Landslide identification early warning system and method based on remote sensing satellite image

Abstract

The invention discloses a landslide identification and early warning system and method based on remote sensing satellite images, and belongs to the technical field of geological monitoring. The problem that the accuracy and reliability of data are easy to interfere due to the fact that the precision and stability of part of monitoring equipment need to be improved in the prior art is solved, the satellite image of the target mountain area is obtained through the remote sensing satellite, landslide feature extraction is conducted on the satellite image through the synthetic aperture radar interferometry technology, and the landslide feature extraction accuracy is improved. Identifying a potential landslide area in the target mountain area; a landslide risk assessment model is constructed in combination, the landslide risk assessment model is utilized to predict real-time mountain data, the degree of landslide risks existing in a potential landslide area is obtained, and a risk assessment report is generated and fed back to the man-machine interaction unit to give an early warning; through fusion of satellite remote sensing images and ground monitoring data, real-time monitoring and early warning of landslide are realized, accuracy and reliability of identification and early warning are improved, and loss caused by geological disasters is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of geological monitoring, and specifically to a landslide identification and early warning system and method based on remote sensing satellite images. Background Technique

[0002] A landslide is a collapse phenomenon that occurs when rocks are under the action of gravity, and often occurs in steep areas of mountainous regions. The occurrence of landslides is usually related to various factors, including river erosion, groundwater activity, rainwater immersion, earthquakes, and artificial slope cutting, etc.; China not only has a vast territory, but the mountainous areas are more complex in terrain. For example, the common Sichuan Basin, Chongqing mountainous areas, etc. are all accident-prone areas for landslides and earthquakes; Landslides can cause great disasters, and in severe cases, they can even destroy entire villages, kill people and livestock, damage factories and power stations, and block roads; At the same time, a large number of stones and soil blocks after a landslide fall into the river, which will also block the river and cause flood disasters; In order to reduce the impact and harm brought by landslides, it is very necessary to identify and early warn of landslides, which not only buys precious time to deal with disasters, but also enhances people's ability to cope with natural disasters.

[0003] However, in the existing technology, the accuracy and stability of some monitoring devices in the landslide identification and early warning system still need to be improved. Especially in complex geological environments, the accuracy and reliability of data are easily interfered; In addition, the degree of automation and intelligence level of the early warning system still need to be improved to ensure more efficient data processing and early warning response.

[0004] Therefore, it does not meet the existing needs, and for this reason, we have proposed a landslide identification and early warning system and method based on remote sensing satellite images. Summary of the Invention

[0005] The purpose of the present invention is to provide a landslide identification and early warning system and method based on remote sensing satellite images. By obtaining satellite images of the target mountain area through remote sensing satellites, using synthetic aperture radar interferometry technology to extract landslide features from the satellite images, thereby identifying potential landslide areas in the target mountain area; Combining the construction of a landslide risk assessment model, using the landslide risk assessment model to predict real-time mountain data, obtaining the degree of landslide risk existing in the potential landslide area, and generating a risk assessment report and feeding it back to the human-computer interaction unit to issue an early warning; By integrating satellite remote sensing images and ground monitoring data, realizing real-time monitoring and early warning of landslides, improving the accuracy and reliability of identification and early warning, reducing the losses caused by geological disasters, and solving the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A landslide identification and early warning system based on remote sensing satellite images, including:

[0008] An image acquisition unit, configured to collect historical mountain data of a target mountain area, and photograph the target mountain area through a remote sensing satellite to obtain a high-resolution satellite image of the target mountain area;

[0009] An image processing unit, configured to preprocess the satellite image, and the preprocessing includes: removing noise, image enhancement, radiometric calibration, and atmospheric correction;

[0010] A feature extraction unit, configured to extract landslide features from the satellite image based on synthetic aperture radar interferometry, analyze the features of terrain changes and surface displacements in the satellite image, and then identify potential landslide areas;

[0011] A landslide identification unit, configured to obtain real-time mountain data of potential landslide areas through Beidou satellite technology and sensor technology, establish and train a landslide risk assessment model based on historical mountain data, perform risk prediction on the real-time mountain data through the landslide risk assessment model, evaluate the landslide risk level of potential landslide areas, and generate a risk assessment report and feedback it to the human-computer interaction unit;

[0012] A human-computer interaction unit, configured to receive and display the mountain data and landslide risk assessment report of the target mountain area, and issue corresponding control instructions according to the data and report.

[0013] Further, the image acquisition unit includes:

[0014] An image acquisition module, configured to determine the scope and monitoring requirements of the target mountain area, and select a suitable remote sensing satellite according to the monitoring requirements to obtain image data of the target mountain area;

[0015] A historical data collection module, configured to collect historical mountain data within the target mountain area and save it in a time series to provide basic data for subsequent mountain change detection.

[0016] Further, the image processing unit includes:

[0017] A preliminary processing module, configured to filter out spike noise through Fourier transform filtering, remove strip noise through Fourier transform and low-pass filtering, and adjust the readability and information extraction ability of the satellite image through histogram transformation method to improve the contrast and visual effect of the satellite image;

[0018] A secondary processing module, configured to convert the brightness value in the original remote sensing image into the reflectance or radiance value of the outer atmosphere surface through radiometric calibration, so that the observed data is converted from the brightness unit of the sensor to a physically comparable and analyzable unit; then convert the radiance or reflectance data into the actual reflectance of the surface through atmospheric correction to obtain the true reflectance, radiance and other physical model parameters of the ground object.

[0019] Furthermore, the feature extraction unit includes:

[0020] An image selection module, which is used to select satellite images of a suitable target mountain area, and after preprocessing the satellite images, select the landslide area that needs to be monitored and identified key points according to the early warning requirements and the scope of the target mountain area;

[0021] An image processing module, which is used to register the selected satellite images and select a suitable interference pair by comparing satellite images at different times;

[0022] A feature extraction module, which is used to decompose the interference result by wavelet transform, extract the time series information of the surface deformation, and analyze the interference result to estimate the terrain phase of the surface;

[0023] A feature determination module, which is used to analyze the terrain phase of the surface, evaluate the characteristics and dynamic changes of the landslide, and obtain the landslide feature information in the satellite image.

[0024] Furthermore, the landslide identification unit includes:

[0025] A sample processing module, which is used to clean the historical mountain data of the target mountain area and divide the historical mountain data into a training set and a test set;

[0026] A model construction module, which is used to create a neural network model, import the training set into the neural network model for training to obtain the simulation prediction result of the target mountain area; then import the test set into the trained neural network model for verification, and obtain an optimized landslide risk assessment model;

[0027] A real-time prediction module, which is used to import the real-time mountain data of the potential landslide area into the landslide risk assessment model for landslide prediction, obtain the landslide risk degree of the potential landslide area, and generate a risk assessment report;

[0028] A result feedback module, which is used to feedback the risk assessment report to the human-computer interaction unit in real time through wireless communication technology.

[0029] Furthermore, the human-computer interaction unit includes:

[0030] A data receiving module, which is used to receive and display the mountain data and the landslide risk assessment report of the target mountain area in real time.

[0031] A method for identifying and warning mountain landslides based on remote sensing satellite images includes the following steps:

[0032] Step 1: Obtain high-resolution satellite images of the target mountain area through a remote sensing satellite, and collect the historical mountain data of the target mountain area;

[0033] Step 2: After preprocessing the satellite image, use synthetic aperture radar interferometry to extract landslide features from the satellite image, and analyze and identify potential landslide areas in the target mountain area;

[0034] Step 3: Build a landslide risk assessment model, clean the historical mountain data, and divide it into a training set and a test set. Import the training set and the test set into the landslide risk assessment model in sequence for risk training to obtain an optimized landslide risk assessment model;

[0035] Step 4: Based on the landslide identification results, use Beidou satellite technology and sensor technology to obtain real-time mountain data of potential landslide areas. Use the landslide risk assessment model to predict the real-time mountain data, obtain the degree of landslide risk in the potential landslide areas, and generate a risk assessment report;

[0036] Step 5: Based on wireless transmission technology, the risk assessment report is fed back to the human-computer interaction unit in real time for early warning. The human-computer interaction unit promptly issues early warning information to notify the residents and relevant departments in the affected areas to take countermeasures.

[0037] Further, in the above Step 2, using synthetic aperture radar interferometry to extract landslide features from the satellite image specifically includes the following steps:

[0038] Select the landslide areas that need to be monitored and identified key points according to the early warning requirements and the scope of the target mountain area, register the selected satellite images to ensure the spatial consistency of the images at different time points; by comparing the satellite images at different times, find the image pairs with coherence to obtain suitable interference pairs; perform wavelet transform decomposition on the interference results to extract the temporal information of surface deformation; select a part of the ground stable points as references, and estimate the topographic phase of the surface by analyzing the interference results; analyze the processing results to evaluate the characteristics and dynamic changes of the landslide, including: deformation rate, deformation trend, to obtain the landslide feature information in the satellite image.

[0039] Further, in the above Step 2, analyzing and identifying potential landslide areas in the target mountain area specifically includes the following steps:

[0040] Based on the landslide feature information in the satellite image, perform zonal identification on the satellite image to determine the remote sensing sub-images with landslide terrain; and based on the geological state information of the landslide terrain, judge whether the mudslide event occurs and obtain the mudslide movement information; by comparing the multi-temporal remote sensing images before and after the landslide, use the normalization method to detect the change of the vegetation index to determine the potential landslide areas; and combine the geometric rules of the satellite image to exclude the non-landslide parts of roads, buildings, and bare land to complete the fine identification of potential landslide areas.

[0041] Further, in step 4, after predicting the real-time mountain data by using the landslide risk assessment model, the following steps are specifically included:

[0042] Use Beidou satellite technology and sensor technology to continuously monitor the mountain body in the potential landslide area, and combine visual interpretation and field investigation to verify and correct the evaluation results of the landslide risk assessment model. The field investigation includes: on-site inspection of the landslide site to confirm the type, scale and cause of the landslide.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] In the present invention, satellite images of the target mountain area are obtained through remote sensing satellites, and synthetic aperture radar interferometry is used to extract landslide features from the satellite images, thereby identifying potential landslide areas in the target mountain area; combined with the construction of a landslide risk assessment model, the landslide risk assessment model is used to predict the real-time mountain data, obtain the degree of landslide risk existing in the potential landslide area, generate a risk assessment report and feedback it to the human-computer interaction unit to issue a warning; by fusing satellite remote sensing images and ground monitoring data, real-time monitoring and warning of mountain landslides are realized, improving the accuracy and reliability of identification and warning, and reducing the losses caused by geological disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a composition diagram of the mountain landslide identification and warning system based on remote sensing satellite images of the present invention;

[0046] Figure 2 It is a flowchart of the mountain landslide identification and warning method based on remote sensing satellite images of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] The following will combine the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0048] In order to solve the technical problems in the current technology that the accuracy and stability of some monitoring devices in the mountain landslide identification and warning system need to be improved, especially in complex geological environments, the accuracy and reliability of data are easily interfered; in addition, the automation degree and intelligent level of the warning system still need to be improved to ensure more efficient data processing and warning response, please refer to Figure 1 - Figure 2 , the following technical solutions are provided in this embodiment:

[0049] A landslide identification and early warning system based on remote sensing satellite images, comprising:

[0050] An image acquisition unit, configured to collect historical landslide data of a target mountain area and capture a satellite image of the target mountain area with high resolution through a remote sensing satellite; the image acquisition unit includes:

[0051] An image collection module, configured to determine the scope and monitoring requirements of the target mountain area, and select a suitable remote sensing satellite according to the monitoring requirements to obtain image data of the target mountain area; specifically, the slope is monitored in real time through remote sensing satellite images to identify potential landslide areas; in this embodiment, medium-resolution satellites can be used for periodic scanning, high-resolution satellites can be used for key monitoring, and airborne LiDAR can be used for fine detection. By comprehensively utilizing the advantages of different technologies, the accuracy of identification can be improved.

[0052] A historical data collection module, configured to collect historical landslide data within the target mountain area and save it in time series to provide basic data for subsequent landslide change detection; specifically, the purpose of collecting historical landslide data within the target mountain area is to comprehensively analyze the remotely sensed data to determine whether there are slope anomalies within the target mountain area; specifically, it includes: comprehensively considering data such as remote sensing images, precipitation, groundwater level, seismic waves, and slope displacement, and further establishing a neural network model for data analysis to determine whether there are anomalies.

[0053] An image processing unit, configured to preprocess the satellite image, and the preprocessing includes: removing noise, image enhancement, radiometric correction, and atmospheric correction to ensure the accuracy and reliability of the image data; the image processing unit includes:

[0054] A preliminary processing module, configured to filter out spike noise through Fourier transform filtering, remove strip noise through Fourier transform and low-pass filtering, and adjust the readability and information extraction ability of the satellite image through histogram transformation to improve the contrast and visual effect of the satellite image; specifically, due to the reasons of the sensor, periodic noise, spike noise, and strip noise may appear in the satellite image, and these noises need to be eliminated through band-pass or notch filters to ensure the clarity and accuracy of the satellite image.

[0055] Secondary processing module: Through radiometric calibration, the brightness values in the original remote sensing image are converted into the reflectance or radiance values of the outer atmosphere surface, converting the observation data from the brightness unit of the sensor into physically comparable and analyzable units; then through atmospheric correction, the radiance or reflectance data is converted into the actual reflectance of the ground surface, obtaining the true reflectance, radiance and other physical model parameters of the ground objects; specifically, first, radiometric calibration is to convert the brightness values in the original remote sensing image into the reflectance or radiance values of the outer atmosphere surface, aiming to eliminate the errors caused by factors such as the characteristics of the sensor itself, the spectrum, and the acquisition time; through radiometric calibration, the observation data can be converted from the brightness unit of the sensor into physically interpretable units, making the data comparable and analyzable; second, the task of atmospheric correction is to convert the radiance or reflectance data into the actual reflectance of the ground surface. The gases and particulate matter in the atmosphere will cause changes in the brightness values in the image during the propagation of light. Through atmospheric correction, the influence of factors such as the atmosphere and illumination on the reflection of ground objects can be eliminated, obtaining the true reflectance, radiance and other physical model parameters of the ground objects, which helps to accurately understand the characteristics and changes of the ground surface; finally, the solar altitude angle and terrain correction correct the radiance error caused by the differences in the ground terrain and solar altitude angle through statistical and physical models. The different heights of the ground surface and the changes in the solar altitude angle will affect the propagation and reflection of light. Therefore, these factors are avoided to ensure accurate ground surface information; based on this, it can be known that: Radiometric calibration is a key link in remote sensing image processing, which helps to ensure that the image data collected from satellite sensors is accurate and comparable, and can effectively improve the quality and accuracy of satellite images.

[0056] Feature extraction unit: Based on synthetic aperture radar interferometry technology, landslide features are extracted from satellite images, and the features of terrain changes and surface displacements in the satellite images are analyzed, and then potential landslide areas are identified; the feature extraction unit includes:

[0057] Image selection module: Used to select satellite images of suitable target mountain areas, and after preprocessing the satellite images, the preprocessing includes: removing noise, correcting geometric distortion, etc., to ensure the quality and accuracy of the image data; then according to the early warning requirements and the scope of the target mountain area, the landslide areas that need to be monitored and identified are selected, aiming to concentrate resources and improve the monitoring efficiency and accuracy;

[0058] Image processing module: Used to register the selected satellite images to ensure the spatial consistency of the images at different time points for subsequent analysis; and by comparing the satellite images at different times, image pairs with coherence are found, and appropriate interferometric pairs are selected for interferometric processing;

[0059] A feature extraction module, which is used to perform wavelet transform decomposition on the interference result to extract the temporal information of the ground deformation for analyzing the dynamic changes of the landslide; and analyze the interference result to estimate the topographic phase of the ground, aiming to remove the influence of the atmosphere and other factors; and during the analysis process, select some ground stable points as references to more accurately evaluate the deformation of the landslide.

[0060] A feature determination module, which is used to analyze the topographic phase of the ground to evaluate the features and dynamic changes of the landslide, including: deformation rate, deformation trend, and thus obtain the landslide feature information in the satellite image; since the atmospheric conditions will affect the interference result of the satellite image, it is necessary to adopt appropriate methods to remove the atmospheric phase and residuals to improve the accuracy of the analysis; through the above steps, the synthetic aperture radar interferometry technology can be effectively used to extract the landslide features from the satellite image, providing an important basis for the monitoring and early warning of geological disasters.

[0061] A landslide identification unit, which is used to obtain the real-time mountain data of the potential landslide area through the Beidou satellite technology and sensor technology, establish and train a landslide risk assessment model based on the historical mountain data, perform risk prediction on the real-time mountain data through the landslide risk assessment model, evaluate the landslide risk degree of the potential landslide area, and generate a risk assessment report and feedback it to the human-computer interaction unit; specifically, by deploying Beidou satellite receiving terminals in the potential landslide area, the system can obtain the accurate position information of the potential landslide area in real time; and utilize the high-precision positioning function of the Beidou satellite to enable the system to achieve three-dimensional spatial positioning of the potential landslide area, providing accurate spatial data support for disaster early warning; then deploy a variety of sensors in the potential landslide area, including: rain sensors, displacement sensors, stress sensors, etc.; through the above-mentioned various sensors, the changes in geological environments such as rainfall, surface displacement, and underground stress can be monitored in real time, and the monitoring data is transmitted to the human-computer interaction unit in real time, and then the landslide identification unit imports it into the landslide risk assessment model for risk prediction.

[0062] The landslide identification unit includes:

[0063] A sample processing module, which is used to clean the historical mountain data of the target mountain area and divide the historical mountain data into a training set and a test set; specifically, as can be seen from the above: the historical mountain data includes but is not limited to: remote sensing images, precipitation, groundwater level, seismic waves, and slope displacement, etc. By cleaning the above data, the error data and duplicate data are removed, and the missing data is supplemented to ensure the effectiveness and authenticity of the historical mountain data, and further ensure the accuracy of the neural network model prediction.

[0064] The model construction module is used to create a neural network model, import the training set into the neural network model for training, and obtain the simulation prediction results of the target mountain area; then import the test set into the trained neural network model for verification, and obtain the optimized landslide risk assessment model; specifically, by importing the training set into the neural network model for training, while establishing the characteristic points of the neural network model, obtain the trained landslide risk assessment model, and output the training results, that is: landslide risk prediction results; including any of the following results: no landslide possibility, landslide possibility, and obvious landslide possibility; then import the corresponding test set into the landslide risk assessment model for testing, and output the test results for result verification. When the test results are consistent with the training results, it is determined that the current landslide risk assessment model is the optimal type; when the test results are inconsistent with the training results, it is determined that the current landslide risk assessment model is abnormal, check the accuracy of the training set and the test set, and then import them into the neural network model for training and testing until the verification passes, and obtain the optimized landslide risk assessment model.

[0065] The real-time prediction module is used to import the real-time mountain data of the potential landslide area into the landslide risk assessment model for landslide prediction, obtain the landslide risk degree of the potential landslide area, and generate a risk assessment report; specifically, after cleaning the real-time mountain data, import it into the optimized landslide risk assessment model for risk prediction, and output the prediction results, analyze the prediction results to judge whether a landslide will occur in the potential landslide area, and feedback the prediction results to the human-computer interaction unit in real time for early warning.

[0066] The result feedback module is used to feedback the risk assessment report to the human-computer interaction unit in real time through wireless communication technology; the wireless communication technology includes but is not limited to: WIFI, GPRS, etc.

[0067] The human-computer interaction unit is used to receive and display the mountain data and landslide risk assessment report of the target mountain area, and issue corresponding control instructions according to the data and report; the human-computer interaction unit includes:

[0068] The data receiving module is used to receive and display the mountain data and landslide risk assessment report of the target mountain area in real time; specifically, by installing a display terminal adapted to it in the customer center and logging in to the identification and early warning system on its terminal, aiming to display the mountain data and landslide risk assessment report of the current target mountain area, and issue corresponding instructions according to the data situation, such as: after discovering a potential landslide area, issue an instruction to collect data on this area, and after the data collection, issue an instruction to perform risk prediction on this data, so as to improve the automation and intelligence level of the early warning system, ensure more efficient data processing and early warning response; and timely release early warning information to notify the residents and relevant departments in the affected area to take countermeasures.

[0069] To better demonstrate the operation process of the landslide identification and early warning system based on remote sensing satellite images, the present invention provides a landslide identification and early warning method based on remote sensing satellite images, including the following steps:

[0070] Step 1: Obtain high-resolution satellite images of the target mountain area through remote sensing satellites, and collect historical landslide data of the target mountain area;

[0071] Step 2: After preprocessing the satellite images, use synthetic aperture radar interferometry to extract landslide features from the satellite images, and analyze and identify potential landslide areas in the target mountain area; among them, using synthetic aperture radar interferometry to extract landslide features from the satellite images specifically includes the following steps:

[0072] Select the landslide areas that need to be monitored and identified key points according to the early warning requirements and the scope of the target mountain area, register the selected satellite images to ensure the spatial consistency of the images at different time points; find the coherent image pairs by comparing the satellite images at different times to obtain suitable interference pairs; perform wavelet transform decomposition on the interference results to extract the temporal information of surface deformation; select a part of the ground stable points as references, and estimate the terrain phase of the surface by analyzing the interference results; analyze the processing results to evaluate the characteristics and dynamic changes of the landslide, including: deformation rate, deformation trend, to obtain the landslide feature information in the satellite images.

[0073] Analyze and identify potential landslide areas in the target mountain area, specifically including the following steps:

[0074] Based on the landslide feature information in the satellite images, perform zonal identification on the satellite images to determine the remote sensing sub-images with landslide terrain; and based on the geological state information of the landslide terrain, judge whether the mudslides occur and obtain the mudslide movement information; detect the change of the vegetation index by using the normalization method by comparing the multi-temporal remote sensing images before and after the landslide to determine the potential landslide areas; and combine the geometric rules of the satellite images to exclude non-landslide parts such as roads, buildings, and bare land to complete the fine identification of potential landslide areas; specifically, the remote sensing monitoring technology performs zonal identification on the satellite remote sensing images to determine the areas with landslide terrain, and based on this, judges the occurrence of mudslides, so as to quickly predict whether the target area will be affected by mudslides; this method can not only reduce the manpower and material resources required for landslide exploration, but also improve the accuracy and reliability of identification and early warning.

[0075] Step 3: Build a landslide risk assessment model, clean the historical mountain data, and divide it into a training set and a test set. Import the training set and the test set into the landslide risk assessment model in sequence for risk training to obtain an optimized landslide risk assessment model;

[0076] Step 4: Based on the landslide identification results, obtain the real-time mountain data of the potential landslide area through Beidou satellite technology and sensor technology. Use the landslide risk assessment model to predict the real-time mountain data to obtain the degree of landslide risk in the potential landslide area and generate a risk assessment report. Among them, after using the landslide risk assessment model to predict the real-time mountain data, the following steps are specifically included:

[0077] Use Beidou satellite technology and sensor technology to continuously monitor the mountains in the potential landslide area, and combine visual interpretation and field investigation to verify and correct the evaluation results of the landslide risk assessment model. The field investigation includes: on-site inspection of the landslide site to confirm the type, scale, and cause of the landslide

[0078] Step 5: Based on the wireless transmission technology, the risk assessment report is fed back to the human-computer interaction unit in real time for early warning. The human-computer interaction unit timely issues early warning information to notify the residents and relevant departments in the affected area to take countermeasures.

[0079] Beneficial effects achieved by the above content: Through the above operations, satellite remote sensing images and ground monitoring data are fully integrated to achieve real-time monitoring and early warning of landslides, improve the accuracy and reliability of identification and early warning, and reduce the losses caused by geological disasters.

[0080] Working principle: Obtain satellite images of the target mountain area through remote sensing satellites, use synthetic aperture radar interferometry technology to extract landslide features from the satellite images, and analyze and identify potential landslide areas in the target mountain area; build a landslide risk assessment model, obtain real-time mountain data of the potential landslide area through Beidou satellite technology and sensor technology, use the landslide risk assessment model to predict the real-time mountain data to obtain the degree of landslide risk in the potential landslide area, generate a risk assessment report and feed it back to the human-computer interaction unit for early warning, and the human-computer interaction unit timely issues early warning information to notify the residents and relevant departments in the affected area to take countermeasures.

[0081] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or elements inherent to such process, method, article or device.

[0082] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

Claims

1. A landslide identification and early warning system based on remote sensing satellite images, characterized in that: include: An image acquisition unit is used to collect historical mountain data of the target mountain area, and photograph the target mountain area through a remote sensing satellite to obtain a high-resolution satellite image of the target mountain area; An image processing unit is used to pre-process satellite images, including: noise removal, image enhancement, radiation correction and atmospheric correction; The feature extraction unit extracts landslide features from satellite images based on synthetic aperture radar interferometry technology, analyzes the characteristics of terrain changes and surface displacements in satellite images, and then identifies potential landslide areas; The landslide identification unit is used to obtain real-time mountain data of potential landslide areas through Beidou satellite technology and sensor technology, and to establish and train a landslide risk assessment model based on historical mountain data. The landslide risk assessment model is used to predict the risk of real-time mountain data, assess the degree of landslide risk in potential landslide areas, and generate a risk assessment report to be fed back to the human-computer interaction unit; The human-computer interaction unit is used to receive and display the mountain data and landslide risk assessment report of the target mountain area, and issue corresponding control instructions based on the data and reports.

2. The landslide identification and early warning system based on remote sensing satellite images according to claim 1 is characterized by: The image acquisition unit comprises: Image acquisition module, used to determine the scope and monitoring requirements of the target mountain area, and select appropriate remote sensing satellites to obtain image data of the target mountain area according to the monitoring requirements; The historical data collection module is used to collect historical mountain data in the target mountain area and save it in time series to provide basic data for subsequent mountain change detection.

3. The landslide identification and early warning system based on remote sensing satellite images according to claim 1 is characterized in that: The image processing unit comprises: A preliminary processing module is used to process spike noise through Fourier transform filtering, remove stripe noise through Fourier transform and low-pass filtering, and adjust the readability and information extraction ability of satellite images through histogram transformation method to improve the contrast and visual effect of satellite images; The secondary processing module converts the brightness values ​​in the original remote sensing image into the reflectivity or radiation brightness value of the outer surface of the atmosphere through radiation calibration, so that the observation data is converted from the brightness unit of the sensor to a physically comparable and analyzable unit; and then converts the radiation brightness or reflectivity data into the actual reflectivity of the surface through atmospheric correction to obtain the true reflectivity, radiation rate and other physical model parameters of the ground object.

4. The landslide identification and early warning system based on remote sensing satellite images according to claim 1 is characterized in that: The feature extraction unit comprises: The image selection module is used to select satellite images of appropriate target mountain areas, and after preprocessing the satellite images, select the landslide areas that need to be monitored and identified based on the warning requirements and the scope of the target mountain area; The image processing module is used to register the selected satellite images and select the appropriate interferometric pairs by comparing satellite images taken at different times; The feature extraction module is used to perform wavelet change decomposition on the interference results, extract the time series information of the surface deformation, analyze the interference results, and estimate the terrain phase of the surface; The feature establishment module is used to analyze the terrain phase of the surface, evaluate the characteristics and dynamic changes of the landslide, and obtain the landslide feature information in satellite images.

5. The landslide identification and early warning system based on remote sensing satellite images according to claim 1 is characterized by: The landslide identification unit comprises: The sample processing module is used to clean the historical mountain data of the target mountain area and divide the historical mountain data into a training set and a test set; The model building module is used to create a neural network model, import the training set into the neural network model for training, and obtain the simulation prediction results of the target mountain area; then import the test set into the trained neural network model for verification, and obtain the optimized landslide risk assessment model; The real-time prediction module is used to import the real-time mountain data of the potential landslide area into the landslide risk assessment model for landslide prediction, obtain the landslide risk level of the potential landslide area, and generate a risk assessment report; The result feedback module is used to feed back the risk assessment report to the human-computer interaction unit in real time through wireless communication technology.

6. The landslide identification and early warning system based on remote sensing satellite images according to claim 1 is characterized in that: The human-computer interaction unit comprises: The data receiving module is used to receive and display the mountain data and landslide risk assessment report of the target mountain area in real time.

7. A method for identifying and warning landslides based on remote sensing satellite images, implemented based on the system for identifying and warning landslides based on remote sensing satellite images as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: Obtain high-resolution satellite images of the target mountain area through remote sensing satellites, and collect historical mountain data of the target mountain area; Step 2: After preprocessing the satellite image, the synthetic aperture radar interferometry technique is used to extract landslide features from the satellite image, and the potential landslide areas in the target mountain area are analyzed and identified; Step 3: Build a landslide risk assessment model, clean the historical mountain data, and divide it into a training set and a test set. Import the training set and the test set into the landslide risk assessment model in turn for risk training to obtain an optimized landslide risk assessment model. Step 4: Based on the landslide identification results, the real-time mountain data of the potential landslide area is obtained through Beidou satellite technology and sensor technology. The landslide risk assessment model is used to predict the real-time mountain data, the degree of landslide risk in the potential landslide area is obtained, and a risk assessment report is generated; Step 5: Based on wireless transmission technology, the risk assessment report is fed back to the human-computer interaction unit in real time for early warning. The human-computer interaction unit promptly releases early warning information to notify residents and relevant departments in the affected areas to take countermeasures.

8. The landslide identification and early warning method based on remote sensing satellite images according to claim 7 is characterized in that: In the step 2, the landslide features are extracted from the satellite image using synthetic aperture radar interferometry technology, which specifically includes the following steps: According to the early warning needs and the scope of the target mountain area, the landslide area that needs to be monitored and identified is selected, and the selected satellite images are aligned; by comparing satellite images at different times, coherent image pairs are found to obtain suitable interference pairs; the interference results are decomposed by wavelet transform to extract the time series information of surface deformation; a part of ground stable points are selected as reference, and the terrain phase of the surface is estimated by analyzing the interference results; the processing results are analyzed to evaluate the characteristics and dynamic changes of the landslide, including: deformation rate, deformation trend, and obtain the characteristic information of the landslide in the satellite image.

9. The landslide identification and early warning method based on remote sensing satellite images according to claim 7 is characterized in that: In the second step, the potential landslide area in the target mountain area is analyzed and identified, which specifically includes the following steps: Based on the landslide characteristic information in the satellite image, the satellite image is identified in a zoned manner to determine the remote sensing sub-images where the landslide terrain exists; and based on the geological status information of the landslide terrain, it is judged whether a mud dumping event has occurred and the mud dumping movement information is obtained; by comparing the multi-phase remote sensing images before and after the landslide, the change of the vegetation index is detected using the normalization method to determine the potential landslide area; and combined with the geometric rules of the satellite image, the non-landslide parts of roads, buildings and bare land are excluded to complete the fine identification of the potential landslide area.

10. The landslide identification and early warning method based on remote sensing satellite images according to claim 8, characterized in that: In the step 4, after the real-time mountain data is predicted using the landslide risk assessment model, the following steps are specifically included: BeiDou satellite technology and sensor technology are used to continuously monitor the mountains in potential landslide areas, and the assessment results of the landslide risk assessment model are verified and revised in combination with visual interpretation and field investigation. The field investigation includes: on-site inspection of the landslide site to confirm the type, scale and cause of the landslide.

Images