Landslide displacement image analysis method and system based on geological disaster monitoring

Abstract

The invention discloses a landslide displacement image analysis method and system based on geological disaster monitoring, and the method comprises the steps: 1), obtaining an image and hydrological information of a landslide region, and carrying out the preprocessing of the image; 2) extracting multi-scale features in the image; 3) classifying each pixel in the image to generate a segmentation result; 4) comparing the feature points to obtain landslide displacement data; 5) calculating the pixel area of the landslide area, and converting the pixel area into an actual area to obtain the total area of the landslide; 6) calculating an area conversion scale; 7) judging the possible scale of the landslide through the hydrological information and treatment information model; 8) adding the scale obtained by the area conversion scale, the hydrological information and the governance information model according to the weight to obtain a landslide prediction scale; (9) when the landslide displacement data exceed a threshold value, giving an alarm, and generating a risk assessment report; according to the invention, the related information of the landslide area can be rapidly obtained through image analysis, the personnel response speed is accelerated, and the loss caused by disasters is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster monitoring, and in particular to a landslide displacement image analysis method and system based on geological disaster monitoring. Background Art

[0002] Landslides are a common geological disaster, often causing severe damage to human life and infrastructure. Typhoons, a common disaster in coastal areas, are highly susceptible to landslides. Traditional landslide monitoring methods typically rely on on-site surveys and manual observation, which are not only time-consuming and labor-intensive but also have limited accuracy.

[0003] With the development of remote sensing and image processing technologies, image analysis has become an efficient and accurate method for monitoring landslides and estimating their scale. Fujian, a mountainous typhoon-prone region, requires significant manpower and resources to assess geological hazard risks after heavy rains, making it difficult to obtain data on landslide area and scale immediately. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a method for landslide displacement monitoring based on image analysis and estimating the scale of the disaster using data such as the affected area, so as to improve the efficiency and accuracy of landslide monitoring after severe weather.

[0005] In order to achieve the above technical objectives, the technical solution adopted by the present invention is: a landslide displacement image analysis method based on geological disaster monitoring, comprising:

[0006] 1) Obtain high-resolution remote sensing images of the landslide area and hydrological information of the image location, and preprocess the acquired images;

[0007] 2): Extract multi-scale features from images through convolutional neural networks;

[0008] 3): According to the extracted features, each pixel in the image is classified to generate the segmentation result;

[0009] 4): Obtain landslide displacement data by comparing feature points with historical images;

[0010] 5): According to the image segmentation results, the pixel area of ​​the landslide area is calculated, and according to the spatial resolution of the remote sensing image, the pixel area is converted into the actual area to obtain the total area of ​​the landslide;

[0011] 6) Calculate the area conversion scale by using the total area of ​​the landslide and the total area of ​​the image acquisition area;

[0012] 7): Determine the possible scale of landslide by using hydrological information model and control information model respectively;

[0013] 8) The scale obtained by area conversion, hydrological information and governance information model is added proportionally according to the weight to obtain the predicted scale value of the possible landslide;

[0014] 9) Set displacement warning thresholds based on historical data and expert experience. Once landslide displacement data exceeds the threshold, an alarm will be automatically issued; a risk assessment will be conducted and a corresponding report will be generated;

[0015] The report contains hydrological information of the current location, landslide area data, and predicted scale values ​​of possible landslides.

[0016] Furthermore, the pre-processing in step 1) specifically includes:

[0017] Use filtering algorithm (median filtering) to denoise the acquired remote sensing images;

[0018] Perform image correction on the image to ensure the accuracy of the image; image correction includes geometric correction and radiation correction;

[0019] Through histogram equalization and contrast enhancement technology, the clarity of the image is improved to ensure that the features are more obvious.

[0020] Furthermore, the multi-scale features in step 2) include small-scale features, medium-scale features, and large-scale features;

[0021] The small-scale features are local textures and edges, including small cracks, local soil texture changes, small changes in vegetation, and minor slips or breaks;

[0022] The mesoscale features are the overall outline of the landslide area and the sliding of local soil layers;

[0023] The large-scale features are the overall appearance of the entire landslide area and the large-scale changes in the terrain.

[0024] Furthermore, the image feature extraction step in step 2) includes:

[0025] Use SIFT (Scale Invariant Feature Transform) and SURF (Speeded Up Robust Features) methods to extract key feature points in the image; ensure accurate matching in images at different time points.

[0026] Perform feature point matching through FLANN (Fast Nearest Neighbor Search Library) to find corresponding feature point pairs in images at different times;

[0027] Based on the matched feature point pairs, the image transformation matrix (affine transformation matrix) is calculated to stitch multiple images into a panoramic image. The transformation calculation method is the RANSAC (Random Sample Consensus) algorithm to improve the robustness of the matching.

[0028] Furthermore, step 3) specifically includes:

[0029] A convolutional neural network is used to generate a corresponding feature vector for each pixel. The feature vector contains information about the pixel at multiple scales, including local edges, shape features, and global semantic information of the landslide area.

[0030] After feature extraction, the feature vector of each pixel is sent to the classification layer (Softmax function) for processing;

[0031] The classifier outputs the probability of each pixel belonging to different categories based on these feature vectors; the categories include landslide area, non-landslide area, boundary area, and vegetation covered area;

[0032] In semantic segmentation, all pixels with the same category are labeled as the same category, and finally a segmented image is generated to represent the boundary, size and morphology of the landslide area.

[0033] Furthermore, in step 4), the displacement vector of the landslide area is calculated by optical flow technology; the specific method is as follows:

[0034] The Lucas-Kanade optical flow method is used to calculate the displacement of objects (landslide areas) in the image sequence, which can handle real-time dynamic changes. The displacement between different time points is calculated using the extracted key feature points using the matching algorithm FLANN.

[0035] Based on the displacement vectors of the feature points, the overall displacement of the landslide area is calculated using geometric transformation (similarity transformation).

[0036] Furthermore, the expression for the area conversion scale in step 6) is: area conversion scale = total area of ​​the landslide / total area of ​​the image acquisition region.

[0037] Furthermore, the hydrological information model in step 7) predicts the possible scale of the landslide based on the current hydrological information;

[0038] The hydrological information model is trained by using hydrological information as input to a convolutional neural network model and the scale of landslides that occurred historically at locations corresponding to the hydrological information as output, and training the model until the model converges.

[0039] The control information model predicts the possible scale of landslides by using current hydrological information and investment information of landslide control projects; wherein the investment information of landslide control projects includes control capital investment, control duration, and control manpower investment;

[0040] The training method of the control information model is as follows: using hydrological information and landslide control project investment information as inputs to a convolutional neural network model, and using the scale of historical landslides at locations corresponding to the hydrological information and landslide control project investment information as model outputs, and training the model until the model converges;

[0041] The landslide scale predicted by the hydrological information model and the governance information model is mapped to a numerical value range of 0 to 1 according to the severity, where the value 1 represents the situation corresponding to the scale of the most serious landslide in history.

[0042] Furthermore, step 8) is as follows:

[0043] The landslide scale values ​​obtained from the area conversion scale, hydrological information model, and governance information model are multiplied by their corresponding weights and added together to obtain the predicted scale value of the possible landslide.

[0044] Among them, the weight of area conversion scale is 0.6 to 0.9; the sum of the weights of landslide scale values ​​obtained by the hydrological information model and the governance information model is 0.1 to 0.4.

[0045] The present invention also provides a landslide displacement image analysis system based on geological disaster monitoring, which applies the above-mentioned landslide displacement image analysis method based on geological disaster monitoring. The system includes:

[0046] Data acquisition module, used to obtain high-resolution images of the landslide area and hydrological information of the image location;

[0047] Image preprocessing module, used for filtering, rectifying, histogram equalization and contrast enhancement of images;

[0048] The image processing module is used to extract multi-scale features in the image and classify each pixel in the image based on the extracted features to generate segmentation results;

[0049] Landslide displacement analysis module, used to obtain landslide displacement data by comparing feature points with historical images;

[0050] The landslide area analysis module is used to calculate the pixel area of ​​the landslide area based on the image segmentation results, and convert the pixel area into the actual area according to the spatial resolution of the remote sensing image to obtain the total area of ​​the landslide;

[0051] A conversion module, for calculating the area conversion scale by using the total area of ​​the landslide and the total area of ​​the image acquisition area;

[0052] The landslide scale prediction module is used to determine the possible scale of the landslide through the hydrological information model and the governance information model, and to add the area conversion scale, the scale obtained from the hydrological information and the governance information model in proportion according to the weight to obtain the predicted scale value of the possible landslide;

[0053] The logic judgment module is used to determine whether the landslide displacement data exceeds the threshold. If so, it controls the early warning module to issue an alarm; and controls the risk assessment module to conduct risk assessment and generate corresponding reports;

[0054] Risk assessment module, used to conduct landslide risk assessment and generate corresponding reports;

[0055] The early warning module is used to issue an alarm prompt.

[0056] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0057] This method uses image analysis to quickly obtain information about large landslide areas. The data used is immediately available, accelerating response times and minimizing disaster losses. Furthermore, the method utilizes high-resolution remote sensing imagery and advanced image processing techniques to provide precise data on landslide displacement and area. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 It is a simplified flow chart of the present invention. DETAILED DESCRIPTION

[0060] The present invention will be described in further detail below with reference to the accompanying drawings and examples. It is particularly noted that the following examples are intended only to illustrate the present invention and are not intended to limit the scope of the present invention. Similarly, the following examples are only some embodiments of the present invention and are not intended to be exhaustive. All other embodiments obtained by those of ordinary skill in the art without creative effort are intended to fall within the scope of protection of the present invention.

[0061] Refer to the attached Figure 1 As shown, this embodiment provides a landslide displacement image analysis method based on geological disaster monitoring, including:

[0062] 1) Use remote sensing equipment such as drones, satellites, and ground cameras in high-risk areas to obtain high-resolution remote sensing images of the landslide area and hydrological information at the image location, and preprocess the acquired images. Preprocessing specifically includes:

[0063] Use filtering algorithm (median filtering) to denoise the acquired remote sensing images; perform image correction processing (image correction includes geometric correction and radiation correction) on the images to ensure the accuracy of the images; use histogram equalization and contrast enhancement technology to improve the clarity of the images and ensure that the features are more obvious.

[0064] 2) Extract multi-scale features from the image through convolutional neural networks; the multi-scale features include:

[0065] Small-scale features: local textures and edges, including small cracks, local soil texture changes, subtle changes in vegetation, and tiny slips or breaks; medium-scale features (larger local): the overall outline of the landslide area, slips in local soil layers; large-scale features (global): the overall picture of the entire landslide area, and large-scale changes in terrain.

[0066] The image feature extraction step includes:

[0067] Use SIFT (Scale Invariant Feature Transform) and SURF (Speeded Up Robust Features) methods to extract key feature points in the image; ensure accurate matching in images at different time points.

[0068] Perform feature point matching through FLANN (Fast Nearest Neighbor Search Library) to find corresponding feature point pairs in images at different times;

[0069] Based on the matched feature point pairs, the image transformation matrix (affine transformation matrix) is calculated using the RANSAC (Random Sample Consensus) algorithm to improve the robustness of the matching. Multiple images are stitched together into a panoramic image, providing a more comprehensive view of the landslide area for easier analysis.

[0070] 3): Based on the extracted features, each pixel in the image is classified and the segmentation result is generated; specifically, it includes:

[0071] A convolutional neural network generates a corresponding feature vector for each pixel. These feature vectors contain information about the pixel at multiple scales, including local edges, shape features, and global semantic information about the landslide area. After feature extraction, the feature vector of each pixel is fed into a classification layer (Softmax function) for processing. Based on these feature vectors, the classifier outputs the probability that each pixel belongs to a different category. Categories include landslide areas (including sliding rock and soil, collapsed areas, etc.); non-landslide areas (including background, stable surface, or unchanged areas); boundary areas (such as these areas often have high gradient changes and are areas where landslides are expanding or about to occur); and vegetation-covered areas (in landslide monitoring, the disappearance or change of vegetation can be a precursor to landslides).

[0072] In semantic segmentation, all pixels with the same category (e.g., landslide area, non-landslide area) are labeled as the same category, and a segmented image is finally generated to represent the boundary, size, and morphology of the landslide area.

[0073] 4): Obtain landslide displacement data by comparing feature points based on historical images. In this embodiment, the displacement vector of the landslide area is calculated using the optical flow method. The specific method is as follows:

[0074] The Lucas-Kanade optical flow method is used to calculate the displacement of objects (landslide areas) in the image sequence, which can handle real-time dynamic changes. The displacement between different time points is calculated using the extracted key feature points using the matching algorithm FLANN.

[0075] Based on the displacement vectors of the feature points, the overall displacement of the landslide area is calculated using geometric transformation (similarity transformation).

[0076] 5) Based on the image segmentation results, the pixel area of ​​the landslide area is calculated. According to the spatial resolution of the remote sensing image, the pixel area is converted into the actual area to obtain the total area of ​​the landslide. The affected area is determined based on the difference in image segmentation before and after the landslide occurs.

[0077] 6) Calculate the area conversion scale using the total area of ​​the landslide and the total area of ​​the image acquisition region; wherein the area conversion scale is expressed as: area conversion scale = total area of ​​the landslide / total area of ​​the image acquisition region.

[0078] 7): Determine the possible scale of landslide by using hydrological information model and control information model respectively;

[0079] The hydrological information model predicts the possible scale of landslides based on current hydrological information;

[0080] The hydrological information model is trained by using hydrological information (including precipitation, groundwater levels, soil moisture, and other hydrological information) as input to a convolutional neural network model. The model output is the scale of historical landslides at the locations corresponding to the hydrological information, and the model is trained until convergence. A convolutional neural network (CNN) is used to process this temporal data (such as changes in rainfall) and spatial data (such as the geographic location of the landslide and the topography of that location as assessed by geologists). The CNN extracts characteristics that influence landslides from this data, such as the possibility that long-term, heavy rainfall may lead to soil saturation, increasing the risk of landslides. Areas historically prone to landslides may experience landslides again under similar conditions. The model ultimately outputs a predicted value, indicating the scale of the landslide that is likely to occur given current hydrological conditions and historical landslide data.

[0081] The governance information model predicts the possible scale of landslides through current hydrological information and investment information of landslide governance projects. The investment information of landslide governance projects includes governance capital investment, governance duration, and governance manpower investment, that is, how much money was spent on landslide governance, how much manpower was invested, and how long the governance project lasted.

[0082] The remediation information model is trained using hydrological information and information on investment in landslide remediation projects as inputs to a convolutional neural network model. The model outputs the historical scale of landslides at locations corresponding to these hydrological and landslide remediation project investment information, and the model is trained until convergence. A convolutional neural network (CNN) is used to analyze remediation information. From this data, the CNN learns the relationship between remediation investment and landslide risk: greater remediation investment or longer-term remediation efforts may effectively slow the scale or reduce the frequency of landslides. Factors such as whether remediation projects are conducted in high-risk areas and whether sufficient resources are invested also influence the final scale of landslides. The model outputs a predicted value, representing the scale of the landslide that is likely to occur under existing remediation measures.

[0083] The landslide scale predicted by the hydrological information model and the governance information model is mapped to a numerical value range of 0 to 1 according to the severity of the scale, where the value 1 represents the situation corresponding to the scale of the most serious landslide in history.

[0084] 8) The scale obtained by area conversion, hydrological information and governance information model is added proportionally according to the weight to obtain the predicted scale value of the possible landslide; the details are as follows:

[0085] The landslide scale values ​​obtained from the area conversion scale, hydrological information model, and governance information model are multiplied by their corresponding weights and added together to obtain the predicted scale value of the possible landslide.

[0086] Among them, the weight of area conversion scale is 0.6 to 0.9; the sum of the weights of landslide scale values ​​obtained by the hydrological information model and the governance information model is 0.1 to 0.4.

[0087] 9) Displacement warning thresholds are set based on historical data and expert experience. Once the landslide displacement data exceeds the threshold, an alarm is automatically issued; a risk assessment is conducted and a corresponding report is generated; the report includes hydrological information of the current location, landslide area data, and the predicted scale of the possible landslide.

[0088] As a further improvement, the vector field is used to display the displacement of the landslide area to intuitively reflect the changes; and a visual comparison of historical displacement data is provided to facilitate the monitoring of landslide change trends, which are sent to staff together with warning information.

[0089] This embodiment further provides a landslide displacement image analysis system based on geological disaster monitoring, which applies the above-mentioned landslide displacement image analysis method based on geological disaster monitoring. The system includes:

[0090] Data acquisition module, used to obtain high-resolution images of the landslide area and hydrological information of the image location;

[0091] Image preprocessing module, used for filtering, rectifying, histogram equalization and contrast enhancement of images;

[0092] The image processing module is used to extract multi-scale features in the image and classify each pixel in the image based on the extracted features to generate segmentation results;

[0093] Landslide displacement analysis module, used to obtain landslide displacement data by comparing feature points with historical images;

[0094] The landslide area analysis module is used to calculate the pixel area of ​​the landslide area based on the image segmentation results, and convert the pixel area into the actual area according to the spatial resolution of the remote sensing image to obtain the total area of ​​the landslide;

[0095] A conversion module, for calculating the area conversion scale by using the total area of ​​the landslide and the total area of ​​the image acquisition area;

[0096] The landslide scale prediction module is used to determine the possible scale of the landslide through the hydrological information model and the governance information model, and to add the area conversion scale, the scale obtained from the hydrological information and the governance information model in proportion according to the weight to obtain the predicted scale value of the possible landslide;

[0097] The logic judgment module is used to determine whether the landslide displacement data exceeds the threshold. If so, it controls the early warning module to issue an alarm; and controls the risk assessment module to conduct risk assessment and generate corresponding reports;

[0098] The risk assessment module is used to conduct landslide risk assessment and generate corresponding reports; the report contains hydrological information of the current location, landslide area data, and predicted scale values ​​of possible landslides.

[0099] The early warning module is used to issue an alarm prompt; specifically, the alarm prompt can be issued in the form of sound, text, image, etc.

[0100] The above descriptions are only some embodiments of the present invention and do not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A landslide displacement image analysis method based on geological disaster monitoring, characterized in that: include: 1): Obtain high-resolution remote sensing images of the landslide area and the hydrological information of the image location, and preprocess the acquired images; 2): Extract multi-scale features in images through convolutional neural networks; 3): According to the extracted features, each pixel in the image is classified to generate the segmentation result; 4): Obtain landslide displacement data by comparing feature points based on historical images; 5): According to the image segmentation results, the pixel area of ​​the landslide area is calculated, and according to the spatial resolution of the remote sensing image, the pixel area is converted into the actual area to obtain the total area of ​​the landslide; 6) Calculate the area conversion scale by the total area of ​​the landslide and the total area of ​​the image acquisition area; 7): Determine the possible scale of landslide through hydrological information model and governance information model respectively; 8): Add the area conversion scale, hydrological information and governance information model in proportion to the weights to obtain the predicted scale value of the possible landslide; 9) Set displacement warning thresholds based on historical data and expert experience. Once the landslide displacement data exceeds the threshold, an alarm will be automatically issued; perform risk assessment and generate corresponding reports; The report contains hydrological information of the current location, landslide area data and predicted scale values ​​of possible landslides.

2. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: The pretreatment in step 1) specifically includes: Use filtering algorithms to denoise the acquired remote sensing images; Perform image correction on the image to ensure the accuracy of the image; image correction includes geometric correction and radiation correction; Improve image clarity through histogram equalization and contrast enhancement technology.

3. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: In step 2), the multi-scale features include small-scale features, medium-scale features and large-scale features; The small-scale features are local textures and edges, including small cracks, local soil texture changes, small changes in vegetation, and tiny slides or breaks; The mesoscale features are the overall outline of the landslide area and the sliding of local soil layers; The large-scale features are the overall view of the entire landslide area and the large-scale changes in terrain.

4. The landslide displacement image analysis method based on geological disaster monitoring according to claim 3 is characterized in that: The image feature extraction step in step 2) includes: Use SIFT and SURF methods to extract key feature points in the image; Use FLANN to match feature points and find corresponding feature point pairs in images at different times; According to the matched feature point pairs, the image transformation matrix is ​​calculated, and multiple images are stitched into a panoramic image; wherein, the transformation calculation method is the RANSAC algorithm.

5. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: Step 3) specifically includes: A convolutional neural network is used to generate a corresponding feature vector for each pixel, wherein the feature vector contains information of the pixel at multiple scales, including local edges, shape features, and global semantic information of the landslide area; After extracting features, the feature vector of each pixel is sent to the classification layer for processing; The classifier outputs the probability of each pixel belonging to different categories based on these feature vectors; the categories include landslide area, non-landslide area, boundary area, and vegetation covered area; In semantic segmentation, all pixels with the same category are labeled as the same category, and finally a segmented image is generated, which represents the boundary, size and morphology of the landslide area.

6. The landslide displacement image analysis method based on geological disaster monitoring according to claim 4 is characterized in that: In step 4), the displacement vector of the landslide area is calculated by optical flow method; the specific method is as follows: The displacement of objects in the image sequence is calculated by the Lucas-Kanade optical flow method, and the displacement between different time points is calculated by the matching algorithm FLANN using the extracted key feature points; Based on the displacement vectors of the feature points, the overall displacement of the landslide area is calculated using geometric transformation.

7. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: The expression of the area conversion scale in step 6) is: area conversion scale = total area of ​​the landslide / total area of ​​the image acquisition area.

8. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: The hydrological information model in step 7) predicts the possible scale of landslide through current hydrological information; The training method of the hydrological information model is: using the hydrological information as the input of the convolutional neural network model, using the scale of landslides that occurred historically at the location corresponding to the hydrological information as the model output, and training the model until the model converges; The control information model predicts the possible scale of landslides through current hydrological information and input information of landslide control projects; wherein the input information of landslide control projects includes control capital investment, control duration, and control manpower investment; The training method of the control information model is: using hydrological information and input information of landslide control projects as input of the convolutional neural network model, using the scale of landslides that have occurred in history at locations corresponding to the hydrological information and input information of landslide control projects as model output, and training the model until the model converges; The scale of landslides predicted by the hydrological information model and the governance information model is mapped to a numerical range of 0 to 1 according to the severity, where the value 1 represents the situation corresponding to the scale of the most serious landslide in history.

9. The landslide displacement image analysis method based on geological disaster monitoring according to claim 1 is characterized in that: Step 8) is as follows: The landslide scale values ​​obtained from the area conversion scale, hydrological information model and governance information model are multiplied by their corresponding weights and added together to obtain the predicted scale value of the possible landslide; Among them, the weight of area conversion scale is 0.6-0.9; the sum of the weights of landslide scale values ​​obtained by the hydrological information model and the governance information model is 0.1-0.

4.

10. A landslide displacement image analysis system based on geological disaster monitoring, characterized in that: The landslide displacement image analysis method based on geological disaster monitoring as claimed in any one of claims 1 to 9 is applied, and the system comprises: A data acquisition module is used to obtain high-resolution images of the landslide area and hydrological information of the image location; Image preprocessing module, used for filtering, rectifying, histogram equalization and contrast enhancement of images; The image processing module is used to extract multi-scale features in the image and classify each pixel in the image according to the extracted features to generate segmentation results; The landslide displacement analysis module is used to obtain landslide displacement data by comparing feature points based on historical images; The landslide area analysis module is used to calculate the pixel area of ​​the landslide area according to the image segmentation results, and convert the pixel area into the actual area according to the spatial resolution of the remote sensing image to obtain the total area of ​​the landslide; A conversion module, used for calculating the area conversion scale by the total area of ​​the landslide and the total area of ​​the image acquisition area; The landslide scale prediction module is used to determine the possible scale of the landslide through the hydrological information model and the governance information model, and to add the area conversion scale, the scale obtained by the hydrological information and the governance information model in proportion according to the weight to obtain the predicted scale value of the possible landslide; The logic judgment module is used to judge whether the landslide displacement data exceeds the threshold. If so, the early warning module is controlled to issue an alarm; and the risk assessment module is controlled to perform risk assessment and generate a corresponding report; Risk assessment module, used to conduct landslide risk assessment and generate corresponding reports; The early warning module is used to issue an alarm prompt.

Images