Plateau mountain geological disaster prediction method and device based on millimeter wave SAR

Abstract

The invention relates to the technical field of disaster identification, and particularly discloses a plateau mountain geological disaster prediction method and device based on a millimeter wave SAR, and the method comprises the steps: obtaining the complete polarization SAR remote sensing information and large-scale DEM data of a plateau mountain; geological analysis information is generated based on the fully polarized SAR remote sensing information and the large-scale DEM data; performing feature analysis on typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features; disaster inversion analysis is executed based on the typical disaster features, and a disaster interpretation model is generated; and predicting the geological disaster of the plateau mountain land based on the disaster interpretation model, and generating a prediction result. The original fully-polarized SAR remote sensing image is enhanced and corrected by using the multi-source heterogeneous data, and the mountain disaster is accurately analyzed by using the constructed disaster interpretation model, so that the recognition accuracy of the mountain disaster is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster identification, and in particular to a method for predicting plateau mountain geological disasters based on millimeter wave SAR and a device for predicting plateau mountain geological disasters based on millimeter wave SAR. Background Art

[0002] The western region has a undulating terrain, intense tectonic activity, and complex and changeable strata, lithology, and climate. Mountain disasters such as large-scale landslides, high-altitude dangerous rock falls, mudslides, debris slopes, water damage, avalanches, ice disasters, and high and steep rock slopes during the growing season are well developed. These disasters are large in scale, highly destructive, and occur frequently. In addition, the climate changes in various ways, making it impossible to quickly obtain the disaster terrain and geological characteristics when a disaster occurs, providing timely, accurate, and high-precision geographical and geological data support for disaster prevention, mitigation, and rescue, which has a great impact on disaster prevention and mitigation decision-making.

[0003] In order to reduce the losses caused by natural disasters, technicians identify and warn of natural disasters. With the continuous development of science and technology, technicians identify natural disasters through satellite images. In the prior art, technicians obtain remote sensing images through conventional means, such as using synthetic aperture radar (SAR) remote sensing technology to obtain remote sensing images, and extract image features to identify possible natural disasters. However, in the actual application process, technicians found that the prior art has at least the following technical problems:

[0004] When SAR remote sensing technology is applied to specific areas, such as plateau mountainous areas, the complex and changeable terrain, climate and forest coverage in plateau mountainous areas will greatly restrict the SAR remote sensing interpretation of geological structure and geological disasters, thereby reducing the accuracy of natural disaster identification, and failing to quickly and accurately identify typical mountain disasters, and failing to meet actual needs. Summary of the invention

[0005] In order to overcome the above-mentioned technical problems existing in the prior art, the embodiment of the present invention provides a method and device for predicting plateau mountain geological disasters based on millimeter wave SAR. The device can break through the influence of complex weather such as clouds, rain, and fog, and form a complete set of real-time rapid mapping technology and equipment. The device enhances and corrects the original fully polarized SAR remote sensing image by utilizing multi-source heterogeneous data, and uses the constructed disaster interpretation model to accurately analyze mountain disasters, thereby improving the accuracy of mountain disaster identification.

[0006] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a method for predicting geological disasters in plateau mountains based on millimeter wave SAR, and the method includes: obtaining full-polarization SAR remote sensing information and large-scale DEM data for plateau mountains; generating geological analysis information based on the full-polarization SAR remote sensing information and the large-scale DEM data; performing feature analysis on typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features; performing disaster inversion analysis based on the typical disaster features to generate a disaster interpretation model; predicting the geological disasters in the plateau mountains based on the disaster interpretation model to generate prediction results.

[0007] Preferably, the generating of geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data comprises: performing a preprocessing operation on the fully polarized SAR remote sensing information to obtain preprocessed information; extracting key terrain data based on the large-scale DEM data; and performing geo-referencing on the preprocessed information based on the key terrain data to obtain geological analysis information.

[0008] Preferably, the method also includes: before performing feature analysis on the typical mountain disaster remote sensing information, obtaining a backscattering map; enhancing the fully polarized SAR remote sensing information based on the backscattering map to obtain enhanced information; obtaining first object features of different types of objects in the fully polarized SAR remote sensing information, and second object features in the enhanced information; training an integrated multi-attention Pytorch framework based on the first object features and the second object features to generate a corresponding contrast enhancement model; performing feature analysis on the typical mountain disaster remote sensing information based on the contrast enhancement model and the geological analysis information to generate typical disaster features.

[0009] Preferably, the feature analysis of typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features includes: determining disaster information of typical mountain disasters, the disaster information including disaster type and disaster occurrence time period; extracting corresponding disaster remote sensing information from the typical mountain disaster remote sensing information during the disaster occurrence time period based on the disaster type; performing causal mechanism analysis on the disaster remote sensing information to obtain at least one disaster cause; and performing feature extraction on each disaster cause based on the disaster remote sensing information to obtain typical disaster features.

[0010] Preferably, performing disaster inversion analysis based on the typical disaster characteristics to generate a disaster interpretation model includes: obtaining basic meteorological and hydrological data of the target mountain during the disaster occurrence period; extracting hydrological evolution characteristics of the basic meteorological and hydrological data; performing disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics to generate disaster physical evolution characteristics; and training a preset deep learning model based on the disaster physical evolution characteristics to generate a disaster interpretation model.

[0011] Correspondingly, the present invention also provides a device for predicting geological disasters in plateau mountains based on millimeter wave SAR, and the device includes: a data acquisition unit, used to acquire full-polarization SAR remote sensing information and large-scale DEM data for plateau mountains; an analysis unit, used to generate geological analysis information based on the full-polarization SAR remote sensing information and the large-scale DEM data; a feature analysis unit, used to perform feature analysis on typical mountain disaster remote sensing information based on the geological analysis information, and generate typical disaster features; a model construction unit, used to perform disaster inversion analysis based on the typical disaster features, and generate a disaster interpretation model; a disaster prediction unit, used to predict the geological disasters in the plateau mountains based on the disaster interpretation model, and generate prediction results.

[0012] Preferably, the analysis unit includes: a preprocessing module, used to perform preprocessing operations on the fully polarized SAR remote sensing information to obtain preprocessed information; a data extraction module, used to extract key terrain data based on the large-scale DEM data; and an analysis module, used to perform geographic registration on the preprocessed information based on the key terrain data to obtain geological analysis information.

[0013] Preferably, the device also includes an enhancement unit, which is specifically used to: obtain a backscattering map before performing feature analysis on the typical mountain disaster remote sensing information; enhance the fully polarized SAR remote sensing information based on the backscattering map to obtain enhanced information; obtain first object features of different types of objects in the fully polarized SAR remote sensing information, and second object features in the enhanced information; train an integrated multi-attention Pytorch framework based on the first object features and the second object features to generate a corresponding contrast enhancement model; perform feature analysis on the typical mountain disaster remote sensing information based on the contrast enhancement model and the geological analysis information to generate typical disaster features.

[0014] Preferably, the feature analysis unit includes: an information determination module, which is used to determine the disaster information of a typical mountain disaster, wherein the disaster information includes the disaster type and the disaster occurrence period; a remote sensing information extraction module, which is used to extract corresponding disaster remote sensing information from the typical mountain disaster remote sensing information based on the disaster type during the disaster occurrence period; a mechanism analysis module, which is used to perform a causal mechanism analysis on the disaster remote sensing information to obtain at least one disaster cause; and a feature analysis module, which is used to perform feature extraction on each disaster cause based on the disaster remote sensing information to obtain typical disaster features.

[0015] Preferably, the model building unit includes: a hydrological data acquisition module, used to obtain the basic meteorological and hydrological data of the target mountain during the disaster period; an evolution feature extraction module, used to extract the hydrological evolution characteristics of the meteorological and hydrological basic data; an inversion analysis module, used to perform disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics, and generate disaster physical evolution characteristics; a model building module, used to train a preset deep learning model based on the disaster physical evolution characteristics, and generate a disaster interpretation model.

[0016] Through the technical solution provided by the present invention, the present invention has at least the following technical effects:

[0017] Based on the acquired high-precision fully polarized SAR remote sensing images, we further optimize and process the multi-source high-precision DEM data and backscatter maps to obtain more accurate geological analysis information. On this basis, we use the disaster interpretation model to identify mountain disasters, track and accurately identify the progress of disaster evolution, disaster types, etc., providing data support for subsequent disaster prevention and treatment, and meeting actual needs.

[0018] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0020] Figure 1 It is a specific implementation flow chart of the method for predicting plateau mountain geological disaster bodies based on millimeter wave SAR provided by an embodiment of the present invention;

[0021] Figure 2 It is a specific implementation flow chart of generating geological analysis information provided by an embodiment of the present invention;

[0022] Figure 3It is a structural schematic diagram of a device for predicting geological disasters in plateau mountains based on millimeter wave SAR provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0024] The terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" means two or more than two. In view of this, "multiple" can also be understood as "at least two" in the embodiments of the present invention. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previous and subsequent associated objects are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present invention, the words "first", "second", etc. are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0025] See also Figure 1 The embodiment of the present invention provides a method for predicting plateau and mountainous geological disasters based on millimeter wave SAR, the method comprising:

[0026] S10) Obtain full-polarization SAR remote sensing information and large-scale DEM data for plateau and mountainous areas;

[0027] S20) generating geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data;

[0028] S30) performing feature analysis on typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features;

[0029] S40) performing disaster inversion analysis based on the typical disaster characteristics to generate a disaster interpretation model;

[0030] S50) Predicting the geological disasters in the plateau and mountainous areas based on the disaster interpretation model to generate prediction results.

[0031] In a possible implementation, firstly obtain full polarization SAR remote sensing information and large-scale DEM data for the target mountain. The full polarization SAR remote sensing information is a millimeter wave SAR remote sensing image. In the embodiment of the present invention, the target mountain is a plateau mountain. The full polarization SAR remote sensing image of the target mountain is obtained by using the Gaofen-3 (GF-3) satellite, and the large-scale DEM data is obtained by high-resolution commercial satellites such as WorldView and QuickBird. The large-scale DEM data is, for example, DEM data with a scale of at least 1:10000, and its data accuracy can reach an accuracy of 5m or more. After obtaining the above data, it is parsed and data fused.

[0032] See also Figure 2 In an embodiment of the present invention, the generating of geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data includes:

[0033] S21) performing a preprocessing operation on the fully polarized SAR remote sensing information to obtain preprocessed information;

[0034] S22) extracting key terrain data based on the large-scale DEM data;

[0035] S23) Geo-referencing the pre-processed information based on the key terrain data to obtain geological analysis information.

[0036] Specifically, the fully polarized SAR remote sensing image is first preprocessed, and the preprocessing operation includes, for example, four steps: polarization matrix generation, polarization filtering, polarization decomposition, and geocoding. In the process of generating the polarization matrix, each pixel of the fully polarized SAR remote sensing image is first subjected to matrix conversion processing. Specifically, the information of each pixel in the polarized radar image can be represented by the Sinclair scattering matrix.

[0037] (1)

[0038] Where S is a 2×2 scattering matrix describing the scattering phenomenon of the incident wave by the scatterer. Represents the target backscatter coefficient of the transmitted horizontal H polarized wave and the received vertical V polarized wave. According to the above matrix, the fully polarized SAR remote sensing image is converted into the corresponding coherence matrix T3, for example, represented as:

[0039] (2)

[0040] Since there is image interference, such as coherent speckle, in the imaging process, in order to improve the accuracy of the image, it is filtered. In the embodiment of the present invention, the image is filtered using an Improved Lee sigma filter to ensure the clarity and polarization information of the polarimetric SAR image.

[0041] After the polarization filtering is completed, polarization decomposition is performed on it, for example, polarization analysis is performed on the filtered image based on the Cloude-Pottier decomposition method, and the corresponding decomposition parameters are obtained, which include but are not limited to entropy (Entropy, H), scattering angle (Scattering Angle, α) and anisotropy (Anisotropy, A) parameters. Then, geocoding is performed to complete the preprocessing operation of the fully polarized SAR remote sensing information and obtain the corresponding preprocessing information.

[0042] Due to the imaging mechanism of radar, terrain factors will cause geometric distortion of the image and have a certain impact on the backscatter received by the radar. Therefore, it is impossible to accurately obtain the image information of the target mountain using only fully polarized SAR remote sensing images. Therefore, large-scale DEM data is further used to georeference the remote sensing images. Specifically, large-scale DEM data is used to extract key terrain data such as slope, slope direction, elevation, and the local incident angle data derived from it are combined with the above preprocessing information to eliminate the above distortion effects in the remote sensing images, and finally generate geological analysis information that can accurately represent the actual geographic information, such as a multi-dimensional feature vector composed of polarization features and key terrain data. At this time, characteristic analysis of typical mountain disasters can be performed.

[0043] In actual applications, in the fully polarized SAR remote sensing images obtained, due to the influence of illumination, dielectric constant, chemical composition, etc., the image edges of different types of objects may be relatively similar, so their boundaries cannot be clearly distinguished, which affects the subsequent precise analysis of disaster characteristics and disaster evolution analysis.

[0044] In order to solve the above technical problems, in an embodiment of the present invention, the method also includes: before performing feature analysis on the typical mountain disaster remote sensing information, obtaining a backscattering map; enhancing the fully polarized SAR remote sensing information based on the backscattering map to obtain enhanced information; obtaining first object features of different types of objects in the fully polarized SAR remote sensing information, and second object features in the enhanced information; training an integrated multi-attention Pytorch framework based on the first object features and the second object features to generate a corresponding contrast enhancement model; performing feature analysis on the typical mountain disaster remote sensing information based on the contrast enhancement model and the geological analysis information to generate typical disaster features.

[0045] In a possible implementation, after determining the geological analysis information, a backscatter map is further obtained, and the original fully polarized SAR remote sensing image is enhanced through the backscatter map to obtain an enhanced image, through which different types of land objects can be more clearly identified. Specifically, the corresponding backscatter coefficient can be determined according to the backscatter map, and the backscatter intensity, surface roughness, dielectric constant, penetration scattering parameters, and attenuation coefficient of the target mountain area can be analyzed and calculated according to the backscatter coefficient. According to the above parameters, the fully polarized SAR remote sensing image is enhanced to obtain enhanced information. After enhancement, the distinctiveness and image features between different land objects can be effectively increased, which is convenient for the subsequent extraction and identification of the boundaries of different land objects, and improves the accuracy of subsequent disaster evolution analysis.

[0046] Based on the above processing, on the one hand, the features of different types of objects in the fully polarized SAR remote sensing image are extracted to obtain the corresponding first object features, and the features of different types of objects in the enhanced image are extracted to obtain the corresponding second object features. At this time, the first object features and the second object features are input into the Pytorch framework with integrated multi-attention for learning, so as to learn the feature enhancement process of different types of objects and train to obtain the contrast enhancement model. Before the subsequent analysis of mountain disaster features, the mountain disaster remote sensing image is first contrast enhanced by the contrast enhancement model, so that the output mountain disaster remote sensing image has better contrast characteristics and is more conducive to disaster identification.

[0047] In the embodiment of the present invention, by further enhancing the remote sensing image in combination with the backscatter map based on the combination of the fully polarized SAR remote sensing image with the large-scale DEM data, the various ground objects in the remote sensing image are more clearly distinguished from each other, which is convenient for subsequent feature recognition and disaster analysis, and improves the accuracy of disaster identification and determination.

[0048] In order to ensure the accuracy in the subsequent mountain disaster identification process, historical typical mountain disasters can be analyzed to generate a corresponding disaster interpretation model. In the subsequent mountain disaster identification process, the mountain disaster interpretation model can be directly called to quickly and accurately identify the disaster situation of the target mountain.

[0049] In an embodiment of the present invention, the feature analysis of typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features includes: determining disaster information of typical mountain disasters, the disaster information including disaster type and disaster occurrence time period; extracting corresponding disaster remote sensing information from the typical mountain disaster remote sensing information during the disaster occurrence time period based on the disaster type; performing causal mechanism analysis on the disaster remote sensing information to obtain at least one disaster cause; and performing feature extraction on each disaster cause based on the disaster remote sensing information to obtain typical disaster features.

[0050] In a possible implementation, since mountain disasters include multiple types, such as but not limited to landslide disasters, debris flow disasters, glacier disasters, ice and snow cover disasters, and the causes of each disaster are different, they need to be processed separately. Specifically, when determining the characteristics of typical disasters, the disaster information of typical mountain disasters is first determined. The disaster information includes the type of disaster, the time period of disaster occurrence, and other information of each mountain disaster. In the embodiment of the present invention, since different typical disasters require different time from condition accumulation to disaster occurrence, their disaster occurrence time periods are also different. For example, in this embodiment, for landslide disasters, the one-year period before the disaster occurs is used as the disaster occurrence time period; for ice and snow cover disasters, the one-month period before the disaster occurs is used as the disaster occurrence time period.

[0051] At this time, typical mountain disaster remote sensing information is retrieved from the database, and according to each disaster type, the disaster remote sensing information within the corresponding disaster occurrence period is obtained from the typical mountain disaster remote sensing information. Specifically, the disaster remote sensing images within the corresponding disaster occurrence period are extracted, and then the causal mechanism of the disaster remote sensing images is analyzed. During the analysis process, it is found that the geographical features associated with the occurrence of mountain disasters include but are not limited to topographic and geomorphic features (rugged terrain in mountainous areas, high mountains and steep slopes, which are prone to landslides, collapses and other gravity disasters), geological structure characteristics (variable strata, complex lithology, strong rock weathering, developed joints, broken structures, forming weak structural surfaces, prone to geological disasters), precipitation characteristics (precipitation is concentrated in mountainous areas, especially High-intensity, long-term rainstorms are the triggering factors of disasters such as landslides and mud-rock flows), soil erosion characteristics (severe soil erosion causes the slopes to lose stability, further triggering disasters such as collapse and landslides), vegetation destruction characteristics (due to deforestation and the decline in forest coverage, the surface structure and ecological environment have been damaged, increasing the risk of mountain torrents and geological disasters), seismic activity characteristics (earthquakes can trigger mountain disasters such as landslides, collapses, mudslides, etc., especially in areas with active geological structures), climate change characteristics (the increase in extreme weather events caused by climate change may increase the probability and intensity of disasters in mountainous areas) and human activity characteristics (including geological disasters caused by human activities such as engineering construction and slope changes).

[0052] In the specific implementation process, the parameter characteristics of the above-mentioned characteristics that have correlated changes during the evolution of the disaster are analyzed, and the above-mentioned correlated parameters are determined as the causes of the disaster based on data such as the intensity of their characteristic changes and the amplitude of parameter changes. Feature extraction is then performed based on the above-mentioned determined change data to obtain typical disaster characteristics.

[0053] After obtaining the above-mentioned typical disaster characteristics, a disaster inversion analysis is performed on them. In an embodiment of the present invention, the disaster inversion analysis is performed based on the typical disaster characteristics to generate a disaster interpretation model, including: obtaining the basic meteorological and hydrological data of the target mountain during the disaster period; extracting the hydrological evolution characteristics of the basic meteorological and hydrological data; performing a disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics to generate disaster physical evolution characteristics; and training a preset deep learning model based on the disaster physical evolution characteristics to generate a disaster interpretation model.

[0054] In one possible implementation, the basic meteorological and hydrological data of the target mountain during the disaster period are first obtained, and the basic meteorological and hydrological data include but are not limited to precipitation, temperature, humidity, air pressure, wind speed, wind direction, evaporation, runoff and soil moisture content, and then the hydrological evolution characteristics of the above-mentioned basic meteorological and hydrological data are extracted. At this time, the evolution of the structural attributes, lithological attributes, hydrological attributes, etc. of the target mountain is further analyzed in combination with the typical disaster characteristics obtained above, and the corresponding structural attribute evolution information, lithological attribute evolution information and hydrological attribute evolution information are generated. The structural attribute evolution characteristics, lithological attribute evolution characteristics and hydrological attribute evolution characteristics corresponding to different typical mountain disasters are further extracted, that is, the physical evolution characteristics of the disaster are obtained. At this time, the above-mentioned characteristic data are input into a pre-set deep learning model, for example, the deep learning model is a CNN neural network model, and the CNN model performs disaster inversion analysis on the above-mentioned data to generate a corresponding disaster interpretation model.

[0055] In the subsequent prediction process of geological disasters in plateau mountains, by real-time collection of fully polarized SAR remote sensing information in plateau mountains, the fully polarized SAR remote sensing information is combined with large-scale DEM data and backscatter maps for enhancement and geographic correction, and then input into the disaster interpretation model. It can quickly and accurately identify whether the plateau mountain area is evolving towards a certain specific geological disaster, as well as identify what type of disaster may occur and to which stage the disaster has evolved, and generate corresponding prediction information in a timely manner, thereby providing managers with accurate decision-making basis, realizing early prediction and prevention of disasters, and ensuring people's personal and property safety.

[0056] The millimeter wave SAR-based rapid identification device for geological disaster bodies provided by an embodiment of the present invention is described below in conjunction with the accompanying drawings.

[0057] See also Figure 3 Based on the same inventive concept, an embodiment of the present invention provides a device for predicting geological disasters in plateau mountains based on millimeter wave SAR, and the device includes: a data acquisition unit, used to acquire full-polarization SAR remote sensing information and large-scale DEM data for plateau mountains; an analysis unit, used to generate geological analysis information based on the full-polarization SAR remote sensing information and the large-scale DEM data; a feature analysis unit, used to perform feature analysis on typical mountain disaster remote sensing information based on the geological analysis information, and generate typical disaster features; a model construction unit, used to perform disaster inversion analysis based on the typical disaster features, and generate a disaster interpretation model; a disaster prediction unit, used to predict the geological disasters in the plateau mountains based on the disaster interpretation model, and generate prediction results.

[0058] In an embodiment of the present invention, the analysis unit includes: a preprocessing module, which is used to perform preprocessing operations on the fully polarized SAR remote sensing information to obtain preprocessed information; a data extraction module, which is used to extract key terrain data based on the large-scale DEM data; and an analysis module, which is used to perform geographic alignment on the preprocessed information based on the key terrain data to obtain geological analysis information.

[0059] In an embodiment of the present invention, the device also includes an enhancement unit, which is specifically used to: obtain a backscattering map before performing feature analysis on the typical mountain disaster remote sensing information; enhance the fully polarized SAR remote sensing information based on the backscattering map to obtain enhanced information; obtain first object features of different types of objects in the fully polarized SAR remote sensing information, and second object features in the enhanced information; train an integrated multi-attention Pytorch framework based on the first object features and the second object features to generate a corresponding contrast enhancement model; perform feature analysis on the typical mountain disaster remote sensing information based on the contrast enhancement model and the geological analysis information to generate typical disaster features.

[0060] In an embodiment of the present invention, the feature analysis unit includes: an information determination module, which is used to determine the disaster information of a typical mountain disaster, wherein the disaster information includes the disaster type and the disaster occurrence period; a remote sensing information extraction module, which is used to extract corresponding disaster remote sensing information from the typical mountain disaster remote sensing information based on the disaster type during the disaster occurrence period; a mechanism analysis module, which is used to perform a causal mechanism analysis on the disaster remote sensing information to obtain at least one disaster cause; and a feature analysis module, which is used to perform feature extraction on each disaster cause based on the disaster remote sensing information to obtain typical disaster features.

[0061] In an embodiment of the present invention, the model building unit includes: a hydrological data acquisition module, which is used to obtain the basic meteorological and hydrological data of the target mountain during the disaster period; an evolution feature extraction module, which is used to extract the hydrological evolution characteristics of the meteorological and hydrological basic data; an inversion analysis module, which is used to perform disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics to generate disaster physical evolution characteristics; a model building module, which is used to train a preset deep learning model based on the disaster physical evolution characteristics to generate a disaster interpretation model.

[0062] The optional implementation modes of the embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation modes. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical scheme of the embodiments of the present invention, and these simple modifications all belong to the protection scope of the embodiments of the present invention.

[0063] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0064] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0065] In addition, various implementations of the embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.

Claims

1. A method for predicting plateau mountain geological disasters based on millimeter wave SAR, characterized in that: The method comprises: Obtain full-polarization SAR remote sensing information and large-scale DEM data for plateaus and mountains; Generate geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data; Based on the geological analysis information, characteristic analysis is performed on remote sensing information of typical mountain disasters to generate typical disaster characteristics; Perform disaster inversion analysis based on the typical disaster characteristics to generate a disaster interpretation model; Based on the disaster interpretation model, the geological disasters in the plateau and mountainous areas are predicted to generate prediction results.

2. The method according to claim 1, characterized in that The generating of geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data includes: Performing a preprocessing operation on the fully polarized SAR remote sensing information to obtain preprocessing information; Extracting key terrain data based on the large-scale DEM data; The pre-processed information is geo-referenced based on the key terrain data to obtain geological analysis information.

3. The method according to claim 1, characterized in that The method further comprises: Before performing feature analysis on the typical mountain disaster remote sensing information, obtaining a backscattering map; enhancing the fully polarized SAR remote sensing information based on the backscattering image to obtain enhanced information; Acquire first ground object features of different types of ground objects in the fully polarized SAR remote sensing information, and second ground object features in the enhanced information; Based on the first land feature and the second land feature, the Pytorch framework of integrated multi-attention is trained to generate a corresponding contrast enhancement model; Based on the contrast enhancement model and the geological analysis information, characteristic analysis is performed on typical mountain disaster remote sensing information to generate typical disaster characteristics.

4. The method according to claim 1, characterized in that: The characteristic analysis of typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster characteristics includes: Determine disaster information of typical mountain disasters, wherein the disaster information includes disaster type and disaster occurrence period; Extracting corresponding disaster remote sensing information from the typical mountain disaster remote sensing information during the disaster occurrence period based on the disaster type; Analyzing the cause mechanism of the disaster remote sensing information to obtain at least one cause of the disaster; Based on the disaster remote sensing information, feature extraction is performed on each disaster cause to obtain typical disaster features.

5. The method according to claim 4, characterized in that The performing disaster inversion analysis based on the typical disaster characteristics to generate a disaster interpretation model includes: Obtaining basic meteorological and hydrological data of the target mountain area during the period when the disaster occurs; Extracting hydrological evolution characteristics of the meteorological and hydrological basic data; Perform disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics to generate disaster physical evolution characteristics; The preset deep learning model is trained based on the physical evolution characteristics of the disaster to generate a disaster interpretation model.

6. A device for predicting plateau and mountainous geological disasters based on millimeter wave SAR, characterized in that: The device comprises: Data acquisition unit, used to obtain full-polarization SAR remote sensing information and large-scale DEM data for plateaus and mountains; An analysis unit, used for generating geological analysis information based on the fully polarized SAR remote sensing information and the large-scale DEM data; A feature analysis unit, used to perform feature analysis on typical mountain disaster remote sensing information based on the geological analysis information to generate typical disaster features; A model building unit, used to perform disaster inversion analysis based on the typical disaster characteristics and generate a disaster interpretation model; The disaster prediction unit is used to predict the geological disasters in the plateau and mountainous areas based on the disaster interpretation model and generate prediction results.

7. The device according to claim 6, characterized in that The parsing unit comprises: A preprocessing module, used for performing a preprocessing operation on the fully polarized SAR remote sensing information to obtain preprocessing information; A data extraction module, used for extracting key terrain data based on the large-scale DEM data; The analysis module is used to perform geo-reference on the pre-processed information based on the key terrain data to obtain geological analysis information.

8. The device according to claim 6, characterized in that The device further comprises an enhancement unit, which is specifically used for: Before performing feature analysis on the typical mountain disaster remote sensing information, obtaining a backscattering map; enhancing the fully polarized SAR remote sensing information based on the backscattering image to obtain enhanced information; Acquire first ground object features of different types of ground objects in the fully polarized SAR remote sensing information, and second ground object features in the enhanced information; Based on the first land feature and the second land feature, the Pytorch framework of integrated multi-attention is trained to generate a corresponding contrast enhancement model; Based on the contrast enhancement model and the geological analysis information, characteristic analysis is performed on typical mountain disaster remote sensing information to generate typical disaster characteristics.

9. The device according to claim 6, characterized in that The feature analysis unit comprises: An information determination module, used to determine disaster information of typical mountain disasters, wherein the disaster information includes the type of disaster and the time period when the disaster occurs; A remote sensing information extraction module, configured to extract corresponding disaster remote sensing information from the typical mountain disaster remote sensing information during the disaster occurrence period based on the disaster type; A mechanism analysis module, used to perform a cause mechanism analysis on the disaster remote sensing information to obtain at least one cause of the disaster; The feature analysis module is used to extract the features of each disaster cause based on the disaster remote sensing information to obtain typical disaster features.

10. The device according to claim 9, characterized in that The model building unit comprises: A hydrological data acquisition module, used to acquire basic meteorological and hydrological data of the target mountain area during the disaster occurrence period; An evolution feature extraction module, used to extract the hydrological evolution features of the meteorological and hydrological basic data; An inversion analysis module, used for performing disaster inversion analysis based on the typical disaster characteristics and the hydrological evolution characteristics; The model building module is used to train a preset deep learning model based on the physical evolution characteristics of the disaster to generate a disaster interpretation model.

Images