Method, device, computer equipment, and storage medium for identifying the dry beach length of a tailings pond

Abstract

The present disclosure relates to a method, device, computer equipment, and storage medium for identifying the dry beach length of a tailings pond. The method includes: obtaining a remote sensing image of the tailings pond; inputting the remote sensing image of the tailings pond into a pre-trained target detection model, and outputting a target detection result of the tailings pond through the target detection model; classifying multiple types of tailings pond elements contained in the target detection result of the tailings pond by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements; merging each type of the same tailings pond elements, determining a reference line corresponding to the tailings pond elements according to the merging result, and obtaining the dry beach length based on the reference line corresponding to the tailings pond elements. By using this method, the dry beach length can be automatically and accurately identified.

Description

Technical Field

[0001] The present disclosure relates to the field of image processing technologies, and particularly to a data transmission method, apparatus, computer device, and storage medium. Background Art

[0002] With the development of image processing technologies, the dry beach length of a tailings pond is the horizontal distance from the water edge line inside the pond to the beach top, and it is one of the flood control elements of the tailings pond. Extracting the dry beach length of the nearest point from the water edge line inside the pond to the beach top on high-resolution remote sensing images as a dynamic monitoring index, the dry beach length is the horizontal distance from the nearest point on the water edge line inside the pond to the beach top line. Currently, for the identification of the dry beach length of tailings ponds, most methods use on-site installation of optical cameras, distance monitoring sensors, etc. Such equipment has a high layout cost, and most of them can only monitor the dry beach length of a single tailings pond, making it difficult to achieve large-scale automated monitoring.

[0003] Therefore, existing detection technologies have problems such as low detection accuracy of tailings ponds and insufficient calculation accuracy of dry beach length, which seriously restrict the automated identification ability of the dry beach length of tailings ponds. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, apparatus, computer device, and storage medium for identifying the dry beach length of a tailings pond in view of the above technical problems.

[0005] In a first aspect, the present disclosure provides a method for identifying the dry beach length of a tailings pond. The method includes:

[0006] Obtain a remote sensing image of a tailings pond;

[0007] Input the remote sensing image of the tailings pond into a pre-trained target detection model, and output a target detection result of the tailings pond through the target detection model;

[0008] Use a superpixel segmentation algorithm and a random forest algorithm to classify multiple types of tailings pond elements contained in the target detection result of the tailings pond to obtain multiple types of tailings pond elements;

[0009] Merge each type of the same tailings pond elements, determine a reference line corresponding to the tailings pond elements according to the merging result, and obtain the dry beach length based on the reference line corresponding to the tailings pond elements.

[0010] In one embodiment, the step of using a superpixel segmentation algorithm and a random forest algorithm to classify multiple types of tailings pond elements contained in the target detection result of the tailings pond to obtain multiple types of tailings pond elements includes:

[0011] Assign a coordinate reference system to the target detection result of the tailings pond to obtain a target box vector;

[0012] Based on a preset buffer distance, expand the target box vector to obtain an expanded target box vector;

[0013] Determine a seed center point in the expanded target box vector, and through a clustering method, merge the same-class pixels based on the seed center point to form superpixels;

[0014] Detect the edges of the superpixels, and through centerline extraction and line-to-surface conversion, obtain a superpixel segmentation surface;

[0015] Use the random forest algorithm to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

[0016] In one embodiment, the using the random forest algorithm to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements includes:

[0017] Based on pre-determined classification samples, determine the element features of each type of tailings pond element;

[0018] Use the random forest algorithm to learn the element features to form a superpixel classification model;

[0019] Use the superpixel classification model to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

[0020] In one embodiment, the tailings pond elements include: embankment elements, dry beach elements, and water body elements; the merging of the same type of each tailings pond element, determining the reference line corresponding to the tailings pond element according to the merging result, and obtaining the dry beach length based on the reference line corresponding to the tailings pond element includes:

[0021] Merge the same tailings pond elements in the multiple types of tailings pond elements to obtain the merging result of each type of tailings pond element;

[0022] Use a curve simplification algorithm to process the merging result of the embankment elements to obtain the initial dam baseline;

[0023] Determine the outer water line of the water body element, and determine the dry beach length based on the distance between the initial dam baseline and the outer water line of the water body.

[0024] In one embodiment, the method further includes:

[0025] Obtain training remote sensing images, where the training remote sensing images contain multiple tailings pond areas;

[0026] Determine the tailings pond areas contained in the training remote sensing images. For the tailings pond areas, draw vector boxes and assign values to obtain the vector boundary of the tailings pond target detection box;

[0027] Train an object detection model based on the vector boundary of the tailings pond object detection box and the training remote sensing image.

[0028] In one embodiment, the object detection model includes a backbone network, a neck network, and a head network, and the backbone network includes a rectangular feature extraction module; the training of the object detection model based on the vector boundary of the tailings pond object detection box and the training remote sensing image includes:

[0029] Based on the backbone network, extract the tailings pond feature information contained in the vector boundary of the tailings pond object detection box in the training remote sensing image;

[0030] Based on the rectangular feature extraction module, perform secondary feature extraction on the tailings pond feature information to obtain high-order tailings pond feature information;

[0031] After dimension elevation of the high-order tailings pond feature information, fuse it with the tailings pond feature information, and use the fused feature information and the vector boundary of the tailings pond object detection box to train the object detection model.

[0032] In a second aspect, the present disclosure also provides a device for the dry beach length of a tailings pond. The device includes:

[0033] An image acquisition module for acquiring remote sensing images of the tailings pond;

[0034] A model processing module for inputting the remote sensing image of the tailings pond into a pre-trained object detection model, and outputting a tailings pond object detection result through the object detection model;

[0035] A classification module for classifying multiple types of tailings pond elements contained in the tailings pond object detection result by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements;

[0036] A length calculation module for merging each type of the same tailings pond elements, determining a reference line corresponding to the tailings pond elements according to the merging result, and obtaining the dry beach length based on the reference line corresponding to the tailings pond elements.

[0037] In a third aspect, the present disclosure also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in any of the above method embodiments are implemented.

[0038] In a fourth aspect, the present disclosure also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0039] Fifth aspect, the present disclosure also provides a computer program product. The computer program product includes a computer program which, when executed by a processor, implements the steps in any of the above method embodiments.

[0040] In the above embodiments, by obtaining the remote sensing images of the tailings pond and processing them using the pre-trained object detection model, it is possible to accurately locate the tailings pond-related objects at the macroscopic remote sensing image level, avoiding problems such as limited vision and low efficiency that may exist in manual on-site exploration. For example, for a tailings pond area located in a remote and large-scale area, relying on the remote sensing image combined with the object detection model can quickly and relatively accurately determine the specific location and approximate range of the tailings pond in the image, improving the comprehensiveness and accuracy of the tailings pond object detection. This model-based automated detection method can process a large amount of remote sensing image data at the same time, and is suitable for monitoring multiple tailings ponds or large areas containing tailings ponds, which helps to realize the extraction of tailings pond information on a large scale and in batches. Using the superpixel segmentation algorithm and the random forest algorithm to classify various types of tailings pond elements contained in the detection results of the tailings pond object detection can further refine the understanding of different elements inside the tailings pond, and can more deeply understand the details of the tailings pond, thereby improving the accuracy of dry beach identification. Merging the same type of tailings pond elements, and then determining the reference line corresponding to the tailings pond elements according to the merging result to obtain the dry beach length. The dry beach length is a key indicator for measuring the safety status of the tailings pond. Accurately mastering the dry beach length can help relevant personnel timely judge whether the tailings pond is in a safe operating state. For example, a too short dry beach length may indicate safety risks such as overtopping of the tailings pond. And through this method, accurate calculation of the dry beach length provides reliable data support for the safety warning and risk prevention and control of the tailings pond. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present disclosure. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0042] Figure 1 It is a schematic diagram of the application environment of the method for identifying the dry beach length of a tailings pond in an embodiment;

[0043] Figure 2 It is a schematic flowchart of the method for identifying the dry beach length of a tailings pond in an embodiment;

[0044] Figure 3 It is a schematic flowchart of step S206 in an embodiment;

[0045] Figure 4 It is a schematic flowchart of step S310 in an embodiment;

[0046] Figure 5 It is a schematic flowchart of step S208 in an embodiment;

[0047] Figure 6 It is a schematic flowchart of the model training step in an embodiment;

[0048] Figure 7 It is a schematic flowchart of step S606 in an embodiment;

[0049] Figure 8 It is a schematic block diagram of the structure of the tailings pond dry beach length identification device in an embodiment;

[0050] Figure 9 It is a schematic internal structure diagram of a computer device in an embodiment;

[0051] Figure 10 It is a schematic internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0052] In order to make the objectives, technical solutions and advantages of the present disclosure clearer and more understandable, the present disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present disclosure, and are not used to limit the present disclosure.

[0053] It should be noted that the terms "first", "second", etc. in the specification and claims of this article and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units does not necessarily need to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or equipment.

[0054] In this article, the term "and / or" is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0055] An embodiment of the present disclosure provides a method for identifying the dry beach length of a tailings pond, which can be applied to an application environment as shown in Figure 1 The terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or other network servers. The terminal 102 can obtain the remote sensing image of the tailings pond stored by the server 104. The terminal 102 can input the remote sensing image of the tailings pond into a pre-trained object detection model, and output the object detection result of the tailings pond through the object detection model. The terminal 102 classifies multiple types of tailings pond elements included in the object detection result of the tailings pond by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements. The terminal 102 merges each type of the same tailings pond elements, determines the baseline corresponding to the tailings pond elements according to the merging result, and obtains the dry beach length based on the reference line corresponding to the tailings pond elements. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

[0056] In one embodiment, as shown in Figure 2 A method for identifying the dry beach length of a tailings pond is provided. Taking the terminal 102 in Figure 1 as an example, the method includes the following steps:

[0057] S202, obtain the remote sensing image of the tailings pond.

[0058] Among them, the remote sensing image of the tailings pond can be image data about the tailings pond obtained by aerial or satellite remote sensing technology. The remote sensing image of the tailings pond can be a high-resolution remote sensing image.

[0059] S204, input the remote sensing image of the tailings pond into a pre-trained object detection model, and output the object detection result of the tailings pond through the object detection model.

[0060] Among them, the target detection result of the tailings pond can be a bounding box. The target detection model is an artificial intelligence model based on deep learning. Common ones include architectures based on convolutional neural networks (CNNs), such as Faster R-CNN, YOLO (You Only Look Once) series, SSD (Single Shot MultiBox Detector), etc. After being trained on a large number of annotated datasets, these models can automatically identify specific targets in images or videos and determine their positions (usually represented by bounding box coordinates) and category information, etc. In some embodiments of the present disclosure, the target detection model can be the YOLOV8 model.

[0061] Specifically, remote sensing images may exist in various formats (such as TIFF, GeoTIFF, etc.), and they need to be converted into a format that the target detection model can accept, generally a common image format, such as JPEG, PNG, etc. At the same time, it is necessary to ensure that the size of the image meets the requirements of the model input. If the size is too large or too small, scaling operations may be required. For example, a large-size image with high resolution can be scaled down proportionally to the input size specified by the model. Common input sizes can be 512×512 pixels, 1024×1024 pixels, etc. Normalize the pixel values of the image and map them to a specific numerical range, usually between 0 and 1 or -1 and 1. This helps the model converge faster and better, improving the training effect and the accuracy of subsequent detections. There are various normalization methods. For example, the mean and standard deviation of the image pixel values can be calculated, and then converted according to the corresponding formula. If the remote sensing image has geographic coordinate information, and the target detection model itself does not directly process geographic coordinates, it may be necessary to convert the geographic coordinates into image plane coordinates, or establish a correspondence between geographic coordinates and image coordinates in subsequent processing to facilitate accurate annotation and positioning of the location of the tailings pond in the actual geographical space.

[0062] The detection process of the object detection model includes: 1. Feature extraction: When the preprocessed remote sensing image of the tailings pond is input into the object detection model, the model first uses internal structures such as convolutional layers in the model to extract features from the image. For example, the convolutional layer slides convolutional kernels over the image to extract features of different levels such as textures and shapes. These features are continuously abstracted and integrated as the network deepens, forming high-level semantic feature maps valuable for subsequent object detection. In Faster R-CNN, features are extracted through a backbone network (such as ResNet as the feature extraction network); the YOLO series relies on its unique network structure to quickly extract features and perform object detection. 2. Object localization and classification: Based on the extracted features, the model further performs object localization and classification operations. Different models adopt different strategies. For example, Faster R-CNN uses a Region Proposal Network (RPN) to first generate a series of candidate regions that may contain objects, and then classifies these candidate regions (judging whether it is a tailings pond or other objects) and makes more precise position adjustments (refining the coordinates of the bounding box); while the YOLO series divides the image into multiple grids, and each grid directly predicts whether an object exists, its position, category, and other information, and can obtain the detection results of all objects through a single forward propagation, with relatively faster speed.

[0063] 3. Output detection results: Finally, the model outputs the object detection results of the tailings pond, and its content usually includes position information: presenting the position of the tailings pond in the remote sensing image in the form of a bounding box. The bounding box is generally represented by the coordinates of the upper left corner and the lower right corner (such as pixel coordinates in the image coordinate system). Through these coordinates, the specific range of the tailings pond in the image can be intuitively seen.

[0064] S206, classify multiple types of tailings pond elements contained in the object detection results of the tailings pond by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements.

[0065] Among them, the superpixel segmentation algorithm aims to divide an image into a series of small regions with similar features, which are called superpixels. Compared with traditional pixel-based processing, superpixels can greatly reduce the data volume while retaining the main contours and structural information of objects in the image. It is based on features such as the color, texture, and brightness of the image, and aggregates adjacent and similar pixels together to form superpixels through clustering and other methods. Common superpixel segmentation algorithms include the SLIC (Simple Linear Iterative Clustering) algorithm, etc. Random forest is an ensemble learning algorithm based on decision trees, which consists of multiple decision trees. In the training stage, multiple decision trees are constructed by randomly sampling the training dataset with replacement. Each decision tree learns and grows based on the sampled data. When splitting nodes, a part of the features are randomly selected from many features to determine the best splitting method. In the prediction stage, for a new data sample, each decision tree will give a prediction result, and then the opinions of multiple decision trees are integrated by voting (for classification tasks) or averaging (for regression tasks) to obtain the final prediction result. Random forest has strong anti-overfitting ability, is not sensitive to noisy data, and can handle high-dimensional data, etc., and is suitable for dealing with complex classification problems, such as classifying multi-category situations like tailings pond elements. Tailings pond elements can include: embankment dams, dry beaches, water bodies, etc.

[0066] Specifically, the remote sensing image containing the tailings pond output by the object detection model (that is, the image corresponding to the tailings pond object detection result) is segmented according to the superpixel segmentation algorithm, and the image is divided into multiple superpixel regions. The pixels within each superpixel region have high similarity in terms of color, texture, etc., and different superpixel regions can correspond to different tailings pond elements or different parts of the same element. For example, for a tailings pond image, different superpixel regions corresponding to the tailings accumulation area, dam body, surrounding vegetation of the reservoir area, etc. may be segmented, and these regions will be used as the basic units for subsequent classification.

[0067] Feature extraction: For each superpixel region, corresponding features are extracted. These features can include color features (such as average color value, color histogram, etc.), texture features (such as gray-level co-occurrence matrix features, local binary pattern features, etc.), geometric features (such as the area, perimeter, shape factor of the region, etc.), and spatial location features (such as the coordinate position of the superpixel region in the image, the relative position relationship with other regions, etc.). Through these rich feature descriptions, different tailings pond elements can be better distinguished, providing a basis for subsequent classification using the random forest algorithm. Collect a large number of labeled sample data of tailings pond elements (which can be manually labeled by professionals or labeled using other reliable methods). These sample data contain various features extracted from the superpixel regions before and the true categories of the corresponding tailings pond elements (such as labeled as tailings accumulation area, dam body, drainage facilities, surrounding vegetation, etc.). Divide these sample data into a training set and a validation set (the commonly used division ratio can be 7:3 or 8:2, etc.). The training set is used to train the random forest model, and the validation set is used to evaluate the performance of the model and adjust the model parameters (such as the number of decision trees, the maximum depth of each tree, etc.). Model training: Input the sample feature data in the training set into the random forest algorithm for training, enabling the model to learn the corresponding relationship between different tailings pond element features and categories. During the training process, the random forest will construct multiple decision trees according to the set parameters. Each decision tree continuously adjusts its own structure and node splitting rules to classify the training samples as accurately as possible. For example, a decision tree may determine that it is more likely to belong to the tailings accumulation area category based on the texture and color features of the superpixel region. As the training progresses, the classification ability of each decision tree is continuously improved, and the comprehensive classification performance of the entire random forest model is gradually enhanced. Input the superpixel region feature data in the tailings pond target detection result image after preprocessing (superpixel segmentation and feature extraction) into the trained and evaluated and optimized random forest model. The model will predict the tailings pond element category to which each superpixel region belongs, and finally obtain the classification results of multiple types of tailings pond elements, such as clearly indicating which regions in the image belong to the tailings accumulation area and which are the dam body parts.

[0068] S208, merge each type of the same tailings pond element, determine the reference line corresponding to the tailings pond element according to the merging result, and obtain the dry beach length based on the reference line corresponding to the tailings pond element.

[0069] Among them, the reference line can include: baseline, outer edge line, etc. The dry beach refers to a section of the slope area on the downstream slope of the tailings accumulation dam in the tailings pond where the beach surface changes from wet to dry. The dry beach length is an important indicator to measure the safety status of the tailings pond, which reflects the safety-related characteristics such as the buffering ability of the tailings pond in special situations such as floods.

[0070] Specifically, after obtaining the classification results of various types of tailings pond elements through the previous superpixel segmentation algorithm and random forest algorithm, it is necessary to first identify the tailings pond elements of the same category. For example, for multiple superpixel regions classified as tailings accumulation areas (these regions have been determined to belong to the same type of element, i.e., tailings accumulation areas, in the previous classification operation), they need to be integrated together to form a relatively complete representation of the tailings accumulation area. This process can be achieved by analyzing and processing information such as the coordinate ranges and boundaries of each region. For example, the boundaries of adjacent superpixel regions belonging to the same category are fused to construct a continuous region, so as to determine the reference line based on these merged regions of the same category of elements subsequently. The reference is used to calculate the dry beach length. The calculation of the dry beach length is based on the baseline, and the distance from the baseline to the outer edge of the tailings accumulation area in a specific direction (usually along the downstream slope direction of the dam) is measured.

[0071] In the above method for identifying the dry beach length of a tailings pond, by obtaining the remote sensing image of the tailings pond and processing it using a pre-trained object detection model, it is possible to accurately locate the tailings pond-related targets at the macroscopic remote sensing image level, avoiding problems such as limited vision and low efficiency that may exist in manual on-site exploration. For example, for a tailings pond area located in a remote and large-scale area, relying on the remote sensing image combined with the object detection model can quickly and relatively accurately determine the specific location and approximate range of the tailings pond in the image, improving the comprehensiveness and accuracy of the tailings pond target detection. This model-based automated detection method can process a large amount of remote sensing image data at the same time, and is suitable for monitoring multiple tailings ponds or large areas containing tailings ponds, which helps to realize the extraction of tailings pond information on a large scale and in batches. Using the superpixel segmentation algorithm and the random forest algorithm to classify various types of tailings pond elements contained in the detected tailings pond target detection results can further refine the understanding of different elements inside the tailings pond, and can more deeply understand the details of the tailings pond, thereby improving the accuracy of dry beach identification. Merging each type of tailings pond element of the same category, and then determining the reference line corresponding to the tailings pond element according to the merging result to obtain the dry beach length. The dry beach length is a key indicator for measuring the safety status of the tailings pond. Accurately mastering the dry beach length can help relevant personnel timely judge whether the tailings pond is in a safe operating state. For example, a too short dry beach length may indicate safety risks such as overtopping of the tailings pond. And through this method, accurate calculation of the dry beach length provides reliable data support for the safety warning and risk prevention and control of the tailings pond. The above process can be repeated based on the remotely sensed images obtained regularly to continuously monitor the change of the dry beach length, realize the dynamic tracking of the safety status of the tailings pond, and facilitate the timely adoption of corresponding maintenance, treatment and other measures to ensure the safety of the environment around the tailings pond and the lives and property of the people.

[0072] In one embodiment, as Figure 3As shown, the superpixel segmentation algorithm and the random forest algorithm are used to classify multiple types of tailings pond elements contained in the tailings pond target detection result, obtaining multiple types of tailings pond elements, including:

[0073] S302, endow the tailings pond target detection result with a coordinate reference system to obtain a target box vector.

[0074] Specifically, the tailings pond target detection result can be endowed with a coordinate reference system to obtain a target box vector in shp format.

[0075] In some exemplary embodiments, first, an appropriate coordinate reference system needs to be selected according to the geographical location of the tailings pond and the specific requirements of the project. For example, common geographic coordinate systems include WGS84 (World Geodetic System 1984, widely used in scenarios such as global positioning); in terms of projected coordinate systems, the Gauss-Krüger projected coordinate system is often used in the drawing of large-scale maps in China. If the relevant monitoring and management work of the tailings pond needs to match and integrate with the existing local geographic information data (such as regional maps, geographic information of surrounding infrastructure, etc.), then the corresponding coordinate reference system should be preferentially selected to ensure data consistency and compatibility. In some embodiments of the present disclosure, no restrictions are imposed on how to determine the specific coordinate reference system. After determining the specific coordinate reference system, the corresponding parameters of this coordinate system need to be obtained. For example, for a projected coordinate system, parameters such as its projection method (such as conformal projection, equal-area projection, etc.), central meridian, projection zone, etc. need to be determined. These parameters will be used for subsequent conversion and correction of the coordinates in the target detection results, so that they accurately fall under the selected coordinate reference system. The target detection results of the tailings pond output by the target detection model are usually presented in a certain specific format, which may be a text format (such as a CSV file containing information such as target categories and position coordinates, with each row recording the information related to a detected target), a JSON format (storing target-related attributes and coordinates and other data in the form of key-value pairs), or a custom format, etc. It is necessary to use the corresponding parsing program (which can use programming languages such as Python) to read and understand these data, and extract the coordinate information corresponding to each target box (for example, the position of a rectangular target box is usually represented by the upper-left corner coordinates and the lower-right corner coordinates, or represented by the center point coordinates and width and height, etc. in different forms). Convert the pixel coordinates in the image coordinate system extracted from the target detection results into the actual geographic coordinates under the selected coordinate reference system. By establishing a mapping relationship between the pixel coordinates and the geographic coordinates (common methods include mathematical models based on affine transformation, etc., constructing a transformation matrix according to the known image corner coordinates and corresponding pixel coordinates, and then using this matrix to convert the target box pixel coordinates into geographic coordinates), the coordinate conversion is realized, so that the coordinates of the target box have the meaning of actual geographical location. Use the programming libraries related to the Geographic Information System (GIS) (such as the Geopandas library in Python, which is a library based on Pandas and expands the geographical spatial data processing ability, or the GDAL / OGR library, which is a powerful geographical data processing library) to create a new vector dataset object, and define its data type as polygon (because the target box can usually be regarded as a rectangular polygon) or other appropriate geometric types such as points and lines (if there are special geometric representation requirements in the target detection results).Set the corresponding fields in this vector dataset object (for example, add a "target category" field to record whether the detected tailings pond target belongs to the dam body, dry beach, or other categories, and other descriptive fields such as "target number" can also be added for subsequent data management and query). Add the target boxes with converted coordinates to the created vector dataset one by one. According to the set geometric type (if it is a polygon, construct a polygon object with the four corner coordinates of the converted target box) and field information, accurately input the relevant attribute and geometric shape information of each target into the vector dataset to form a complete vector data content containing tailings pond target information. Use the functional functions provided by the relevant GIS library to save the constructed vector dataset as a shp format file. For example, in Geopandas, through a simple "to_file" method, specify the save path and file name suffix as ".shp" to complete the output operation, and finally obtain a vector file of the tailings pond target box stored in shp format with a coordinate reference system, which is convenient for subsequent visualization display, spatial analysis, and integration with other geographic information data in GIS software (such as ArcGIS, QGIS, etc.).

[0076] S304, based on a preset buffer distance, expand the target box vector to obtain an expanded target box vector.

[0077] Among them, the setting of the buffer distance will be differentiated to a certain extent according to the scale of the detected tailings pond target detection result. For example, for a tailings pond with a scale of 10000 km 2 a buffer range of 10 m is set; for a tailings pond with a scale of 50000 km 2 a buffer range of 50 m is set; the purpose is to avoid the border not precisely covering the tailings pond and extend it outward by a certain range to ensure that all elements are included. Those skilled in the art can flexibly adjust the buffer distance according to the size of the tailings pond target detection result, and the size of the buffer distance is not limited in some embodiments of the present disclosure.

[0078] Specifically, a buffer distance can be set for the target box vector, and after expansion using the buffer distance, an expanded target box vector is obtained.

[0079] S306, determine the seed center point in the expanded target box vector, and through a clustering method, merge the same-class pixels based on the seed point center to form superpixels.

[0080] Among them, the seed center points refer to some representative points pre-selected within a specific area or determined through a certain algorithm. They usually serve as the starting reference points or core points for subsequent processing (such as clustering, region growing, etc.). Operations such as classification, merging, and expansion are performed on the surrounding pixels, objects, or data elements around these seed center points, thereby forming regions, groups, etc. with specific characteristics or meeting specific requirements. A superpixel refers to a pixel block formed by clustering and merging pixel points with similar features (such as color, texture, brightness, etc.) in an image. It can be regarded as a kind of regional division of the original image at the pixel level, replacing numerous original pixels with relatively few "superpixel blocks" to simplify the representation of the image while retaining most of the key visual information and structural features in the image.

[0081] Specifically, several points can be randomly and evenly selected as seed center points within the area covered by the extended target box vector according to certain rules. For example, the number of seed points to be selected can be set according to the size of the target box range (if the target box range is large and a more refined superpixel division is expected, more seed points can be selected; otherwise, fewer can be selected). Then, through programming (taking Python as an example), using the random number generation function combined with the coordinate range of the target box, the corresponding number of coordinate points are generated within this range as seed center points. The area corresponding to the extended target box vector can also be divided into regular grids, and then the center points of each grid or points at specific positions are selected as seed center points. For example, the target box area is divided into small square grids at a fixed interval (such as dividing into one grid every 10 meters, and the interval size can be adjusted according to actual needs and data precision), and the center coordinate points of the grids are taken as seed points. Then, a suitable clustering method can be selected, such as K-Means clustering or other clustering methods, and clustering is performed with the seed center points as the clustering centers. In the scenario of forming superpixels, pixel points can be regarded as data points to be clustered. By calculating the distance between the pixel and the seed center point (usually using distance metrics such as Euclidean distance), the cluster (i.e., superpixel) to which the pixel belongs is continuously adjusted, and finally the merging of similar pixels is achieved.

[0082] In addition, a superpixel segmentation algorithm, such as the SLIC (Simple Linear Iterative Clustering) algorithm, can also be used for processing. SLIC is an algorithm specifically for image superpixel segmentation that combines spatial proximity and color similarity to divide superpixels. At the beginning of the algorithm, it is similar to determining the seed center points mentioned above (however, the SLIC algorithm has its specific seed point initialization method, usually uniformly spaced on the pixel grid of the image and fine-tuned according to features such as the color of the pixels). Then, with these seed points as the center, within a certain search range (usually a square area with a fixed side length, and the side length is related to the desired superpixel size), considering both the color distance and spatial distance of the pixels, the superpixel regions corresponding to each seed point are iteratively optimized, making the pixels within the superpixel as similar as possible in color and space. Eventually, superpixel segmentation is achieved, and superpixels are obtained.

[0083] S308, detect the edges of the superpixels, and through centerline extraction and line-to-surface conversion, obtain the superpixel segmentation surface.

[0084] Among them, the algorithms for edge detection can include: Canny edge detection algorithm, Laplacian operator, Sobel operator and other algorithms. The superpixel segmentation surface refers to a way of expressing each superpixel region in a planar geometric form after dividing the image into multiple superpixels. It is a form presented after performing superpixel segmentation on the image through certain processing (such as a series of operations like superpixel edge detection, centerline extraction, and line-to-surface conversion), making the originally discrete superpixels presented in a continuous surface form in space. Each surface represents a set of pixels with similar features (such as color, texture, etc.), that is, a superpixel.

[0085] Specifically, an edge detection algorithm can be used to detect the edges of the superpixels. The detected superpixel edges can be regarded as a region with a certain width. A thinning algorithm or a skeletonization algorithm can be used to obtain the centerline. Then, the centerline is converted into a surface using polygon construction methods or by utilizing the functions of GIS software or related libraries to obtain the superpixel segmentation surface.

[0086] In some exemplary embodiments, the thinning algorithm aims to gradually "thin" the edges by continuously removing redundant pixel points in the edge regions, eventually leaving a single-pixel-wide centerline to represent the original edge. For example, the Zhang-Suen thinning algorithm is a classic thinning algorithm. It is based on the neighborhood information of pixel points in the image (such as the connectivity of pixels within the 8-neighborhood, the distribution of black and white pixels, etc.). Through multiple iterations, it gradually deletes the removable pixel points according to certain rules until no more deletion is possible, thus obtaining the thinned centerline. The skeletonization algorithm is similar to the thinning algorithm. The goal of skeletonization is also to extract a single-pixel-wide centerline that can represent the shape of the object, but its idea is more from the perspective of maintaining the topological structure and shape characteristics of the object. For example, in the skeletonization method based on distance transformation, first, the distance from each pixel in the image to the background (such as the Euclidean distance, etc.) is calculated, and then the pixel points located at the center position of the object are determined according to the distance information, gradually constructing the skeleton of the object, that is, the centerline.

[0087] Method using polygon construction: For the extracted centerline, it can be regarded as the boundary line or skeleton line of a polygon, and the area enclosed by these lines is constructed into a polygon surface through a certain algorithm. A common approach is to trace along the centerline, determine the vertices of the polygon at key positions such as the turning points and endpoints of the centerline, and then connect these vertices in a certain order to form a closed polygon, thereby obtaining the superpixel segmentation surface. For example, starting from an endpoint of the centerline, moving along the direction of the line, whenever a point where the direction changes (judged by calculating the slope change of adjacent line segments, etc.) is encountered, it is recorded as a vertex, and so on until returning to the starting endpoint or being able to close and connect with other already constructed polygons, finally forming polygon surfaces to represent the superpixel segmentation surface. Additionally, if the superpixel-related data has geographical coordinate information (such as in scenarios like geographical remote sensing image analysis), the functions of geographical information system (GIS) software (such as ArcGIS, QGIS, etc.) or related geospatial data processing libraries (such as Geopandas, GDAL / OGR in Python, etc.) can be used to achieve the line-to-surface operation. For example, in ArcGIS, after importing the centerline vector data (usually in the.shp format), through the "Feature to Polygon" tool in the "Features" module under the "Data Management Tools", setting parameters such as the input centerline feature layer and the output polygon feature layer path according to the operation process prompted by the software, clicking the "OK" button can complete the line-to-surface operation, generating the corresponding superpixel segmentation surface layer, and it can be viewed, edited, and integrated with other geographical information data for analysis, etc. in the software.

[0088] S310. Classify the superpixel segmentation surface using the random forest algorithm to obtain multiple types of tailings pond elements.

[0089] Specifically, features in the superpixel segmentation surface can be extracted, and then the unclassified superpixel segmentation surface can be classified using a pre-trained random forest model.

[0090] In this embodiment, the target detection results of the tailings pond are assigned to a coordinate reference system to obtain a target box vector, so that the detected tailings pond-related targets have clear and accurate position information in the geographical space. This is very crucial for subsequent various geographical space analyses and integration with other geographical information data (such as surrounding terrain, infrastructure distribution, etc.). For example, when conducting regional planning and environmental impact assessment, it is possible to accurately know the specific location of the tailings pond within the entire region, facilitating the precise measurement of its relationship with surrounding elements. Expand the target box vector based on a preset buffer distance to obtain an expanded target box vector, which can simulate the impact area within a certain range around the tailings pond. For example, when considering the potential impact of the tailings pond on the surrounding ecological environment and residents' safety, by reasonably setting the buffer distance (such as determined according to the pollutant diffusion model, safety protection standards, etc.), the scope that needs to be focused on and analyzed can be clearly defined, helping to formulate corresponding prevention and treatment measures in advance and providing an intuitive and quantitative spatial basis for risk management. Determine the seed center points in the expanded target box vector, and then merge similar pixels into superpixels through a clustering method, greatly simplifying the representation form of image data. The originally large number of pixels are aggregated into a relatively small number of superpixels, reducing the complexity of subsequent data processing and analysis, while retaining the main visual features and structural information in the image. For example, when analyzing a large-area tailings pond area, if directly processing a large amount of original pixel data, the computational cost and storage cost will be very high, while the formation of superpixels can improve the processing efficiency and make the analysis more feasible on the premise of ensuring key information. As a relatively stable region with certain semantic information, the pixels within a superpixel are similar, facilitating the extraction of more representative and robust features. For example, extracting features such as color, texture, and geometry of the superpixels to describe the characteristics of different regions of the tailings pond. Compared with feature extraction based on individual pixels, these superpixel-level features can better reflect the essential differences between different elements in the tailings pond (such as the dam body, dry beach, water accumulation area, etc.), providing a high-quality feature basis for subsequent classification and other operations.

[0091] In one embodiment, as Figure 4 shown, the classification of the superpixel segmentation surface using the random forest algorithm to obtain multiple types of tailings pond elements includes:

[0092] S402. Based on the pre-determined classification samples, determine the element features of each type of tailings pond element.

[0093] Among them, the classified samples mainly include the sample point data (vector format) of the tailings pond stacking dam, dry beach, water body and other elements. The classified samples can be obtained by pre-annotation. For example, there is an image. Through superpixel segmentation, a superpixel segmentation surface is obtained. Then, according to visual inspection by humans, some sample points are selected on the surfaces of elements such as water, stacking dam, and dry beach, and the categories are marked, so as to obtain classified samples for subsequent classification algorithms.

[0094] Specifically, the element features of each type of tailings pond element can be determined according to the respective features in the pre-determined classified samples.

[0095] S404, using the random forest algorithm to learn the element features to form a superpixel classification model.

[0096] Specifically, the random forest algorithm, element features, and each type of tailings pond corresponding to the element features can be used to train the model to obtain a superpixel classification model.

[0097] In some exemplary embodiments, the Scikit-learn library of Python is used to build a random forest model. First, the corresponding modules are imported and a random forest classifier object is created. At the same time, some key parameters of the model can be set, such as the number of decision trees (n_estimators), the maximum depth of each tree (max_depth), the minimum number of samples for splitting (min_samples_split), etc. The initial values of the parameters can be set according to experience first, and then optimized through cross-validation and other methods later.

[0098] S406, using the superpixel classification model to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

[0099] Among them, the tailings pond elements can be represented by element vector pixel blocks in some embodiments of the present disclosure.

[0100] Specifically, the superpixel classification model can be used to classify the superpixel segmentation surface to finally obtain the element vector pixel blocks within the range of each tailings pond.

[0101] In this embodiment, using the formed superpixel classification model to classify the superpixel segmentation surface, multiple types of tailings pond elements can finally be obtained. Each part such as the tailings dam, dry beach, water accumulation area, and vegetation around the pond can be clearly distinguished, and then the area changes, spatial distribution changes, etc. of them can be grasped in real time, ensuring the accuracy of dry beach identification.

[0102] In one embodiment, such as Figure 5As shown, the tailings pond elements include: the embankment dam element, the dry beach element, and the water body element; merging each type of the same tailings pond elements, determining the reference line corresponding to the tailings pond element according to the merging result, and obtaining the dry beach length based on the reference line corresponding to the tailings pond element, including:

[0103] S502, merging the same tailings pond elements in the multiple types of tailings pond elements to obtain the merging result of each type of tailings pond element.

[0104] S504, processing the merging result of the embankment dam element by using a curve simplification algorithm to obtain the initial dam baseline.

[0105] S506, determining the outer water boundary line of the water body element, and determining the dry beach length based on the distance between the initial dam baseline and the outer water boundary line.

[0106] Among them, the initial dam baseline generally refers to a representative curve obtained by simplifying the geometric shape of the tailings pond embankment dam. It can reflect the key geometric features such as the trend and contour of the main body of the embankment dam, and can be regarded as a line extracted from the complex actual boundary line of the embankment dam to reflect its core shape, and is approximately similar to the bottom contour line of the embankment dam or the foundation contour line at the key position to a certain extent. The outer water boundary line is essentially a definition of the water body range, which outlines the boundary line between the water body and different elements such as the surrounding tailings dam, dry beach, and reservoir perimeter land in the form of a line. From a geometric perspective, if the water body is regarded as a planar region (usually represented by a polygon to indicate its range), then the outer water boundary line is the boundary contour line of this polygon, which clarifies the distribution range of the water body in the two-dimensional plane, just like drawing a "boundary line" for the water body area to clearly distinguish it in space.

[0107] Specifically, based on the category field information of multiple types of tailings pond elements, perform same-element merging to obtain the merging result of each type of tailings pond element (the final vector surface of each element); process the embankment dam vector (embankment dam element) through the Douglas–Peucker algorithm to obtain the initial dam baseline vector; traverse the distance between the initial dam baseline vector and the outer water boundary line of the water body element in the corresponding vertical direction, take the shortest distance and assign it to the corresponding dry beach vector attribute table, and finally obtain the dry beach length of each tailings pond.

[0108] In some exemplary embodiments, the data structures of multiple types of tailings pond elements can be determined. For the data structures representing tailings pond elements, they commonly exist in the vector form of polygons (such as areas representing tailings dams, dry beaches, water accumulation areas, etc.), points (such as certain specific monitoring points, etc.), or lines (such as the boundary lines of the dam body, etc.). Elements of the same type are merged. Then, the Douglas–Peucker algorithm or the Visvalingam-Whyatt algorithm is selected to process the merged result of the stacking dam elements, obtaining a simplified curve, that is, the initial dam baseline. If it is in GIS software (taking ArcGIS as an example): Load the vector data layer containing water body elements (such as water accumulation areas, etc.) into ArcGIS. The water body elements generally also exist in the form of polygons. Through the "Boundary" tool under the "Elements" module in the "Data Management Tools", select the water body element layer as the input element, specify the path and name of the output edge element layer (usually saved in the.shp format), click the "OK" button, and ArcGIS will extract the outer boundary lines of each water body polygon, that is, the outer water edge lines. These edge lines are also stored in the form of vector lines, which can be conveniently used for subsequent spatial analysis and distance calculation operations. Use the "Near Neighbor Analysis" tool under the "Distance Analysis" module in the "Analysis Tools" of ArcGIS. Take the initial dam baseline as the input element and the outer water edge lines as the near neighbor elements, set parameters such as the path and name of the output distance result table, click the "OK" button, and ArcGIS will calculate the nearest distance from the dam baseline to each outer water edge line and record the results in the output table. Then, by viewing the table data or further visualizing it in the software, find the distance value that meets the requirements (such as taking the minimum value because the dry beach length usually refers to the shortest distance from the dam body to the water body; or determining the value-taking method according to actual business rules and analysis requirements), which is the dry beach length.

[0109] If it is through programming (taking Python combined with the Geopandas library as an example): Assume that the vector data of the water body elements has been loaded into a GeoDataFrame object (such as water_gdf). By traversing each water body polygon, extract its boundary line. First, ensure that the initial dam baseline and the outer water edge lines are in the same coordinate reference system. Then, by traversing each line object of the outer water edge lines, calculate the minimum distance between it and the initial dam baseline.

[0110] In this embodiment, the same tailings pond elements among multiple types of tailings pond elements are merged, and the scattered elements of the same type are integrated into a unified geometric object, greatly simplifying the data structure. By merging elements of the same type, it is more convenient to extract the overall characteristics of each type of tailings pond element. Using a curve simplification algorithm to process the merging result of the stacking dam elements to obtain the initial dam baseline can highlight the key geometric features of the stacking dam. As a simplified representative curve, the initial dam baseline filters out some trivial detail information that may exist in the original dam body boundary data and has little impact on the overall shape, making the main contour and trend of the dam body clearer. By determining the outer water line of the water body element and based on the distance between the initial dam baseline and the outer water line, the dry beach length can be determined more accurately.

[0111] In one embodiment, as Figure 6 shown, the method further includes:

[0112] S602, obtaining training remote sensing images, where the training remote sensing images contain multiple tailings pond areas.

[0113] S604, determining the tailings pond areas contained in the training remote sensing images, and for the tailings pond areas, drawing a vector box and assigning values to obtain the vector boundary of the tailings pond target detection box.

[0114] S606, training a target detection model based on the vector boundary of the tailings pond target detection box and the training remote sensing images.

[0115] Specifically, cloudless high-resolution training remote sensing images of each time phase are selected, and the mining areas of the mining areas are mainly selected as the training remote sensing images. The resolution of the training remote sensing images is not less than 2m. Use ArcGIS software to create a new vector box surface vector for the tailings pond; for the tailings pond range, draw a vector box and assign an attribute table of 1 to obtain the vector boundary of the tailings pond target detection box. The center position of the target box can also be located through a sliding window algorithm and expanded to form a sample slice with a size of 1024*1024 pixels. Convert the longitude and latitude coordinates of the vector to pixel coordinates in the slice and save them as a txt file with the same name as the corresponding sample slice to obtain the tailings pond sample slice and the target box position information. Use the tailings pond sample slice and the target box position information to train the target detection model.

[0116] In one embodiment, the target detection model includes a backbone network, a neck network, and a head network, and the backbone network includes a rectangular feature extraction module. The backbone network includes: CSPDarknet (rectangular feature extraction module) consists of a standard convolutional layer (convolution with stride) and a pooling layer, which is used for preliminary feature extraction and downsampling. Each CSPBlock contains a Residual Network and a direct transmission path for combining the features of both. By introducing multi-scale feature fusion, it is possible to better capture targets of different sizes, and finally output the feature map to subsequent network layers for target detection or classification tasks. The neck network PAFPN is designed to enhance the detection ability for small and large targets through more efficient multi-scale feature fusion. It is mainly achieved through the following key parts: Feature Pyramid Network (FPN): used for top-down feature fusion to ensure that high-level feature information can be combined with low-level features; Path Aggregation Network (PANet): used for bottom-up feature fusion to further aggregate features at different scales and enhance the expressiveness of the model. At the same time, PAFPN introduces more convolutional and connection operations to ensure that information can be efficiently propagated at multiple scales. The head network is responsible for generating the final detection results. The detection head usually contains several parallel convolutional layers, which are used to process the feature map provided by the neck network and output the detection boxes and classification results on each feature map. In YOLOv8, the detection head adopts an Anchor-Free design, directly predicting the coordinates and class probabilities of the target boxes, which simplifies the complexity of the model and improves the detection speed. As Figure 7 shown, training the target detection model based on the vector boundary of the tailings pond target detection box and the training remote sensing image includes:

[0117] S702, based on the backbone network, extract the tailings pond feature information contained in the vector boundary of the tailings pond target detection box in the training remote sensing image.

[0118] Specifically, the backbone network can be first used to extract the tailings pond feature information contained in the vector boundary of the tailings pond target detection box in the training remote sensing image, and the tailings pond feature information is used for subsequent processing.

[0119] S704, based on the rectangular feature extraction module, perform secondary feature extraction on the tailings pond feature information to obtain high-order tailings pond feature information.

[0120] Specifically, since there may be some characteristic information in the tailings pond characteristic information with a large difference in the length-width ratio, for those with a small difference in length and width, the characteristic modules in the original backbone network can well extract the characteristic information. The rectangular characteristic extraction module is used to extract the characteristic information again on this basis to ensure that the characteristics contained in the characteristic information of the tailings pond with a large difference in length and width are recognized, so as to obtain the high-order characteristic information of the tailings pond.

[0121] S706. After dimensionality elevation of the high-order characteristic information of the tailings pond, it is fused with the characteristic information of the tailings pond, and the target detection model is trained using the fused characteristic information and the vector boundary of the target detection frame of the tailings pond.

[0122] Specifically, since the high-order characteristic information of the tailings pond is obtained by performing secondary feature extraction on the characteristic information of the tailings pond, therefore, the high-order characteristic information of the tailings pond and the characteristic information of the tailings pond have different dimensions, and it can be dimensionally elevated so that it has the same dimension as the characteristic information of the tailings pond. Then, the elevated features are added and normalized to obtain the final feature extraction result (the fused characteristic information). The feature extraction result is a part of the model training data, which can be understood as extracting parameters such as the shape and texture of the image and calculating them together with the target label. Then, the fused characteristic information and the vector boundary of the target detection frame of the tailings pond are used to train the neck network and the head network to obtain the finally trained target detection model.

[0123] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0124] Based on the same inventive concept, the embodiments of the present disclosure also provide a tailings pond dry beach length recognition device for implementing the above-mentioned tailings pond dry beach length recognition method. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the following tailings pond dry beach length recognition device can refer to the limitations on the tailings pond dry beach length recognition method in the above text, and will not be repeated here.

[0125] In one embodiment, as Figure 8As shown, a dry beach length recognition device 800 for a tailings pond is provided, including: an image acquisition module 802, a model processing module 804, a classification module 806, and a length calculation module 808, where:

[0126] The image acquisition module 802 is used to acquire remote sensing images of the tailings pond;

[0127] The model processing module 804 is used to input the remote sensing image of the tailings pond into a pre-trained object detection model, and output the object detection result of the tailings pond through the object detection model;

[0128] The classification module 806 is used to classify multiple types of tailings pond elements contained in the object detection result of the tailings pond by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements;

[0129] The length calculation module 808 is used to merge each type of the same tailings pond elements, determine the reference line corresponding to the tailings pond elements according to the merging result, and obtain the dry beach length based on the reference line corresponding to the tailings pond elements.

[0130] In an embodiment of the device, the classification module 806 includes:

[0131] The data processing module is used to assign a coordinate reference system to the object detection result of the tailings pond to obtain a target box vector;

[0132] The expansion module is used to expand the target box vector based on a preset buffer distance to obtain an expanded target box vector;

[0133] The merging module is used to determine seed center points in the expanded target box vector, and merge the same-class pixels based on the seed center points through a clustering method to form superpixels;

[0134] The segmentation module is used to detect the edges of the superpixels, and obtain a superpixel segmentation surface through centerline extraction and line-to-surface conversion;

[0135] The classification sub-module is used to classify the superpixel segmentation surface by using a random forest algorithm to obtain multiple types of tailings pond elements.

[0136] In an embodiment of the device, the classification sub-module is further used to determine the element features of each type of tailings pond element based on a pre-determined classification sample; use the random forest algorithm to learn the element features to form a superpixel classification model; use the superpixel classification model to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

[0137] In one embodiment of the device, the tailings pond elements include: a stacking dam element, a dry beach element, and a water body element; the length calculation module 808 is further configured to merge the same tailings pond elements among the multiple types of tailings pond elements to obtain a merged result for each type of tailings pond element; process the merged result of the stacking dam element by using a curve simplification algorithm to obtain an initial dam baseline; determine the outer water boundary of the water body element, and determine the dry beach length based on the distance between the initial dam baseline and the outer water boundary of the water body element.

[0138] In one embodiment of the device, the device further includes: a model training module, configured to obtain training remote sensing images, where the training remote sensing images contain multiple tailings pond areas; determine the tailings pond areas contained in the training remote sensing images, and for each tailings pond area, draw a vector box and assign a value to obtain a vector boundary of the tailings pond target detection box; train a target detection model based on the vector boundary of the tailings pond target detection box and the training remote sensing images.

[0139] In one embodiment of the device, the target detection model includes a backbone network, a neck network, and a head network, and the backbone network includes a rectangular feature extraction module; the model training module is further configured to extract, based on the backbone network, the tailings pond feature information contained within the vector boundary of the tailings pond target detection box in the training remote sensing images; perform secondary feature extraction on the tailings pond feature information by using the rectangular feature extraction module to obtain high-order tailings pond feature information; perform dimension elevation on the high-order tailings pond feature information and then fuse it with the tailings pond feature information, and train the target detection model by using the fused feature information and the vector boundary of the tailings pond target detection box.

[0140] Each module in the above-mentioned tailings pond dry beach length recognition device can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0141] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 9As shown in the figure. The computer device includes a processor, a memory, and a network interface connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store remote sensing image data of the tailings pond. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it implements a method for identifying the dry beach length of a tailings pond.

[0142] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 10 shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for identifying the dry beach length of a tailings pond. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0143] Those skilled in the art can understand that Figure 9 the structures shown in FIGS. 9 or 10 are merely block diagrams of some structures related to the present disclosure solution, and do not constitute a limitation on the computer device to which the present disclosure solution is applied. The specific computer device may include more or fewer components than those shown in the figures, or combine certain components, or have different component arrangements.

[0144] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the steps in the above-mentioned method embodiments.

[0145] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the steps in the above-mentioned method embodiments.

[0146] In one embodiment, a computer program product is provided, including a computer program which, when executed by a processor, implements the steps in the above method embodiments.

[0147] It should be noted that the remote sensing images of the tailings reservoir involved in the present disclosure (including but not limited to the remote sensing images of the tailings reservoir for training and the identified remote sensing images of the tailings reservoir) are all information and data authorized by the user or fully authorized by all parties.

[0148] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, a database, or other media used in the embodiments provided in the present disclosure can include at least one of non-volatile and volatile memories. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present disclosure can include at least one of relational databases and non-relational databases. The non-relational databases can include distributed databases based on blockchain, etc., and are not limited thereto. The processors involved in the embodiments provided in the present disclosure can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., and are not limited thereto.

[0149] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0150] The above-described embodiments merely represent several implementation manners of the present disclosure. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present disclosure. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present disclosure, several modifications and improvements can still be made, and these all belong to the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure shall be subject to the appended claims.

Claims

1. A method for identifying the dry beach length of a tailings pond, characterized in that: The method comprises: Obtain remote sensing images of tailings ponds; Inputting the tailings pond remote sensing image into a pre-trained target detection model, and outputting the tailings pond target detection result via the target detection model; A superpixel segmentation algorithm and a random forest algorithm are used to classify multiple types of tailings pond elements contained in the tailings pond target detection result to obtain multiple types of tailings pond elements; Merging the same tailings pond elements of each type, determining the reference line corresponding to the tailings pond element according to the merging result, and obtaining the dry beach length based on the reference line corresponding to the tailings pond element; wherein The tailings pond elements include: a dam element, a dry beach element, and a water body element; the same tailings pond elements of each type are merged, and a reference line corresponding to the tailings pond element is determined according to the merged result. Based on the reference line corresponding to the tailings pond element, the dry beach length is obtained, including: Merging the same tailings pond elements among the multiple types of tailings pond elements to obtain a merged result of each tailings pond element; The combined result of the accumulation dam elements is processed by using a curve simplification algorithm to obtain an initial dam baseline; The outer boundary of the water body element is determined, and the length of the dry beach is determined based on the distance between the initial dam baseline and the outer boundary of the water body element.

2. The method according to claim 1, characterized in that The superpixel segmentation algorithm and the random forest algorithm are used to classify the multiple types of tailings pond elements contained in the tailings pond target detection result to obtain multiple types of tailings pond elements, including: Assigning the tailings pond target detection result to a coordinate reference system to obtain a target frame vector; Based on a preset buffer distance, the target frame vector is expanded to obtain an expanded target frame vector; Determine a seed center point in the expanded target frame vector, and merge similar pixels based on the seed center point through a clustering method to form a superpixel; Detecting the edge of the superpixel, and obtaining the superpixel segmentation surface by extracting the center line and converting the line to the surface; The random forest algorithm is used to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

3. The method according to claim 2, characterized in that The random forest algorithm is used to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements, including: Determine the element characteristics of each type of tailings pond element based on the predetermined classification samples; Using the random forest algorithm to learn the feature features and form a superpixel classification model; The superpixel classification model is used to classify the superpixel segmentation surface to obtain multiple types of tailings pond elements.

4. The method according to claim 1, characterized in that: The method further comprises: Acquire a training remote sensing image, wherein the training remote sensing image contains a plurality of tailings pond areas; Determine the tailings pond area contained in the training remote sensing image, draw a vector frame and assign values ​​to the tailings pond area, and obtain a vector boundary of the tailings pond target detection frame; A target detection model is trained based on the tailings pond target detection box vector boundary and the training remote sensing image.

5. The method according to claim 4, characterized in that The target detection model includes a backbone network, a neck network and a head network, and the backbone network includes a rectangular feature extraction module; The training of the target detection model based on the tailings pond target detection box vector boundary and the training remote sensing image includes: Based on the backbone network, extracting the tailings pond feature information contained in the tailings pond target detection frame vector boundary in the training remote sensing image; Performing secondary feature extraction on the tailings pond feature information based on the rectangular feature extraction module to obtain high-order feature information of the tailings pond; The high-order feature information of the tailings pond is fused with the feature information of the tailings pond after dimension upgrading, and the target detection model is trained using the fused feature information and the vector boundary of the tailings pond target detection box.

6. A tailings pond dry beach length identification device, characterized in that: The device comprises: Image acquisition module, used to obtain remote sensing images of tailings ponds; A model processing module, used for inputting the tailings pond remote sensing image into a pre-trained target detection model, and outputting the tailings pond target detection result via the target detection model; A classification module is used to classify multiple types of tailings pond elements contained in the tailings pond target detection result by using a superpixel segmentation algorithm and a random forest algorithm to obtain multiple types of tailings pond elements; The length calculation module is used to merge the same tailings pond elements of each type, determine the reference line corresponding to the tailings pond element according to the merged result, and obtain the dry beach length based on the reference line corresponding to the tailings pond element; wherein The tailings pond elements include: a dam element, a dry beach element, and a water body element; the same tailings pond elements of each type are merged, and a reference line corresponding to the tailings pond element is determined according to the merged result. Based on the reference line corresponding to the tailings pond element, the dry beach length is obtained, including: Merging the same tailings pond elements among the multiple types of tailings pond elements to obtain a merged result of each tailings pond element; The combined result of the accumulation dam elements is processed by using a curve simplification algorithm to obtain an initial dam baseline; The outer boundary of the water body element is determined, and the length of the dry beach is determined based on the distance between the initial dam baseline and the outer boundary of the water body element.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images