is the image opening after reconstruction of the image in the band with the largest difference, WTH is the white top hat transform after reconstruction of the structure element se with scale s and direction d as parameters; se is the shape and size designed as the filtering operator; The residuals of two consecutive scales in the same direction are calculated as: ; In the formula, Δs represents the scale parameter, and DPM is the multi-scale and multi-directional differential morphological profile; S23. Add the band ratio parameter to generate the open-pit mine morphological characteristic index: ; Among them, Ratio is the ratio of the band with the largest difference to the near-infrared band, its n = d × s represents the total number of DMPs, and λ is the adjustment factor; S3, automatic identification of open-pit mine area Based on the numerical division of the morphological characteristic index of the open-pit mine area, it is used for the area to be detected to realize the automatic identification of the scope of the mining area; Based on the numerical division of the morphological characteristic index of the open-pit mine area, it is used for the area to be detected to realize the automatic identification of the scope of the mining area.
2. The automatic identification method according to claim 1, characterized in that: The open-pit mine area positioning described in step S1 includes the following sub-steps: S11, selecting a vocabulary containing administrative division units of the open-pit mine area to be detected as a training data set; S12, using a bidirectional long short-term memory network and introducing an attention mechanism to preserve the semantic relationship between words; S13. Capture local features in the text, including but not limited to direction, longitude and latitude, administrative division name and open-pit mine type, to achieve accurate capture and calibration of geographic location information.
3. The automatic identification method according to claim 2, characterized in that: In step S12, the attention score is defined as: ; Where, represents the connection of the output and embedding vectors in the bidirectional long short-term memory network by column, tanh is the hyperbolic tangent activation function, and w a and W a are learnable parameters.
4. The automatic identification method according to claim 3, characterized in that: The geographic location information is calibrated by combining the publicly available global mining database, supplementing and updating the geographic location of open-pit mining activities, and calibrating in a variety of satellite data sources.
5. The automatic identification method according to claim 1, characterized in that: The open-pit mine area range is automatically identified in step S3 by selecting the histogram entropy method to segment the open-pit mine area morphological feature index, and the optimal threshold is obtained by maximizing the prior entropy of the target and background classes: ; Where E is the histogram entropy, L is the number of gray levels of the image, and p i is the probability of each gray level.
6. An open-pit mine automatic identification system, characterized in that: include: The open-pit mine positioning module is used to screen and obtain basic geographic information of the open-pit mine based on deep learning and text analysis of keywords related to the open-pit mine, and to calibrate the geographic location in the geographic information in the satellite data source; The open-pit mine morphological feature construction module is used to calculate the typical ground categories of satellite images, extract the contour features of the open-pit mine area, and calculate the open-pit mine morphological feature index based on the contour features. In the satellite image, typical ground categories are randomly selected, the average spectral curve is drawn, and the band with the largest difference in the spectral curve is calculated. The morphological operator is introduced to extract the differential morphological contour features of the open-pit mine area at multiple scales and directions, specifically: ; In the formula, (se) is the image opening after reconstruction of the image in the band with the largest difference, WTH is the white top hat transform after reconstruction of the structure element se with scale s and direction d as parameters; se is the shape and size designed as the filtering operator; The residuals of two consecutive scales in the same direction are calculated as: ; In the formula, Δs represents the scale parameter, and DPM is the multi-scale and multi-directional differential morphological profile; Add the band ratio parameter to generate the open-pit mine morphological characteristic index: ; Among them, Ratio is the ratio of the band with the largest difference to the near-infrared band, its n = d × s represents the total number of DMPs, and λ is the adjustment factor; The module for automatic identification of the scope of open-pit mines is used to realize automatic identification of the scope of mines in areas to be inspected based on the numerical classification of the morphological characteristic index of open-pit mines; the module for automatic identification of the scope of mines in areas to be inspected is used to realize automatic identification of the scope of mines in areas to be inspected based on the numerical classification of the morphological characteristic index of open-pit mines.
