Geological disaster automatic identification system and method based on multi-source remote sensing data

Abstract

The invention discloses an automatic geological disaster identification system and method based on multi-source remote sensing data, and relates to the field of geological disaster identification, and the method comprises the steps: firstly collecting an optical remote sensing image, an SAR radar image and LIDAR point cloud data before and after the geological disaster of a target area, and carrying out the preprocessing, and obtaining a standard image, a standard SAR and a standard point cloud; comparing and analyzing the standard images before and after the disaster to determine a disaster area, and analyzing the standard images, the standard SAR and the standard point cloud of the disaster area to obtain a vegetation comprehensive index, an SAR index and a point cloud index; training a convolutional neural network model by using historical disaster data to obtain a disaster recognition model; according to the method, multi-source remote sensing data are fused, so that the defect of a single data source is overcome, the geological disaster type can be accurately identified, and a key premise is provided for treatment work.

Description

Technical Field

[0001] The present invention relates to the field of geological disaster identification, and specifically to an automatic geological disaster identification system and method based on multi-source remote sensing data. Background Art

[0002] Traditional geological disaster identification mainly relies on methods such as manual field surveys and topographic mapping. This not only consumes a large amount of manpower, material resources, and time, but is also restricted by natural conditions such as terrain and climate. In complex environments such as high mountains, canyons, and remote areas, the data collection efficiency is low, the coverage is limited, and it is difficult to achieve real-time dynamic monitoring. Moreover, single remote sensing data has certain defects. For example, although optical remote sensing can obtain large-area surface information, it is easily affected by weather and lighting conditions, and the imaging quality is poor in cases such as clouds, fog, and at night; while synthetic aperture radar (SAR) remote sensing has the ability to work all day and all weather, but its ability to extract features of small terrain changes is insufficient, making it difficult to meet the requirements of high-precision identification;

[0003] How to fuse multi-source remote sensing data to achieve accurate identification of disaster types is the problem we need to solve. For this reason, an automatic geological disaster identification system and method based on multi-source remote sensing data are provided. Summary of the Invention

[0004] The purpose of the present invention is to provide an automatic geological disaster identification system and method based on multi-source remote sensing data.

[0005] The purpose of the present invention can be achieved through the following technical solutions: An automatic geological disaster identification method based on multi-source remote sensing data includes:

[0006] Collect remote sensing data of the target area before and after the occurrence of a geological disaster. The remote sensing data includes: optical remote sensing images, SAR radar images, and LIDAR point cloud data;

[0007] Preprocess the remote sensing data to obtain the standard remote sensing data of the target area before and after the occurrence of a geological disaster. The standard remote sensing data includes: standard images, standard SAR, and standard point clouds;

[0008] Compare and analyze the standard images of the target area before and after the disaster occurrence to obtain the disaster area of the target area, and respectively analyze the standard images, standard SAR, and standard point clouds corresponding to the disaster area to obtain the vegetation comprehensive index, SAR index, and point cloud index;

[0009] Obtain the disaster types of each historical disaster and their corresponding vegetation comprehensive index, SAR index, and point cloud index, and use the vegetation comprehensive index, SAR index, and point cloud index to train a convolutional neural network model to obtain a disaster recognition model;

[0010] Obtain the disaster type of the target area through the vegetation comprehensive index, SAR index, point cloud index, and disaster recognition model of the target area.

[0011] Preferably, the process of preprocessing remote sensing data to obtain the standard remote sensing data of the target area before and after the occurrence of geological disasters is as follows:

[0012] Perform radiometric calibration, geometric correction, noise removal, and image enhancement on the optical remote sensing image and SAR radar image respectively to obtain the standard image and standard SAR;

[0013] Perform format conversion, coordinate conversion, noise removal, and resampling on the LIDAR point cloud data to obtain the standard point cloud.

[0014] Preferably, the process of analyzing the standard image of the disaster area to obtain the vegetation comprehensive index is as follows:

[0015] Divide the disaster area into several sub-areas, number each sub-area, and then perform a secondary division on each sub-area to obtain each grandson area, and obtain the vegetation index on each grandson area;

[0016] Preset a vegetation index threshold, and screen the grandson areas based on the comparison result between the vegetation index threshold and the vegetation index of each grandson area to obtain high-value areas;

[0017] Obtain the high-value frequency value of each sub-area through the number of high-value areas in each sub-area, and process the high-value frequency value of each sub-area through the spatial entropy formula to obtain the chaos degree of the target area;

[0018] Obtain the minimum bounding rectangle of the disaster area, obtain the length and width of this rectangle to obtain the aspect ratio, and obtain the vegetation comprehensive index of the target area through the aspect ratio and the chaos degree.

[0019] Preferably, the process of analyzing the standard SAR of the disaster area to obtain the SAR index is as follows:

[0020] Obtain the gray-level co-occurrence matrix of the standard SAR at different angles, and obtain the correlation of the gray-level co-occurrence matrix at different angles;

[0021] Take the standard deviation and mean of the correlation of the gray-level co-occurrence matrix at each angle as the angular wave value and angular mean respectively;

[0022] Perform fusion processing on the angular wave value and angular mean to obtain the SAR index.

[0023] Preferably, the process of analyzing the standard point cloud of the disaster area to obtain the point cloud index is as follows:

[0024] Construct digital elevation models before and after the disaster using the standard point clouds before and after the disaster. Denote the digital elevation model before the disaster as the pre-disaster model and the digital elevation model after the disaster as the post-disaster model;

[0025] Randomly select two points on the pre-disaster model and randomly generate multiple paths between the two points. At the same time, select the same two points on the post-disaster model and generate the same paths;

[0026] Equidistantly select points on each path as marked points, obtain the elevation differences at each marked point on each path, and get the elevation set corresponding to each path;

[0027] Based on the number of positive and negative values in the elevation set corresponding to each path, obtain the elevation ratio, and take the mean of the elevation ratios of all paths as the elevation ratio indicator value;

[0028] Select several points on the pre-disaster model as analysis points and obtain the elevation differences of each analysis point;

[0029] Obtain the spatial distances between pairwise analysis points to get the spatial distance set, preset the spatial distance threshold, and obtain the spatial weight matrix through the spatial distance set and the spatial distance threshold;

[0030] Perform merging processing on the elevation differences of each analysis point through the spatial weight matrix to obtain the elevation difference indicator value;

[0031] Perform fusion processing on the elevation ratio indicator value and the elevation difference indicator value to obtain the point cloud index.

[0032] Preferably, the process of training the convolutional neural network model using the vegetation comprehensive index, SAR index, and point cloud index to obtain the disaster recognition model is as follows:

[0033] Combine the vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster to obtain the recognition vector corresponding to each disaster;

[0034] Based on the disaster type corresponding to each historical disaster, label the recognition vector to obtain the labeled recognition vector;

[0035] Input the labeled recognition vector into the convolutional neural network model for training, and continuously adjust the weights and biases of each neuron based on the preset loss function to obtain the disaster recognition model.

[0036] Preferably, the process of obtaining the disaster type of the target area through the vegetation comprehensive index, SAR index, point cloud index, and disaster recognition model of the target area is as follows:

[0037] Combine the vegetation comprehensive index, SAR index, and point cloud index of the target area in sequence to obtain the target recognition vector;

[0038] Input the target recognition vector into the disaster recognition model to obtain a disaster discrimination vector;

[0039] Take the disaster type corresponding to the largest element in the disaster discrimination vector as the disaster type of the target area.

[0040] Preferably, a system for an automatic geological disaster recognition method based on multi-source remote sensing data includes:

[0041] Data acquisition module: Collect remote sensing data of the target area, remote sensing data of each historical disaster, and the disaster types of each historical disaster;

[0042] Preprocessing module: Preprocess the remote sensing data of the target area and the remote sensing data corresponding to each historical disaster to obtain the standard remote sensing data corresponding to the target area and each historical disaster;

[0043] Data analysis module: Analyze the standard remote sensing data to obtain a vegetation comprehensive index, an SAR index, and a point cloud index;

[0044] Model training module: Use the vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster to train a convolutional neural network model to obtain a disaster recognition model;

[0045] Output module: Input the vegetation comprehensive index, SAR index, and point cloud index of the target area into the disaster recognition model to obtain the disaster type of the target area.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] The present invention collects optical remote sensing images, SAR radar images, and LIDAR point cloud data of the target area and analyzes them respectively to obtain a vegetation comprehensive index, an SAR index, and a point cloud index; by combining the vegetation comprehensive index, SAR index, and point cloud index corresponding to historical disasters, a recognition vector is obtained, and the recognition vector is labeled based on the disaster types of each historical disaster to obtain a labeled recognition vector, and the convolutional neural network model is trained with the labeled recognition vector to obtain a disaster recognition model; finally, the vegetation comprehensive index, SAR index, and point cloud index of the target area are input into the disaster recognition model to obtain the disaster type of the target area, realizing the fusion of multi-source remote sensing data, improving the accuracy of disaster recognition, and providing a key prerequisite for the treatment work. Description of the Drawings

[0048] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for use in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other accompanying drawings can also be obtained based on these drawings.

[0049] Figure 1 This is the schematic diagram of the present invention. Detailed implementation manners

[0050] As Figure 1 shown, a geological disaster automatic recognition method based on multi-source remote sensing data includes:

[0051] Collect remote sensing data of the target area before and after the occurrence of geological disasters. The remote sensing data includes: optical remote sensing images, SAR radar images, and LIDAR point cloud data;

[0052] Preprocess the remote sensing data to obtain the standard remote sensing data of the target area before and after the occurrence of geological disasters; the standard remote sensing data includes: standard images, standard SARs, and standard point clouds;

[0053] Perform radiometric calibration, geometric correction, noise removal, and image enhancement on the optical remote sensing images and SAR radar images respectively to obtain standard images and standard SARs;

[0054] Perform format conversion, coordinate conversion, noise removal, and resampling on the LIDAR point cloud data to obtain standard point clouds;

[0055] Conduct a comparative analysis on the standard images of the target area before and after the disaster to obtain the disaster area of the target area, and analyze the standard images, standard SARs, and standard point clouds corresponding to the disaster area respectively to obtain the vegetation comprehensive index, SAR index, and point cloud index;

[0056] The process of analyzing the standard image of the disaster area to obtain the vegetation comprehensive index is as follows:

[0057] Divide the disaster area into several sub-areas, and number each sub-area. The number is represented by i, i = 1, 2... n, where n represents the total number of sub-areas. Then, conduct a secondary division on each sub-area to obtain each grandson area, and obtain the vegetation index on each grandson area;

[0058] Specifically, the vegetation index The calculation formula is:

[0059] ;

[0060] Among them, represents the reflectance of the near-infrared band of the grandson area, represents the reflectivity of the red light band in the Sun region;

[0061] Specifically, the leaf cell structure of plants has high reflectivity to near-infrared light and strong absorption to red light, while non-vegetation objects have low reflectivity to near-infrared light. When there are more plants in the sun area, the vegetation index is larger, otherwise, the vegetation index is smaller.

[0062] A vegetation index threshold is preset, and the sub-regions are screened based on the comparison result of the vegetation index threshold and the vegetation index of each sub-region to obtain the high-value region;

[0063] In detail, if the vegetation index of a grandchild region is higher than the vegetation index threshold, the grandchild region is marked as a high-value region;

[0064] The high-value frequency value of each sub-region is obtained by the number of high-value regions in each sub-region, and the high-value frequency value of each sub-region is processed by the spatial entropy formula to obtain the chaos degree of the target region;

[0065] In detail, the number of high-value areas corresponding to the sub-region is divided by the number of high-value areas corresponding to the disaster region to obtain the high-value frequency value corresponding to the sub-region;

[0066] In detail, the spatial entropy formula is:

[0067] ;

[0068] in, Represents chaos, Represents the high frequency value of the sub-region numbered i;

[0069] Get the minimum bounding rectangle of the disaster area, get the length and width of the rectangle, and get the aspect ratio;

[0070] Specifically, the length of the rectangle is divided by the width of the rectangle to obtain the aspect ratio;

[0071] Specifically, landslides are usually tongue-shaped, fan-shaped or semicircular, extending from the landslide wall to the slope foot, but the difference between the length and width of its minimum circumscribed rectangle is small; while debris flows have obvious valley dependence, and their flow area and accumulation area extend along the valley, presenting a long strip or belt shape, and the length and width of the minimum circumscribed rectangle differ greatly, so compared with landslides, the length-width ratio is larger;

[0072] The comprehensive vegetation index of the target area is obtained through aspect ratio and disorder;

[0073] Specifically, the chaos degree HLD and the aspect ratio Normalized descendant into the formula:

[0074] ;

[0075] Obtain the vegetation comprehensive index of the target area , where and are the weight influence factors corresponding to the chaos degree HLD and the aspect ratio respectively ;

[0076] Specifically, the geological disasters in this embodiment include: debris flow and landslide; the distributions of the washed-away plants in the disaster areas of debris flow and landslide are different. In the disaster area caused by debris flow, the distribution of the washed-away plants is relatively uniform, and the corresponding chaos degree will be relatively large; while in the disaster area caused by landslide, the distribution of the washed-away plants is relatively concentrated, so the corresponding chaos degree will be smaller than that of debris flow

[0077] By fusing the aspect ratio and the chaos degree, the vegetation comprehensive index is obtained, so that the vegetation coefficient can more accurately reflect the type of geological disaster; the smaller the vegetation comprehensive index, the higher the probability that the disaster area is a landslide, and the larger the vegetation comprehensive index, the higher the probability that the disaster area is a debris flow

[0078] The process of analyzing the standard SAR of the disaster area to obtain the SAR index is as follows

[0079] Obtain the gray-level co-occurrence matrix of the standard SAR at different angles, and obtain the correlation of the gray-level co-occurrence matrix at different angles

[0080] Take the standard deviation and mean of the correlation of the gray-level co-occurrence matrix at each angle as the angular wave value and the angular mean respectively

[0081] Specifically, since the SAR image generated by a landslide has strong texture, the correlation of the gray-level co-occurrence matrix in a specific direction will be significantly higher, while the correlation of the gray-level co-occurrence matrix at other angles is lower; while the texture of the debris flow shows mixed directionality, and the correlations corresponding to the gray-level co-occurrence matrices in all directions are relatively high

[0082] Therefore, the SAR image generated by the debris flow has a smaller angular wave value and a larger angular mean; the SAR image generated by the landslide has a larger angular wave value and a smaller angular mean

[0083] Fuse the angular wave value and the angular mean to obtain the SAR index

[0084] Specifically, after normalizing the angular wave value and the angular mean , substitute them into the formula

[0085] ;

[0086] Obtain the SAR index , where and They are the angular wave value and the angular mean value corresponding weight influence factors;

[0087] Specifically, the larger the SAR index, the higher the probability that the corresponding SAR image is a debris flow; the smaller the SAR index, the higher the probability that the corresponding SAR image is a landslide;

[0088] The process of analyzing the standard point cloud of the disaster area to obtain the point cloud index is as follows:

[0089] Construct digital elevation models before and after the disaster through the standard point clouds before and after the disaster. Denote the digital elevation model before the disaster as the pre-disaster model, and the digital elevation model after the disaster as the post-disaster model;

[0090] Randomly select two points on the pre-disaster model, randomly generate multiple paths between the two points, and at the same time select the same two points on the post-disaster model and generate the same paths;

[0091] Take points at equal distances on each path as marked points, obtain the elevation differences at each marked point on each path, and get the elevation set corresponding to each path;

[0092] Specifically, for a certain marked point on a certain path, subtract the elevation value corresponding to the pre-disaster model from the elevation value corresponding to the post-disaster model to obtain the elevation difference of this marked point, and obtain the elevation values of all marked points on this path, so as to obtain the elevation set corresponding to this path;

[0093] Based on the number of positive and negative values in the elevation sets corresponding to each path, obtain the elevation ratio, and take the mean of the elevation ratios of all paths as the elevation ratio indication value;

[0094] Specifically, obtain the number of positive and negative values in the elevation set corresponding to a certain path, divide the number of positive values by the number of negative values to obtain the elevation ratio corresponding to this path;

[0095] Specifically, different elevation ratios can reflect different disaster types. Landslides show a two-headed distribution, and the areas of the sliding area and the accumulation area are not very different. Therefore, the difference in the number of positive and negative values in the elevation set corresponding to landslides is not very large, and the elevation ratio is close to 1; Debris flows also show a two-headed distribution, but the area of the upstream erosion area is larger than that of the downstream accumulation area. Therefore, the number of negative values in the elevation set corresponding to debris flows is larger, and the elevation ratio is less than 1;

[0096] Therefore, the smaller the elevation ratio indication value, the higher the probability that the geological disaster is a debris flow;

[0097] Select several points on the pre-disaster model as analysis points, and obtain the elevation differences of each analysis point;

[0098] Obtain the spatial distances between pairwise analysis points to get a set of spatial distances, preset a spatial distance threshold, and obtain a spatial weight matrix through the set of spatial distances and the spatial distance threshold;

[0099] Specifically, number the analysis points, with the number represented by v, where v = 1, 2... g, and g represents the total number of analysis points. The value in the spatial weight matrix represents the weight value between the analysis point numbered e and the analysis point numbered r;

[0100] Specifically, if the spatial distance between the analysis point numbered e and the analysis point numbered r is less than the spatial distance threshold, then , otherwise, , and so on, thereby obtaining the spatial weight matrix;

[0101] Perform a merging process on the elevation differences of each analysis point through the spatial weight matrix to obtain an elevation difference indication value;

[0102] Specifically, through the formula:

[0103] ;

[0104] Obtain the elevation difference indication value , where represents the mean value of the elevation differences of all analysis points, represents the elevation difference of the analysis point numbered e, represents the elevation difference of the analysis point numbered r;

[0105] Specifically, due to the large influence range of debris flows and the long distance between the scouring area and the deposition area, when the distance between two analysis points is relatively close, that is, when the weight value between the two analysis points is 1, the differences between the two points and the mean value of the elevation differences of all analysis points show the same positive or negative state. Therefore, the elevation difference indication value is relatively large;

[0106] While the influence range of landslides is relatively small. When the distance between two analysis points is relatively close, the differences between the two points and the mean value of the elevation differences of all analysis points show a chaotic state, which may be the same positive or negative, or one positive and one negative. Therefore, the elevation difference indication value is relatively small;

[0107] Therefore, the smaller the elevation difference indication value, the greater the probability that the geological disaster is a landslide;

[0108] Perform a fusion process on the elevation ratio indication value and the elevation difference indication value to obtain a point cloud index;

[0109] Specifically, a number of elevation difference indication value intervals are preset. Different elevation difference indication value intervals correspond to different adjustment factors. The smaller the elevation difference indication value, the larger the corresponding adjustment factor. The elevation difference indication value interval where the elevation difference indication value is located is matched to obtain the corresponding adjustment factor. The adjustment factor is multiplied by the elevation ratio indication value to obtain the point cloud index;

[0110] Specifically, the smaller the point cloud index, the greater the probability of debris flow;

[0111] Obtain the types of each historical disaster and their corresponding vegetation comprehensive index, SAR index, and point cloud index. Use the vegetation comprehensive index, SAR index, and point cloud index to train the convolutional neural network model to obtain a disaster recognition model;

[0112] Combine the vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster to obtain the recognition vector corresponding to each disaster;

[0113] Based on the disaster type corresponding to each historical disaster, label the recognition vector to obtain a labeled recognition vector;

[0114] Input the labeled recognition vector into the convolutional neural network model for training. Continuously adjust the weights and biases of each neuron based on a preset loss function to obtain a disaster recognition model;

[0115] Specifically, number each disaster type. The number is represented by c, c = 1, 2... q, where q represents the number of disaster types. The preset loss function is:

[0116] ;

[0117] where SSZ represents the loss value, represents whether the current sample belongs to the disaster type numbered c. If it belongs, then = 1. Otherwise, represents the probability that the model predicts that the current sample belongs to the disaster type numbered c;

[0118] Specifically, based on the loss function, obtain the loss value of the sample. Continuously adjust the weights and biases of each neuron using the backpropagation mechanism based on the loss value of the sample, so that the loss value continuously decreases. At the same time, use the early stopping method to prevent the model from overfitting, thereby obtaining a disaster recognition model;

[0119] Through the vegetation comprehensive index, SAR index, point cloud index, and disaster recognition model of the target area, obtain the disaster type of the target area;

[0120] Combine the vegetation comprehensive index, SAR index, and point cloud index of the target area in sequence to obtain a target recognition vector;

[0121] Input the target recognition vector into the disaster recognition model to obtain a disaster discrimination vector;

[0122] Specifically, each value in the disaster discrimination vector represents the probability value that the disaster recognition model determines that the target area belongs to each type of disaster;

[0123] Take the disaster type corresponding to the largest element in the disaster discrimination vector as the disaster type of the target area;

[0124] A system for an automatic geological disaster recognition method based on multi-source remote sensing data, comprising:

[0125] A data acquisition module: acquiring remote sensing data of the target area, remote sensing data of each historical disaster, and the disaster type of each historical disaster;

[0126] A preprocessing module: preprocessing the remote sensing data of the target area and the remote sensing data corresponding to each historical disaster to obtain the standard remote sensing data corresponding to the target area and each historical disaster;

[0127] A data analysis module: analyzing the standard remote sensing data to obtain a vegetation comprehensive index, a SAR index, and a point cloud index;

[0128] A model training module: training a convolutional neural network model using the vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster to obtain a disaster recognition model;

[0129] An output module: inputting the vegetation comprehensive index, SAR index, and point cloud index of the target area into the disaster recognition model to obtain the disaster type of the target area;

[0130] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or decorations equivalent to changes within the scope of the technical solution of the present invention without departing from the technical solution of the present invention. However, any modification or equivalent replacement made to the above embodiments based on the technical essence of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. An automatic geological disaster recognition method based on multi-source remote sensing data, characterized in that, Including: Collecting remote sensing data of the target area before and after a geological disaster occurs. The remote sensing data includes: optical remote sensing images, SAR radar images, and LIDAR point cloud data; Preprocessing the remote sensing data to obtain the standard remote sensing data of the target area before and after the geological disaster occurs. The standard remote sensing data includes: standard images, standard SARs, and standard point clouds; Performing a comparative analysis on the standard images of the target area before and after the disaster occurs to obtain the disaster area of the target area, and respectively analyzing the standard images, standard SARs, and standard point clouds corresponding to the disaster area to obtain the vegetation comprehensive index, SAR index, and point cloud index; Obtaining the disaster types of each historical disaster and their corresponding vegetation comprehensive index, SAR index, and point cloud index, and using the vegetation comprehensive index, SAR index, and point cloud index to train a convolutional neural network model to obtain a disaster recognition model; Obtaining the disaster type of the target area through the vegetation comprehensive index, SAR index, point cloud index, and disaster recognition model of the target area.

2. The automatic geological disaster recognition method based on multi-source remote sensing data according to claim 1, wherein The process of preprocessing the remote sensing data to obtain the standard remote sensing data of the target area before and after the geological disaster occurs is as follows: Performing radiometric calibration, geometric correction, noise removal, and image enhancement on the optical remote sensing image and the SAR radar image respectively to obtain the standard image and the standard SAR; Performing format conversion, coordinate conversion, noise removal, and resampling on the LIDAR point cloud data to obtain the standard point cloud.

3. The automatic geological disaster recognition method based on multi-source remote sensing data according to claim 2, characterized in that, The process of analyzing the standard image of the disaster area to obtain the vegetation comprehensive index is as follows: Dividing the disaster area into several sub-areas, numbering each sub-area, and then performing a secondary division on each sub-area to obtain each grandson area, and obtaining the vegetation index on each grandson area; Presetting a vegetation index threshold, screening the grandson areas based on the comparison result between the vegetation index threshold and the vegetation index of each grandson area to obtain high-value areas; Obtaining the high-value frequency value of each sub-area through the number of high-value areas in each sub-area, and processing the high-value frequency value of each sub-area through the spatial entropy formula to obtain the chaos degree of the target area; Obtaining the minimum bounding rectangle of the disaster area, obtaining the length and width of the rectangle to obtain the aspect ratio, and obtaining the vegetation comprehensive index of the target area through the aspect ratio and the chaos degree.

4. A geological disaster automatic recognition method based on multi-source remote sensing data according to claim 3, characterized in that, The process of analyzing the standard SAR of the disaster area to obtain the SAR index is as follows: Obtaining the gray-level co-occurrence matrix of the standard SAR at different angles, and obtaining the correlation of the gray-level co-occurrence matrix at different angles; Taking the standard deviation and mean of the correlation of the gray-level co-occurrence matrix at each angle as the angular wave value and angular mean respectively; Performing a fusion process on the angular wave value and the angular mean to obtain the SAR index.

5. The automatic geological disaster recognition method based on multi-source remote sensing data according to claim 4, characterized in that, The process of analyzing the standard point cloud of the disaster area to obtain the point cloud index is as follows: Constructing a digital elevation model before and after the disaster through the standard point cloud before and after the disaster occurs, recording the digital elevation model before the disaster as the pre-disaster model, and recording the digital elevation model after the disaster as the post-disaster model; Randomly selecting two points on the pre-disaster model, randomly generating multiple paths between the two points, and at the same time selecting the same two points on the post-disaster model to generate the same paths; Equidistant points are taken on each path as marked points, and the elevation differences at each marked point on each path are obtained to get the elevation sets corresponding to each path; Based on the numbers of positive and negative values in the elevation sets corresponding to each path, an elevation ratio is obtained, and the mean value of the elevation ratios of all paths is taken as the elevation ratio indication value; On the pre-disaster model, several points are selected as analysis points, and the elevation differences of each analysis point are obtained; The spatial distances between pairwise analysis points are obtained to get a spatial distance set. A preset spatial distance threshold is set, and through the spatial distance set and the spatial distance threshold, a spatial weight matrix is obtained; The elevation differences of each analysis point are merged through the spatial weight matrix to obtain an elevation difference indication value; The elevation ratio indication value and the elevation difference indication value are fused to obtain a point cloud index.

6. The automatic geological disaster recognition method based on multi-source remote sensing data according to claim 5, characterized in that, The process of training a convolutional neural network model using the vegetation comprehensive index, SAR index, and point cloud index to obtain a disaster recognition model is as follows: The vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster are combined to obtain a recognition vector corresponding to each disaster; Based on the disaster types corresponding to each historical disaster, the recognition vector is labeled to obtain a labeled recognition vector; The labeled recognition vector is input into the convolutional neural network model for training, and based on a preset loss function, the weights and biases of each neuron are continuously adjusted to obtain a disaster recognition model.

7. The automatic geological disaster recognition method based on multi-source remote sensing data according to claim 6, characterized in that The process of obtaining the disaster type of the target area through the vegetation comprehensive index, SAR index, point cloud index, and disaster recognition model of the target area is as follows: The vegetation comprehensive index, SAR index, and point cloud index of the target area are combined in sequence to obtain a target recognition vector; The target recognition vector is input into the disaster recognition model to obtain a disaster discrimination vector; The disaster type corresponding to the largest element in the disaster discrimination vector is taken as the disaster type of the target area.

8. A system applied to the geological disaster automatic recognition method based on multi-source remote sensing data according to any one of claims 1-7, characterized in that, It includes: Data acquisition module: Collect remote sensing data of the target area, remote sensing data of each historical disaster, and the disaster types of each historical disaster; Preprocessing module: Preprocess the remote sensing data of the target area and the remote sensing data corresponding to each historical disaster to obtain the standard remote sensing data corresponding to the target area and each historical disaster; Data analysis module: Analyze the standard remote sensing data to obtain the vegetation comprehensive index, SAR index, and point cloud index; Model training module: Use the vegetation comprehensive index, SAR index, and point cloud index corresponding to each historical disaster to train a convolutional neural network model to obtain a disaster recognition model; Output module: Input the vegetation comprehensive index, SAR index, and point cloud index of the target area into the disaster model to obtain the disaster type of the target area.

Images