Method and device for identifying landslide hazards of building on cutting slope based on remote sensing image

Abstract

The application relates to the technical field of geological disaster identification, and discloses a cutting slope house building landslide hidden danger identification method and device based on remote sensing images, which comprises the following steps: acquiring remote sensing image data and SAR image data for a target area; analyzing the remote sensing image data to obtain a landslide hidden danger position identification result corresponding to the target area; the landslide hidden danger position identification result comprises position information corresponding to a plurality of landslide hidden danger points of cutting slope house building; based on PS-InSAR technology, the SAR image data is analyzed to obtain a deformation area identification result corresponding to the target area; according to the landslide hidden danger position identification result and the deformation area identification result, landslide hidden danger information corresponding to a plurality of target hidden danger areas in the target area is determined. It can be seen that the landslide hidden danger is analyzed by combining the remote sensing image data and the SAR image data, so that the landslide hidden danger identification efficiency and identification accuracy of cutting slope house building can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geological disaster identification, and in particular to a cutting slope house landslide hazard identification method and device based on remote sensing images. BACKGROUND

[0002] In recent years, with the rapid advancement of urbanization construction in mountainous areas, residents in mountainous and hilly areas often obtain flat sites by cutting slope bodies, and the phenomenon of cutting slope house building is becoming increasingly common. However, improper cutting slope design or construction can easily damage the original geological balance, thereby reducing the stability of the slope body and causing landslide hazards. Due to the strong concealment and high suddenness of such landslide hazards, especially under the action of external forces such as heavy rain or earthquakes, disasters are easily triggered, threatening the safety of residents' lives and property.

[0003] Traditional landslide hazard identification methods mainly rely on manual field investigation, but such methods are not only inefficient, but also difficult to cover large areas, and the accuracy of hazard identification is difficult to guarantee. At present, in the existing intelligent landslide hazard identification technology, machine learning technology can be used to identify landslide hazard points, but this method relies on high-quality labeled data, and the model has low interpretability, which affects the accuracy of landslide hazard identification.

[0004] Therefore, it is particularly important to propose a technical solution that can improve the efficiency and accuracy of cutting slope house landslide hazard identification. SUMMARY

[0005] The present application provides a cutting slope house landslide hazard identification method and device based on remote sensing images, which can improve the efficiency and accuracy of cutting slope house landslide hazard identification.

[0006] To solve the above technical problems, the present application discloses a cutting slope house landslide hazard identification method based on remote sensing images, which comprises:

[0007] Obtaining remote sensing image data corresponding to a target area and SAR image data corresponding to the target area;

[0008] Analyzing the remote sensing image data to obtain a landslide hazard position identification result corresponding to the target area; the landslide hazard position identification result includes position information of a plurality of cutting slope house landslide hazard points;

[0009] Based on PS-InSAR technology, analyzing the SAR image data to obtain a deformation area identification result corresponding to the target area; the deformation area identification result includes a plurality of target deformation areas;

[0010] According to the landslide hazard position recognition result and the deformation area recognition result, a landslide hazard recognition result corresponding to the target area is determined; the landslide hazard recognition result includes landslide hazard information corresponding to a plurality of target hazard areas in the target area.

[0011] As an optional implementation, in the first aspect of the present application, the analysis of the remote sensing image data to obtain the landslide hazard position recognition result corresponding to the target area comprises:

[0012] The remote sensing image data is preprocessed to obtain preprocessed remote sensing image data;

[0013] Based on a pre-set graph patch intelligent extraction algorithm, a graph patch intelligent extraction operation is performed on the preprocessed remote sensing image data to obtain a graph patch extraction result;

[0014] Based on a trained hazard building recognition model, a hazard building recognition operation is performed on the preprocessed remote sensing image data to obtain a hazard building recognition result; wherein the hazard building recognition model is a U-Net model improved by using a transfer learning method;

[0015] According to the graph patch extraction result and the hazard building recognition result, a plurality of landslide hazard points in the target area and position information corresponding to each landslide hazard point are determined.

[0016] As an optional implementation, in the first aspect of the present application, the graph patch intelligent extraction operation performed on the preprocessed remote sensing image data based on the pre-set graph patch intelligent extraction algorithm to obtain the graph patch extraction result comprises:

[0017] A segmentation operation is performed on the preprocessed remote sensing image data to obtain a graph patch segmentation result corresponding to the preprocessed remote sensing image data; the graph patch segmentation result includes a plurality of graph patches;

[0018] Based on a Gabor filter, a texture feature extraction operation is performed on the graph patch segmentation result to obtain a texture feature extraction result; the texture feature extraction result includes texture feature information corresponding to each graph patch;

[0019] For each graph patch, according to the determined normal texture feature distribution, the deviation value of the texture feature information corresponding to the graph patch with respect to the normal texture feature distribution is calculated;

[0020] From all the graph patches, an abnormal graph patch with a deviation value greater than or equal to a pre-set deviation threshold is selected as an abnormal graph patch screening result;

[0021] The graph patch extraction result includes the abnormal graph patch screening result.

[0022] As an optional implementation, in the first aspect of the present application, the SAR image data comprises a plurality of SAR images.

[0023] The PS-InSAR technology is used to analyze the SAR image data to obtain the deformation region identification result corresponding to the target region, and the deformation region identification result comprises:

[0024] The time-space baseline of each SAR image in the SAR image data is calculated, and the master image is determined from the SAR image data according to the time-space baseline; the time-space baseline comprises a time baseline and a space baseline;

[0025] The DEM data matched with the SAR image data is obtained, and the registration operation is performed on the other SAR images and the DEM data according to the master image to obtain the auxiliary image corresponding to the SAR image data and the target DEM data;

[0026] The differential interferometric phase sequence corresponding to the SAR image data is generated according to the master image, all the auxiliary images and the target DEM data;

[0027] The deformation phase sequence corresponding to each target PS point in the SAR image data is estimated according to the differential interferometric phase sequence;

[0028] The deformation information corresponding to each target PS point is calculated according to the deformation phase sequence corresponding to the target PS point; the deformation information comprises a deformation amount and a deformation rate;

[0029] The target deformation region corresponding to the target region is identified according to the deformation information corresponding to each target PS point to obtain the deformation region identification result; each target deformation region in the deformation region identification result is one of a deformation accumulation area, a rapid deformation area and an accelerated deformation area.

[0030] As an optional implementation, in the first aspect of the present application, the target deformation region corresponding to the target region is identified according to the deformation information corresponding to each target PS point to obtain the deformation region identification result, and the deformation region identification result comprises:

[0031] The target PS points with the deformation rate greater than or equal to a preset deformation rate threshold are selected from all the target PS points as first candidate PS points;

[0032] The clustering operation is performed on all the first candidate PS points based on a spatial position condition to obtain a clustering result; the clustering result comprises a plurality of candidate deformation regions;

[0033] According to the deformation rate corresponding to each target PS point, the clustering result and the obtained deformation rate noise level, a rapid deformation region corresponding to the target region is determined.

[0034] As an optional implementation, in the first aspect of the present application, the determination of the rapid deformation region corresponding to the target region according to the deformation rate corresponding to each target PS point, the clustering result and the obtained deformation rate noise level comprises:

[0035] For each candidate deformation region, a first number proportion of the first candidate PS points corresponding to the candidate deformation region in the candidate deformation region is counted;

[0036] According to the obtained deformation rate noise level, the deformation rate corresponding to each target PS point is corrected to obtain a target deformation rate corresponding to each target PS point;

[0037] For each candidate deformation region, the target PS points with the target deformation rate greater than or equal to the preset deformation rate threshold are selected from all the target PS points corresponding to the candidate deformation region as second candidate PS points, and a second number proportion of the second candidate PS points corresponding to the candidate deformation region in the candidate deformation region is counted;

[0038] For each candidate deformation region, when the first number proportion is greater than or equal to a first preset number proportion and the second number proportion is greater than or equal to a second preset number proportion, the candidate deformation region is determined as one of the rapid deformation regions corresponding to the target region.

[0039] As an optional implementation, in the first aspect of the present application, the determination of the landslide hazard identification result corresponding to the target region according to the landslide hazard position identification result and the deformation region identification result comprises:

[0040] An environmental influence factor corresponding to the target region is obtained; the environmental influence factor comprises one or more combinations of rainfall information, geological information and topographic information; the geological information comprises lithology and / or structure; the topographic information comprises one or more combinations of slope, slope height and slope direction;

[0041] According to the environmental influence factor, the landslide hazard position identification result and the deformation region identification result, a target hazard region corresponding to the target region is determined and a landslide hazard prediction result corresponding to each target hazard region is analyzed; the landslide hazard prediction result comprises a landslide occurrence probability;

[0042] According to the landslide hazard prediction result, a hazard influence range corresponding to each target hazard region is determined;

[0043] The landslide hazard information corresponding to each of the target hidden danger areas comprises the landslide hazard prediction result and the hidden danger influence range.

[0044] The second aspect of the present application discloses a device for identifying landslide hazards of houses built on cut slopes based on remote sensing images, which comprises:

[0045] An acquisition module is configured to acquire remote sensing image data corresponding to a target area and SAR image data corresponding to the target area.

[0046] An analysis module is configured to analyze the remote sensing image data to obtain a landslide hazard position identification result corresponding to the target area. The landslide hazard position identification result comprises position information corresponding to a plurality of landslide hazard points of houses built on cut slopes.

[0047] The analysis module is further configured to analyze the SAR image data based on PS-InSAR technology to obtain a deformation area identification result corresponding to the target area. The deformation area identification result comprises a plurality of target deformation areas.

[0048] A determination module is configured to determine a landslide hazard identification result corresponding to the target area according to the landslide hazard position identification result and the deformation area identification result. The landslide hazard identification result comprises landslide hazard information corresponding to a plurality of target hidden danger areas in the target area.

[0049] As an optional implementation, in the second aspect of the present application, the specific manner in which the analysis module analyzes the remote sensing image data to obtain the landslide hazard position identification result corresponding to the target area comprises:

[0050] The remote sensing image data is preprocessed to obtain preprocessed remote sensing image data.

[0051] A graph patch intelligent extraction algorithm is performed on the preprocessed remote sensing image data to obtain a graph patch extraction result.

[0052] A hidden danger building identification model is trained to identify hidden danger buildings in the preprocessed remote sensing image data to obtain a hidden danger building identification result. The hidden danger building identification model is a U-Net model improved by using a transfer learning method.

[0053] According to the graph patch extraction result and the hidden danger building identification result, a plurality of landslide hazard points in the target area and position information corresponding to each of the landslide hazard points are determined.

[0054] As an optional implementation, in the second aspect of the present application, the specific manner in which the analysis module performs the intelligent extraction of the graph patches on the preprocessed remote sensing image data based on the pre-set intelligent extraction algorithm of the graph patches comprises:

[0055] performing a segmentation operation on the preprocessed remote sensing image data to obtain a graph patch segmentation result corresponding to the preprocessed remote sensing image data; the graph patch segmentation result comprises a plurality of graph patches;

[0056] performing a texture feature extraction operation on the graph patch segmentation result based on a Gabor filter to obtain a texture feature extraction result; the texture feature extraction result comprises texture feature information corresponding to each of the graph patches;

[0057] for each of the graph patches, calculating a deviation value of the texture feature information corresponding to the graph patch with respect to the normal texture feature distribution according to the determined normal texture feature distribution;

[0058] selecting, from all the graph patches, an abnormal graph patch whose deviation value is greater than or equal to a pre-set deviation threshold, as an abnormal graph patch selection result;

[0059] wherein the graph patch extraction result comprises the abnormal graph patch selection result.

[0060] As an optional implementation, in the second aspect of the present application, the SAR image data comprises a plurality of SAR images;

[0061] wherein the specific manner in which the analysis module analyzes the SAR image data based on the PS-InSAR technology to obtain the deformation region identification result corresponding to the target region comprises:

[0062] calculating a space-time baseline of each of the SAR images in the SAR image data, and determining a main image from the SAR image data according to the space-time baseline; the space-time baseline comprises a time baseline and a space baseline;

[0063] obtaining DEM data matched with the SAR image data, and performing a registration operation on other SAR images and the DEM data according to the main image to obtain a secondary image corresponding to the SAR image data and target DEM data;

[0064] generating a differential interferometric phase sequence corresponding to the SAR image data according to the main image, all the secondary images and the target DEM data;

[0065] estimating a deformation phase sequence corresponding to each target PS point in the SAR image data according to the differential interferometric phase sequence;

[0066] According to the deformation phase sequence corresponding to each target PS point, deformation information corresponding to the target PS point is calculated; the deformation information includes a deformation amount and a deformation rate;

[0067] According to the deformation information corresponding to each target PS point, a target deformation region corresponding to the target region is identified to obtain a deformation region identification result; each target deformation region in the deformation region identification result is one of a deformation accumulation area, a rapid deformation area, and an accelerated deformation area.

[0068] As an optional implementation, in the second aspect of the present application, the specific manner in which the analysis module identifies the target deformation region corresponding to the target region according to the deformation information corresponding to each target PS point to obtain the deformation region identification result includes:

[0069] From all the target PS points, a target PS point with a deformation rate greater than or equal to a preset deformation rate threshold is selected as a first candidate PS point;

[0070] Based on a spatial position condition, a clustering operation is performed on all the first candidate PS points to obtain a clustering result; the clustering result includes a plurality of candidate deformation regions;

[0071] According to the deformation rate corresponding to each target PS point, the clustering result, and an obtained deformation rate noise level, a rapid deformation region corresponding to the target region is determined.

[0072] As an optional implementation, in the second aspect of the present application, the specific manner in which the analysis module determines the rapid deformation region corresponding to the target region according to the deformation rate corresponding to each target PS point, the clustering result, and the obtained deformation rate noise level includes:

[0073] For each candidate deformation region, a first quantity proportion of the first candidate PS points corresponding to the candidate deformation region in the candidate deformation region is counted;

[0074] According to the obtained deformation rate noise level, the deformation rate corresponding to each target PS point is corrected to obtain a target deformation rate corresponding to each target PS point;

[0075] For each candidate deformation region, a target PS point with a target deformation rate greater than or equal to the preset deformation rate threshold is selected from all the target PS points corresponding to the candidate deformation region as a second candidate PS point, and a second quantity proportion of the second candidate PS points corresponding to the candidate deformation region in the candidate deformation region is counted;

[0076] For each of the candidate deformation regions, when the first quantity ratio is greater than or equal to a first preset quantity ratio and the second quantity ratio is greater than or equal to a second preset quantity ratio, the candidate deformation region is determined as one of the rapid deformation regions corresponding to the target region.

[0077] As an optional implementation, in the second aspect of the present application, the specific manner of determining the landslide hazard identification result corresponding to the target region according to the landslide hazard position identification result and the deformation region identification result comprises:

[0078] obtaining an environmental influence factor corresponding to the target region; the environmental influence factor comprises a combination of one or more of rainfall information, geological information and topographic information; the geological information comprises lithology and / or structure; the topographic information comprises a combination of one or more of slope, slope height and slope direction;

[0079] determining a target hazard region corresponding to the target region according to the environmental influence factor, the landslide hazard position identification result and the deformation region identification result, and analyzing a landslide hazard prediction result corresponding to each of the target hazard regions; the landslide hazard prediction result comprises a landslide occurrence probability;

[0080] determining a hazard influence range corresponding to each of the target hazard regions according to the landslide hazard prediction result;

[0081] wherein the landslide hazard information corresponding to each of the target hazard regions comprises the landslide hazard prediction result and the hazard influence range.

[0082] The third aspect of the present application discloses another landslide hazard identification device for cutting slope building based on remote sensing images, which comprises:

[0083] a memory storing executable program codes;

[0084] a processor coupled with the memory;

[0085] The processor invokes the executable program codes stored in the memory to execute part or all of the steps of the landslide hazard identification method for cutting slope building based on remote sensing images disclosed in the first aspect of the present application.

[0086] The fourth aspect of the present application discloses a computer storage medium storing computer instructions, which are invoked to execute part or all of the steps of the landslide hazard identification method for cutting slope building based on remote sensing images disclosed in the first aspect of the present application.

[0087] Compared with the prior art, the present application has the following beneficial effects:

[0088] In the present application, remote sensing image data corresponding to the target area and SAR image data corresponding to the target area are obtained; the remote sensing image data is analyzed to obtain a landslide hazard location identification result corresponding to the target area; the landslide hazard location identification result includes position information corresponding to a plurality of landslide hazard points related to building on a cut slope; based on PS-InSAR technology, the SAR image data is analyzed to obtain a deformation area identification result corresponding to the target area; the deformation area identification result includes a plurality of target deformation areas; according to the landslide hazard location identification result and the deformation area identification result, a landslide hazard identification result corresponding to the target area is determined; the landslide hazard identification result includes landslide hazard information corresponding to a plurality of target hazard areas in the target area. As can be seen, by analyzing the remote sensing image data corresponding to the target area, a plurality of landslide hazard points related to building on a cut slope in the target area are identified, and by analyzing the SAR image data corresponding to the target area based on PS-InSAR technology, a plurality of target deformation areas in the target area are obtained, and then the landslide hazard location identification result and the deformation area identification result are combined to determine the landslide hazard identification result corresponding to the target area, which can analyze the landslide hazard by combining remote sensing image data and SAR image data, can improve the comprehensiveness and reliability of the analysis of the target area, thereby facilitating the determination accuracy of the landslide hazard area in the target area, and further facilitating the landslide hazard identification efficiency and identification accuracy of building on a cut slope, and further facilitating timely disaster emergency measures to effectively protect the lives and property of residents. BRIEF DESCRIPTION OF DRAWINGS

[0089] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0090] Figure 1 is a flowchart of a landslide hazard identification method for building on a cut slope based on remote sensing images disclosed by an embodiment of the present application;

[0091] Figure 2 is a flowchart of another landslide hazard identification method for building on a cut slope based on remote sensing images disclosed by an embodiment of the present application;

[0092] Figure 3 is a schematic diagram of the principle of transfer learning disclosed by an embodiment of the present application;

[0093] Figure 4 is a flowchart of another landslide hazard identification method for building on a cut slope based on remote sensing images disclosed by an embodiment of the present application;

[0094] Figure 5 This is a schematic diagram of the structure of a landslide hazard identification device based on remote sensing imagery for building houses on slopes, as disclosed in an embodiment of the present invention.

[0095] Figure 6 This is a schematic diagram of another landslide hazard identification device based on remote sensing imagery for building houses on slopes, as disclosed in an embodiment of the present invention. Detailed Implementation

[0096] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0097] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0098] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0099] This invention discloses a method and apparatus for identifying landslide hazards related to slope-cutting housing construction based on remote sensing imagery. It can identify several landslide hazard points related to slope-cutting housing construction within a target area by analyzing remote sensing imagery data corresponding to that area. Furthermore, it analyzes SAR imagery data corresponding to the target area using PS-InSAR technology to obtain several target deformation areas within the target area. By combining the landslide hazard location identification results and deformation area identification results, the landslide hazard identification result corresponding to the target area is determined. This method combines remote sensing imagery data and SAR imagery data to analyze landslide hazards, improving the comprehensiveness and reliability of the analysis of the target area. This enhances the accuracy of identifying potential landslide hazard areas within the target area, thereby improving the efficiency and accuracy of landslide hazard identification for slope-cutting housing construction. Ultimately, this facilitates timely disaster emergency measures to effectively protect the lives and property of residents. Detailed descriptions follow.

[0100] Example 1

[0101] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, as disclosed in an embodiment of the present invention. Figure 1 The described method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery can be applied to a landslide hazard identification device based on remote sensing imagery. This device may include one of the following: a hazard identification terminal, hazard identification equipment, a hazard identification system, and a server. The server may include a local server or a cloud server; this embodiment of the invention does not limit the specific components. Figure 1 As shown, the method for identifying landslide hazards caused by building houses on slopes based on remote sensing imagery may include the following operations:

[0102] 101. Obtain remote sensing image data and SAR image data corresponding to the target area.

[0103] In this embodiment of the invention, the target area is one of the areas in the mountainous or hilly region; for example, the target area can be one of the areas in the Guangxi clastic rock and granite region that is located in the high hilly area with an altitude of 50-200m or the low mountain area with an altitude of 200-500m, and this embodiment of the invention does not limit it.

[0104] In this embodiment of the invention, the remote sensing image data may optionally include optical remote sensing images and / or radar remote sensing images, and this embodiment of the invention does not limit the scope of the data. For example, the remote sensing image data may be collected by the ZY-3 satellite and / or the WorldView-2 satellite, or by other data sources, and this embodiment of the invention does not limit the scope of the data.

[0105] In this embodiment of the invention, optionally, the SAR (Synthetic Aperture Radar) image data may include multiple SAR images; for example, the SAR image data may include SAR up-orbit data with a resolution of 3m collected by the ALOS-2 satellite, and more specifically, it may include 50 repeated orbit SAR images from June 15, 2017 to September 17, 2024. This embodiment of the invention does not limit the scope of the invention.

[0106] 102. Analyze remote sensing image data to obtain the location identification results of landslide hazards corresponding to the target area.

[0107] In this embodiment of the invention, the landslide hazard location identification result includes location information corresponding to several landslide hazard points related to slope cutting for building construction; optionally, the location information corresponding to each landslide hazard point may include the location coordinates of the landslide hazard point and / or the location range of the landslide hazard point, which is not limited in this embodiment of the invention.

[0108] 103. Based on PS-InSAR technology, analyze SAR image data to obtain the deformation region identification results corresponding to the target area.

[0109] In this embodiment of the invention, PS-InSAR (Persistent Scatterer Interferometric Synthetic Aperture Radar) technology can invert deformation information by identifying and selecting coherent point targets with stable scattering characteristics and analyzing the pixel phase of these point-like scatterers. The deformation region identification result corresponding to the target region includes several target deformation regions.

[0110] 104. Based on the landslide hazard location identification results and deformation area identification results, determine the landslide hazard identification results corresponding to the target area.

[0111] In this embodiment of the invention, the landslide hazard identification result includes landslide hazard information corresponding to several target hazard areas within the target area; optionally, each target hazard area can be one of the landslide hazard points or a target deformation area, or it can be an overlapping area of ​​the landslide hazard point and the target deformation area, which is not limited in this embodiment of the invention; optionally, the landslide hazard information corresponding to each target hazard area can be used to indicate the probability of a landslide disaster occurring in the target hazard area and the predicted specific hazard situation, which is not limited in this embodiment of the invention.

[0112] It should be noted that steps 102 and 103 are not sequential; that is, step 102 can occur before or after step 103, or occur simultaneously with step 103. This embodiment of the invention does not impose any limitations.

[0113] As can be seen, the method described in the embodiments of the present invention can identify several landslide hazard points related to slope-cutting construction within the target area by analyzing remote sensing image data corresponding to the target area, and obtain several target deformation areas within the target area by analyzing SAR image data corresponding to the target area based on PS-InSAR technology. Then, by combining the landslide hazard location identification results and deformation area identification results, the landslide hazard identification results corresponding to the target area can be determined. By combining remote sensing image data and SAR image data to analyze landslide hazards, the comprehensiveness and reliability of the analysis of the target area can be improved, thereby improving the accuracy of determining the landslide hazard areas within the target area. This, in turn, improves the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and facilitates timely disaster emergency measures to effectively protect the lives and property of residents.

[0114] In an optional embodiment, analyzing remote sensing image data to obtain the identification results of landslide hazard locations corresponding to the target area may include the following operations:

[0115] Preprocessing of remote sensing image data yields preprocessed remote sensing image data.

[0116] Based on a pre-defined intelligent patch extraction algorithm, intelligent patch extraction is performed on pre-processed remote sensing image data to obtain patch extraction results.

[0117] Based on the trained hazardous building identification model, hazardous building identification is performed on preprocessed remote sensing image data to obtain hazardous building identification results; the hazardous building identification model is a U-Net model improved by the transfer learning method.

[0118] Based on the results of the extracted landforms and the identification of hazardous buildings, several landslide hazard points within the target area and the location information corresponding to each landslide hazard point are determined.

[0119] Optionally, the training dataset used to train the hazardous building identification model may include high-resolution optical remote sensing images of the sampled area (hilly / low mountainous area) before and after slope cutting, taking the clastic rock and granite area of ​​Guangxi as the research object, as well as the corresponding patches of the high-resolution optical remote sensing images before and after slope cutting, and the rock and soil characteristics, vegetation characteristics, elevation change characteristics and historical landslide data of the sampled area. It may also include real data obtained from field surveys. This embodiment of the invention does not limit the data.

[0120] Optionally, the dataset used to train the hazardous building identification model is obtained in the following way: visually interpreting the original remote sensing image data used for training based on ArcGIS software, and determining the real labels corresponding to the original remote sensing image data through annotation, and then converting the vector file obtained through the above steps into a raster file to generate a real surface image marked with the location of buildings. The real surface image is divided into a training set, a validation set, and a test set according to 70%, 15%, and 15%, respectively. This embodiment of the invention does not limit the scope of the invention.

[0121] Alternatively, the pre-processed remote sensing image data that needs to be input into the hazardous building identification model can also be processed in the above manner, and the embodiments of the present invention are not limited thereto.

[0122] The transfer learning employed in the hazardous building identification model refers to the ability of a model trained on one task to be applied to another, thereby accelerating the training of the new task and improving its performance. For details, please refer to [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram illustrating the principle of transfer learning disclosed in an embodiment of the present invention. Optionally, the hazardous building identification model can be pre-trained on a publicly available ImageNet dataset; this embodiment of the present invention does not impose any limitations on this.

[0123] As can be seen, this optional embodiment can perform intelligent patch extraction on the preprocessed remote sensing image data after preprocessing, obtain patch extraction results, and identify potential hazards in the preprocessed remote sensing image data based on a trained hazard building identification model, obtain hazard building identification results. Combining the patch extraction results and the hazard building identification results, several landslide hazard points in the target area and the corresponding location information of each landslide hazard point can be determined. This can improve the comprehensiveness of remote sensing image data processing and analysis, and the combined patch extraction results and hazard building identification results can improve the accuracy and reliability of determining landslide hazard points and related location information for slope-cutting construction, thereby improving the efficiency and accuracy of subsequent landslide hazard identification for slope-cutting construction.

[0124] In this optional embodiment, optionally, performing intelligent patch extraction on preprocessed remote sensing image data based on a pre-set patch intelligent extraction algorithm to obtain patch extraction results may include the following operations:

[0125] A segmentation operation is performed on the preprocessed remote sensing image data to obtain the patch segmentation results corresponding to the preprocessed remote sensing image data; the patch segmentation results include multiple patches;

[0126] Based on the Gabor filter, a texture feature extraction operation is performed on the patch segmentation results to obtain the texture feature extraction results; the texture feature extraction results include the texture feature information corresponding to each patch.

[0127] For each patch, based on the determined normal texture feature distribution, calculate the deviation of the texture feature information corresponding to the patch from the normal texture feature distribution;

[0128] Filter out abnormal patches from all patches whose deviation values ​​are greater than or equal to a preset deviation threshold, and use these as the abnormal patch filtering results.

[0129] The results of the image patch extraction include the results of the abnormal image patch screening.

[0130] Each patch obtained through the segmentation operation can be used to represent various geographic entities. For example, each patch can be used to represent one of buildings, roads, water bodies, and vegetation, or it can be used to represent other types of geographic entities. This embodiment of the invention does not limit the scope of the invention.

[0131] Optionally, the texture feature information may include one or more of texture thickness, texture direction, and texture periodicity, and may also include other types of texture feature information, which are not limited in this embodiment of the invention.

[0132] Optionally, the normal texture feature distribution may include the surrounding patches corresponding to each patch and / or predefined normal patches, which is not limited in this embodiment of the invention.

[0133] As can be seen, this optional embodiment can also obtain the patch segmentation results corresponding to the remote sensing image data by segmenting the preprocessed remote sensing image data, and then extract the texture feature information corresponding to the patch segmentation results based on the Gabor filter to calculate the deviation value of the texture feature information corresponding to each patch from the normal texture feature distribution. Then, abnormal patches with deviation values ​​greater than or equal to a preset deviation threshold are selected from all patches as abnormal patch selection results. This can improve the accuracy of texture feature extraction and analysis of remote sensing image data, thereby improving the accuracy of abnormal patch selection. In turn, it is beneficial to more efficiently and accurately determine the landslide hazard points in the target area based on the accurate patch extraction results.

[0134] In this optional embodiment, after preprocessing the remote sensing image data to obtain preprocessed remote sensing image data, the method may further include the following operations:

[0135] Based on a pre-defined image classification algorithm, image classification operations are performed on pre-processed remote sensing image data to obtain image classification results;

[0136] The segmentation operation performed on the preprocessed remote sensing image data to obtain the corresponding patch segmentation results can include the following operations:

[0137] Based on the image classification results, a segmentation operation is performed on the preprocessed remote sensing image data to obtain the patch segmentation results corresponding to the preprocessed remote sensing image data.

[0138] Optionally, the image classification algorithm may include the GLC decision tree classification algorithm and / or the SVM (Support Vector Machine) classification algorithm, which is not limited in this embodiment of the invention.

[0139] As can be seen, this optional embodiment can first classify the preprocessed remote sensing image data based on a preset classification algorithm, and then perform segmentation operation based on the obtained image classification results, which is beneficial to improving the accuracy and efficiency of patch segmentation of remote sensing image data.

[0140] Example 2

[0141] Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, as disclosed in an embodiment of the present invention. Figure 2 The described method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery can be applied to a landslide hazard identification device based on remote sensing imagery. This device may include one of the following: a hazard identification terminal, hazard identification equipment, a hazard identification system, and a server. The server may include a local server or a cloud server; this embodiment of the invention does not limit the specific components. Figure 2 As shown, the method for identifying landslide hazards caused by building houses on slopes based on remote sensing imagery may include the following operations:

[0142] 201. Obtain remote sensing image data and SAR image data corresponding to the target area.

[0143] In this embodiment of the invention, the SAR image data includes multiple SAR images.

[0144] 202. Analyze the remote sensing image data to obtain the location identification results of the landslide hazard corresponding to the target area.

[0145] In this embodiment of the invention, the landslide hazard location identification result includes location information corresponding to several landslide hazard points related to slope-cutting construction.

[0146] 203. Calculate the spatiotemporal baseline of each SAR image in the SAR image data, and determine the main image from the SAR image data based on the spatiotemporal baseline.

[0147] In this embodiment of the invention, the spatiotemporal baseline of each SAR image includes a temporal baseline and a spatial baseline; optionally, the main image is determined from the SAR image data based on the spatiotemporal baseline, which can be specifically: the SAR image with the center of the temporal baseline and the center of the spatial baseline is selected from all SAR images in the SAR image data as the main image, and this embodiment of the invention does not limit this.

[0148] 204. Obtain DEM data that matches the SAR image data, and perform registration operations on other SAR images and DEM data based on the main image to obtain the auxiliary image and target DEM data corresponding to the SAR image data.

[0149] In this embodiment of the invention, DEM (Digital Elevation Model) data is a type of digital model data used to represent the topographic relief of the earth's surface.

[0150] In this embodiment of the invention, optionally, after step 204, the method may further include the following operations:

[0151] Identify and process geometrically distorted regions in the main image; the geometrically distorted regions include at least one of perspective contraction, shadow, and overlay regions.

[0152] This is able to

[0153] 205. Based on the main image, all auxiliary images, and target DEM data, generate the differential interferometric phase sequence corresponding to the SAR image data.

[0154] 206. Based on the differential interferometric phase sequence, estimate the deformation phase sequence corresponding to each target PS point in the SAR image data.

[0155] 207. Calculate the deformation information corresponding to each target PS point based on the deformation phase sequence.

[0156] In this embodiment of the invention, the deformation information corresponding to each target PS point may include deformation amount and deformation rate.

[0157] 208. Based on the deformation information corresponding to each target PS point, identify the target deformation region corresponding to the target region and obtain the deformation region identification result.

[0158] In this embodiment of the invention, the deformation region identification result may include several target deformation regions; wherein, each target deformation region in the deformation region identification result is one of a deformation cluster region, a rapid deformation region, and an accelerated deformation region; wherein, the deformation of the deformation cluster region is significantly higher than that of the surrounding region, the deformation rate of the rapid deformation region is significantly higher than that of the surrounding region, and the deformation rate of the accelerated deformation region continuously increases over time.

[0159] In this embodiment of the invention, optionally, a multi-track data fusion equation can be used based on the elevation and descent track data of the target area to obtain quasi-three-dimensional deformation rate information of the target area. Then, based on the quasi-three-dimensional deformation analysis results, rapid deformation zones and accelerated deformation zones are automatically identified; this embodiment of the invention does not impose limitations. The multi-track data fusion equation can be [E] = [A]. -1 [D], where [E] is the three-dimensional deformation component, [D] is the observed LOS deformation (i.e., Line of Sight, radar line-of-sight deformation), and [A] is the geometric projection matrix. This embodiment of the invention is not limited.

[0160] 209. Based on the landslide hazard location identification results and deformation area identification results, determine the landslide hazard identification results corresponding to the target area.

[0161] In this embodiment of the invention, the landslide hazard identification result may include landslide hazard information corresponding to several target hazard areas within the target area.

[0162] For further detailed descriptions of steps 201-202 and 209 in this embodiment of the invention, please refer to the detailed descriptions of steps 101-102 and 104 in Embodiment 1. These descriptions will not be repeated in this embodiment of the invention.

[0163] As can be seen, the method described in the embodiments of the present invention can identify several landslide hazard points related to slope-cutting construction within the target area by analyzing remote sensing image data corresponding to the target area, and obtain several target deformation areas within the target area by analyzing SAR image data corresponding to the target area based on PS-InSAR technology. Then, by combining the landslide hazard location identification results and deformation area identification results, the landslide hazard identification results corresponding to the target area can be determined. By combining remote sensing image data and SAR image data to analyze landslide hazards, the comprehensiveness and reliability of the analysis of the target area can be improved, thereby improving the accuracy of determining the landslide hazard areas within the target area. This, in turn, improves the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and facilitates timely disaster emergency measures to effectively protect the lives and property of residents. Furthermore, it can determine the main image based on the spatiotemporal baseline of each SAR image in the SAR image data, and register other SAR images and acquired DEM data based on the main image to obtain auxiliary images and target DEM data. Then, it generates the differential interferometric phase sequence corresponding to the SAR image data to estimate the deformation phase sequence corresponding to each target PS point. Based on the deformation phase sequence, it calculates deformation information, thereby identifying the target deformation region corresponding to the target area and obtaining the deformation region identification result. This can improve the accuracy and comprehensiveness of SAR image data processing, thereby improving the accuracy of PS point deformation analysis, which in turn helps to improve the accuracy of identifying deformation regions contained in the target area. This is beneficial for further improving the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and ultimately for taking more targeted disaster emergency measures in a timely manner.

[0164] In an optional embodiment, registration operations are performed on other SAR images and DEM data based on the main image to obtain auxiliary images and target DEM data corresponding to the SAR image data. This may include the following operations:

[0165] Based on the first registration algorithm and the first registration parameters, and according to the main image, the first registration and cropping operation is performed on other SAR images to obtain the auxiliary image corresponding to the SAR image data.

[0166] Based on the second registration algorithm and the second registration parameters, and according to the main image, the second registration and cropping operation is performed on the DEM data to obtain preprocessed DEM data;

[0167] The preprocessed DEM data is converted into target DEM data in the image coordinate system corresponding to the SAR image data.

[0168] Wherein, the registration error between each auxiliary image and the main image is less than a first preset error. For example, the first preset error can be 0.25 pixels, but this embodiment of the invention does not limit it; and, the area displayed by each auxiliary image and the main image is the same area range.

[0169] Wherein, the registration error between the preprocessed DEM data and the main image is less than the second preset error. For example, the second preset error can be 0.5 pixels, but this embodiment of the invention does not limit it.

[0170] As can be seen, this optional embodiment can also register and crop other SAR images and DEMs based on the main image to obtain auxiliary images and preprocessed DEM data. Then, the preprocessed DEM data is converted to target DEM data in the coordinate system of the SAR image data. This can improve the accuracy of registration and cropping of other SAR images and DEM data in the SAR image data, which is beneficial for generating more accurate and reliable differential interferometric phase sequences based on the accurately registered SAR images and DEM data in subsequent steps. This is conducive to more accurately identifying the target deformation area in the target region.

[0171] In an optional embodiment, generating a differential interferometric phase sequence corresponding to the SAR image data based on the main image, all secondary images, and target DEM data may include the following operations:

[0172] After pre-filtering the main image and all auxiliary images, the main image and all auxiliary images are multiplied by complex conjugate to obtain the interferometric phase corresponding to the SAR image data, so as to generate an interferometric atlas based on time series.

[0173] Based on the amplitude deviation index method and / or signal-to-noise ratio method, the interferometric atlas is analyzed to identify the target PS points corresponding to the SAR image data and generate the interferometric phase sequence corresponding to all target PS points;

[0174] Based on the target DEM data, differential interferometry is performed on the interferometric phase sequence to correct the interferometric phase sequence, resulting in a differential interferometric phase sequence and the target interferogram set corresponding to the differential interferometric phase sequence.

[0175] The target PS point is a permanent scatterer.

[0176] Optionally, differential interferometry is performed on the interferometric phase sequence based on the target DEM data to correct the interferometric phase sequence, resulting in a differential interferometric phase sequence and the corresponding target interferogram set. This can include the following operations:

[0177] Calculate the flat-land phase based on the spatial baseline and Earth ellipsoid parameters;

[0178] Calculate the terrain phase based on the target DEM data;

[0179] The flat terrain phase and topographic phase mentioned above are removed from the interferometric phase sequence to obtain the differential interferometric phase sequence, and the differential interferometric atlas corresponding to the differential interferometric phase sequence is generated.

[0180] Based on the baseline model and star orbit parameters, a coarse estimate of the spatial baseline phase is determined, and then the residual baseline phase is estimated. The residual baseline phase is then fed back into the baseline model to eliminate the unmodeled baseline error in the coarse estimate of the spatial baseline phase, thus obtaining an optimized baseline model.

[0181] Based on the baseline model, residual interference fringes in the differential interferogram are detected to calculate the spatial baseline residual phase in the differential interferogram. The spatial baseline residual phase in the differential interferogram is then removed to obtain the target interferogram.

[0182] This approach allows for the initial rough estimation of the spatial baseline phase based on star orbit parameters, assessing the differences between observations at different times or in space. Then, based on this assessment, the residual baseline phase is estimated to eliminate baseline errors not modeled in the initial rough estimation. Next, by improving the spatial baseline phase, the residual phase is fed back into the baseline model to enhance accuracy. Finally, by correcting the flat-ground phase and generating an improved interferometric atlas, the accuracy of determining the differential interferometric phase sequence and the target interferometric atlas can be improved.

[0183] As can be seen, this optional embodiment can also perform complex conjugate multiplication on the main image and all auxiliary images after pre-filtering to obtain the interferometric phase corresponding to the SAR image data, thereby generating a time-series-based interferometric atlas. Then, the target PS points corresponding to the SAR image data are identified based on the interferometric atlas, and an interferometric phase sequence is generated. Then, differential interferometry is performed on the interferometric phase sequence based on the target DEM data to correct the interferometric phase sequence, thereby obtaining the differential interferometric phase sequence and the target interferometric atlas corresponding to the differential interferometric phase sequence. This can improve the calculation accuracy between the main image and auxiliary images, and between the main image and DEM data, thereby helping to improve the generation accuracy of the differential interferometric phase sequence.

[0184] In an optional embodiment, estimating the deformation phase sequence corresponding to each target PS point in the SAR image data based on the differential interferometric phase sequence may include the following operations:

[0185] Linear deformation phase estimation is performed on the differential interferometric phase sequence to obtain the linear deformation phase estimation result. The elevation residual error is then removed from the linear deformation phase estimation result to obtain the linear deformation phase.

[0186] Nonlinear deformed phase estimation is performed on the differential interferometric phase sequence to obtain the nonlinear deformed phase estimation result. Atmospheric phase and noise are then removed from the nonlinear deformed phase estimation result to obtain the nonlinear deformed phase.

[0187] Based on the linear deformation phase and the nonlinear deformation phase, the deformation phase sequence corresponding to each target PS point is determined; wherein, the deformation phase sequence is based on the time series.

[0188] As can be seen, this optional embodiment can also determine the deformation phase sequence corresponding to each PS point by performing linear and nonlinear deformation phase estimation on the differential interference phase sequence and removing the error and noise data in the above estimation results, thereby improving the accuracy of the deformation phase sequence determination.

[0189] In an optional embodiment, calculating the deformation information corresponding to each target PS point based on the deformation phase sequence can include the following operations:

[0190] For each target PS point, the vertical deformation variable with geocoding is determined based on DEM data and deformation phase sequence to obtain the initial deformation variable; the initial deformation variable with geocoding is corrected based on high-precision measured data (such as leveling and / or GPS data) to serve as the deformation variable.

[0191] Based on the predetermined deformation model equations, the deformation rate corresponding to each target PS point is calculated.

[0192] As can be seen, this optional embodiment can also combine DEM data and high-precision measured data to determine the deformation variables with geocoding, and calculate the deformation rate based on the deformation model equation, which can improve the accuracy of determining the deformation variables and deformation rates corresponding to PS points.

[0193] In this optional embodiment, the calculation of the deformation rate corresponding to each target PS point based on a predetermined deformation model equation may include the following operations:

[0194] In the calculation of deformation rate, the deformation evolution history of any target PS point can be modeled as a linear motion model, as shown in the following equation:

[0195] (1)

[0196] In the formula, b represents the initial deformation value of the target PS point, v represents the deformation rate, t represents the time relative to the initial moment, and ε represents the unmodeled deformation and deformation measurement error. Combining N deformation data points throughout the entire monitoring period, the deformation model equation for any target PS point at each moment can be expressed in matrix form:

[0197] (2)

[0198] in:

[0199] (3)

[0200] In the above formula, D represents the observation vector; X represents the design matrix; β represents the parameter to be estimated; and E represents the residual term. [d1; d2; ...; dN] represent the LOS-directed deformation observations; [1, t1; 1, t2; ...; 1, tN] represent the design matrix (constant term and time variable); [b, v] represents the parameter vector (intercept + deformation rate); and [ε1; ε2; ..., εN] represents the residual (including atmospheric noise, unwrapping error, etc.).

[0201] The deformation rate v of the target PS point can be obtained based on the least squares theory.

[0202] As can be seen, this optional embodiment can more accurately determine the deformation rate of the PS point based on the more accurate deformation model equation and least squares theory, thereby improving the accuracy of the deformation information of the PS point.

[0203] In an optional embodiment, based on the deformation information corresponding to each target PS point, the target deformation region corresponding to the target region is identified to obtain the deformation region identification result, which may include the following operations:

[0204] Select target PS points whose deformation rate is greater than or equal to a preset deformation rate threshold from all target PS points and use them as the first candidate PS points;

[0205] Based on spatial location conditions, a clustering operation is performed on all first candidate PS points to obtain clustering results; the clustering results include multiple candidate deformation regions.

[0206] Based on the deformation rate, clustering results, and obtained deformation rate noise level of each target PS point, the fast deformation region corresponding to the target region is determined.

[0207] As can be seen, this optional embodiment can identify target PS points with deformation rates greater than or equal to a preset deformation rate threshold as first candidate PS points, perform clustering operations on all first candidate PS points to obtain clustering results, and then determine the fast deformation region corresponding to the target region based on the deformation rate corresponding to each target PS point, the clustering results, and the obtained deformation rate noise level. This can improve the accuracy of the analysis of the deformation rate of PS points, thereby improving the accuracy of PS point screening and clustering, and further improving the accuracy of identifying fast deformation regions.

[0208] In this optional embodiment, determining the fast deformation region corresponding to the target region based on the deformation rate corresponding to each target PS point, the clustering result, and the obtained deformation rate noise level may include the following operations:

[0209] For each candidate deformation region, the percentage of the first candidate PS point in that candidate deformation region is calculated.

[0210] Based on the obtained deformation rate noise level, the deformation rate corresponding to each target PS point is corrected to obtain the target deformation rate corresponding to each target PS point;

[0211] For each candidate deformation region, target PS points with a target deformation rate greater than or equal to a preset deformation rate threshold are selected from all target PS points corresponding to the candidate deformation region and used as second candidate PS points. The proportion of the second candidate PS points corresponding to the candidate deformation region in the candidate deformation region is then calculated.

[0212] For each candidate deformation region, when the first quantity proportion is greater than or equal to the first preset quantity proportion and the second quantity proportion is greater than or equal to the second preset quantity proportion, the candidate deformation region is determined as one of the fast deformation regions corresponding to the target region.

[0213] For example, the identification process for rapidly deformable regions can be found in [reference needed]. Figure 4 , Figure 4 This is a flowchart illustrating another method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, as disclosed in this embodiment of the invention; Figure 4 As shown, in this embodiment of the invention, the process for identifying rapidly deformable regions may include the following operations:

[0214] Compare the deformation rate v corresponding to each target PS point (e.g., the deformation rate can be obtained from the InSAR deformation database or determined in the manner described above) with the preset deformation rate threshold v. Th And from all target PS points, filter those with deformation rates v exceeding a preset deformation rate threshold v. Th The PS points (i.e., the first candidate PS points mentioned above) are used to cluster all the first candidate PS points, resulting in clustering results (including multiple classes, each class being a candidate deformation region). Then, the rate exceeding v in each class is counted. Th The percentage of the first number of PS points, and the rate exceeding v in each class after noise correction by combining deformation rate noise level. Th The second proportion of PS points is used to identify the rapid deformation zone in the target area by combining the first and second proportions mentioned above.

[0215] As can be seen, this optional embodiment can also, for each candidate deformation region, count the first number proportion of the first candidate PS points corresponding to the candidate deformation region in the candidate deformation region, and count the second number proportion of the second candidate PS points in the candidate deformation region whose corrected target deformation rate is greater than or equal to a preset deformation rate threshold in the candidate deformation region. When the first number proportion is greater than or equal to the first preset number proportion and the second number proportion is greater than or equal to the second preset number proportion, the candidate deformation region is determined as one of the fast deformation regions corresponding to the target region. This can improve the statistical accuracy of PS points for deformation rates before and after correction, thereby helping to further improve the identification accuracy of fast deformation regions.

[0216] In another optional embodiment, determining the landslide hazard identification result corresponding to the target area based on the landslide hazard location identification result and the deformation area identification result may include the following operations:

[0217] Obtain the environmental impact factors corresponding to the target area; environmental impact factors include one or more combinations of rainfall information, geological information, and topographic information; geological information includes lithology and / or structure; topographic information includes one or more combinations of slope, slope height, and slope aspect.

[0218] Based on environmental impact factors, landslide hazard location identification results, and deformation zone identification results, target hazard areas corresponding to target areas are determined, and landslide hazard prediction results corresponding to each target hazard area are analyzed; the landslide hazard prediction results include the probability of landslide occurrence.

[0219] Based on the landslide hazard prediction results, the hazard impact range corresponding to each target hazard area is determined;

[0220] The landslide hazard information corresponding to each target hazard area includes the landslide hazard prediction results and the hazard impact range.

[0221] Optionally, based on environmental impact factors, landslide hazard location identification results, and deformation area identification results, the target hazard area corresponding to the target area is determined, and the landslide hazard prediction results corresponding to each target hazard area are analyzed. This may include the following operations:

[0222] Based on the results of landslide hazard location identification and deformation area identification, the detection results are obtained by detecting whether any landslide hazard point or any target deformation area in the target area has geological disaster development characteristics.

[0223] The environmental impact factors are analyzed to determine the degree of environmental impact on any potential landslide site and any target deformation area in the target region, and the environmental impact analysis results are obtained.

[0224] Based on the detection results and environmental impact analysis results, the probability of landslide occurrence for each area in the target area is determined, and areas in the target area with a landslide occurrence probability higher than the preset probability are identified as target potential hazard areas.

[0225] Optionally, the landslide hazard prediction result may also include the predicted geological hazard type that could trigger a landslide; however, this embodiment of the invention does not impose any limitations on this.

[0226] Optionally, the hazard range corresponding to each target hazard area can be drawn by polygons or represented by other means, and the embodiments of the present invention are not limited thereto.

[0227] As can be seen, this optional embodiment can combine the acquired environmental impact factors, landslide hazard location identification results, and deformation area identification results to determine the target hazard area corresponding to the target area and analyze the landslide hazard prediction results corresponding to each target hazard area. Based on the landslide hazard prediction results, it can determine the hazard impact range corresponding to each target hazard area. It can further analyze the landslide hazard area and landslide hazard situation in the target area by combining the disaster-prone environmental factors, improve the accuracy of determining the target hazard area that may suffer from landslide disasters, and improve the accuracy of analyzing the specific hazard situation of the target hazard area. This is conducive to improving the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and thus facilitates timely disaster emergency measures.

[0228] Example 3

[0229] Please see Figure 5 , Figure 5 This is a schematic diagram of a landslide hazard identification device based on remote sensing imagery for slope-cutting construction, as disclosed in an embodiment of the present invention. Figure 5 The described landslide hazard identification device based on remote sensing imagery for slope-cutting construction may include one of the following: a hazard identification terminal, a hazard identification device, a hazard identification system, and a server. The server may be a local server or a cloud server; this embodiment of the invention is not limited thereto. Figure 5 As shown, the landslide hazard identification device based on remote sensing imagery for building houses on slopes may include:

[0230] The acquisition module 301 is used to acquire remote sensing image data and SAR image data corresponding to the target area;

[0231] Analysis module 302 is used to analyze remote sensing image data to obtain the landslide hazard location identification results corresponding to the target area; the landslide hazard location identification results include the location information of several landslide hazard points related to slope cutting for building construction;

[0232] Analysis module 302 is also used to analyze SAR image data based on PS-InSAR technology to obtain the deformation region identification results corresponding to the target area; the deformation region identification results include several target deformation regions;

[0233] The determination module 303 is used to determine the landslide hazard identification result corresponding to the target area based on the landslide hazard location identification result and the deformation area identification result; the landslide hazard identification result includes landslide hazard information corresponding to several target hazard areas within the target area.

[0234] As can be seen, the apparatus described in the embodiments of the present invention can identify several landslide hazard points related to slope-cutting construction within the target area by analyzing remote sensing image data corresponding to the target area, and obtain several target deformation areas within the target area by analyzing SAR image data corresponding to the target area based on PS-InSAR technology. Then, by combining the landslide hazard location identification results and deformation area identification results, the landslide hazard identification result corresponding to the target area is determined. By combining remote sensing image data and SAR image data to analyze landslide hazards, the comprehensiveness and reliability of the analysis of the target area can be improved, thereby improving the accuracy of determining the landslide hazard areas within the target area. This, in turn, improves the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and facilitates timely disaster emergency measures to effectively protect the lives and property of residents.

[0235] In an optional embodiment, the specific method by which the analysis module 302 analyzes remote sensing image data to obtain the landslide hazard location identification result corresponding to the target area may include:

[0236] Preprocessing of remote sensing image data yields preprocessed remote sensing image data.

[0237] Based on a pre-defined intelligent patch extraction algorithm, intelligent patch extraction is performed on pre-processed remote sensing image data to obtain patch extraction results.

[0238] Based on the trained hazardous building identification model, hazardous building identification is performed on preprocessed remote sensing image data to obtain hazardous building identification results; the hazardous building identification model is a U-Net model improved by the transfer learning method.

[0239] Based on the results of the extracted landforms and the identification of hazardous buildings, several landslide hazard points within the target area and the location information corresponding to each landslide hazard point are determined.

[0240] As can be seen, the apparatus described in this optional embodiment can perform intelligent patch extraction on the preprocessed remote sensing image data after preprocessing it, obtaining patch extraction results. Based on a trained hazard building identification model, it can identify hazard buildings in the preprocessed remote sensing image data, obtaining hazard building identification results. Combining the patch extraction results and the hazard building identification results, it can determine several landslide hazard points in the target area and the location information corresponding to each landslide hazard point. This can improve the comprehensiveness of remote sensing image data processing and analysis. Furthermore, by combining the patch extraction results and the hazard building identification results, it can improve the accuracy and reliability of determining landslide hazard points and related location information related to slope-cutting construction, thereby improving the efficiency and accuracy of subsequent landslide hazard identification for slope-cutting construction.

[0241] In this optional embodiment, optionally, the analysis module 302 performs intelligent patch extraction on the preprocessed remote sensing image data based on a pre-set intelligent patch extraction algorithm to obtain the patch extraction results. The specific method may include:

[0242] A segmentation operation is performed on the preprocessed remote sensing image data to obtain the patch segmentation results corresponding to the preprocessed remote sensing image data; the patch segmentation results include multiple patches;

[0243] Based on the Gabor filter, a texture feature extraction operation is performed on the patch segmentation results to obtain the texture feature extraction results; the texture feature extraction results include the texture feature information corresponding to each patch.

[0244] For each patch, based on the determined normal texture feature distribution, calculate the deviation of the texture feature information corresponding to the patch from the normal texture feature distribution;

[0245] Filter out abnormal patches from all patches whose deviation values ​​are greater than or equal to a preset deviation threshold, and use these as the abnormal patch filtering results.

[0246] The results of the image patch extraction include the results of the abnormal image patch screening.

[0247] As can be seen, the apparatus described in this optional embodiment can also obtain the patch segmentation results corresponding to the remote sensing image data by segmenting preprocessed remote sensing image data, and then extract the texture feature information corresponding to the patch segmentation results based on Gabor filter to calculate the deviation value of the texture feature information corresponding to each patch with respect to the normal texture feature distribution. Then, abnormal patches with deviation values ​​greater than or equal to a preset deviation threshold are selected from all patches as abnormal patch selection results. This can improve the accuracy of texture feature extraction and analysis of remote sensing image data, thereby improving the accuracy of abnormal patch selection. In turn, it is beneficial to more efficiently and accurately determine landslide hazard points in the target area based on accurate patch extraction results.

[0248] In another alternative embodiment, the SAR image data includes multiple SAR images;

[0249] The analysis module 302, based on PS-InSAR technology, analyzes SAR image data to obtain the deformation region identification result corresponding to the target area. The specific method for this can include:

[0250] Calculate the spatiotemporal baseline of each SAR image in the SAR image data, and determine the main image from the SAR image data based on the spatiotemporal baseline; the spatiotemporal baseline includes the temporal baseline and the spatial baseline;

[0251] Acquire DEM data that matches the SAR image data, and perform registration operations on other SAR images and DEM data based on the main image to obtain the auxiliary image and target DEM data corresponding to the SAR image data.

[0252] Based on the main image, all auxiliary images, and target DEM data, generate differential interferometric phase sequences corresponding to the SAR image data;

[0253] Based on the differential interferometric phase sequence, estimate the deformation phase sequence corresponding to each target PS point in the SAR image data;

[0254] Based on the deformation phase sequence corresponding to each target PS point, the deformation information corresponding to that target PS point is calculated; the deformation information includes the deformation amount and the deformation rate.

[0255] Based on the deformation information corresponding to each target PS point, the target deformation region corresponding to the target area is identified, and the deformation region identification result is obtained; among them, each target deformation region in the deformation region identification result is one of the deformation aggregation region, rapid deformation region and accelerated deformation region.

[0256] As can be seen, the apparatus described in this optional embodiment can determine the main image based on the spatiotemporal baseline of each SAR image in the SAR image data, and register other SAR images and acquired DEM data based on the main image to obtain auxiliary images and target DEM data. Then, it generates a differential interferometric phase sequence corresponding to the SAR image data to estimate the deformation phase sequence corresponding to each target PS point. Based on the deformation phase sequence, it calculates deformation information, thereby identifying the target deformation region corresponding to the target area based on the deformation information, and obtaining the deformation region identification result. This can improve the accuracy and comprehensiveness of SAR image data processing, thereby improving the accuracy of the analysis of PS point deformation, which in turn helps to improve the accuracy of identifying the deformation region contained in the target area. This further helps to improve the efficiency and accuracy of identifying landslide hazards in slope-cutting construction, and thus facilitates the timely implementation of more targeted disaster emergency measures.

[0257] In this optional embodiment, optionally, the specific method by which the analysis module 302 identifies the target deformation region corresponding to the target region based on the deformation information corresponding to each target PS point, and obtains the deformation region identification result, may include:

[0258] Select target PS points whose deformation rate is greater than or equal to a preset deformation rate threshold from all target PS points and use them as the first candidate PS points;

[0259] Based on spatial location conditions, a clustering operation is performed on all first candidate PS points to obtain clustering results; the clustering results include multiple candidate deformation regions.

[0260] Based on the deformation rate, clustering results, and obtained deformation rate noise level of each target PS point, the fast deformation region corresponding to the target region is determined.

[0261] As can be seen, the apparatus described in this optional embodiment can also identify target PS points with deformation rates greater than or equal to a preset deformation rate threshold as first candidate PS points, perform clustering operations on all first candidate PS points to obtain clustering results, and then determine the fast deformation region corresponding to the target region based on the deformation rate corresponding to each target PS point, the clustering results, and the obtained deformation rate noise level. This can improve the accuracy of the analysis of the deformation rate of PS points, thereby improving the accuracy of PS point screening and clustering, and further improving the accuracy of identifying fast deformation regions.

[0262] In this optional embodiment, the analysis module 302 may determine the specific method of the fast deformation region corresponding to the target region based on the deformation rate, clustering results, and obtained deformation rate noise level corresponding to each target PS point, which may include:

[0263] For each candidate deformation region, the percentage of the first candidate PS point in that candidate deformation region is calculated.

[0264] Based on the obtained deformation rate noise level, the deformation rate corresponding to each target PS point is corrected to obtain the target deformation rate corresponding to each target PS point;

[0265] For each candidate deformation region, target PS points with a target deformation rate greater than or equal to a preset deformation rate threshold are selected from all target PS points corresponding to the candidate deformation region and used as second candidate PS points. The proportion of the second candidate PS points corresponding to the candidate deformation region in the candidate deformation region is then calculated.

[0266] For each candidate deformation region, when the first quantity proportion is greater than or equal to the first preset quantity proportion and the second quantity proportion is greater than or equal to the second preset quantity proportion, the candidate deformation region is determined as one of the fast deformation regions corresponding to the target region.

[0267] As can be seen, the apparatus described in this optional embodiment can also, for each candidate deformation region, count the first proportion of the first candidate PS points corresponding to the candidate deformation region in the candidate deformation region, and count the second proportion of the second candidate PS points in the candidate deformation region whose corrected target deformation rate is greater than or equal to a preset deformation rate threshold in the candidate deformation region. When the first proportion is greater than or equal to the first preset proportion and the second proportion is greater than or equal to the second preset proportion, the candidate deformation region is determined as one of the fast deformation regions corresponding to the target region. This can improve the statistical accuracy of PS points for deformation rates before and after correction, thereby helping to further improve the identification accuracy of fast deformation regions.

[0268] In another optional embodiment, the specific method by which the determining module 303 determines the landslide hazard identification result corresponding to the target area based on the landslide hazard location identification result and the deformation area identification result may include:

[0269] Obtain the environmental impact factors corresponding to the target area; environmental impact factors include one or more combinations of rainfall information, geological information, and topographic information; geological information includes lithology and / or structure; topographic information includes one or more combinations of slope, slope height, and slope aspect.

[0270] Based on environmental impact factors, landslide hazard location identification results, and deformation zone identification results, target hazard areas corresponding to target areas are determined, and landslide hazard prediction results corresponding to each target hazard area are analyzed; the landslide hazard prediction results include the probability of landslide occurrence.

[0271] Based on the landslide hazard prediction results, the hazard impact range corresponding to each target hazard area is determined;

[0272] The landslide hazard information corresponding to each target hazard area includes the landslide hazard prediction results and the hazard impact range.

[0273] As can be seen, the device described in this optional embodiment can combine the acquired environmental influencing factors, landslide hazard location identification results, and deformation area identification results to determine the target hazard area corresponding to the target area and analyze the landslide hazard prediction results corresponding to each target hazard area. Based on the landslide hazard prediction results, it can determine the hazard impact range corresponding to each target hazard area. It can further analyze the landslide hazard area and landslide hazard situation in the target area in combination with disaster-prone environmental factors, improve the accuracy of determining the target hazard area where a landslide disaster may occur, and improve the accuracy of analyzing the specific hazard situation of the target hazard area. This is conducive to improving the efficiency and accuracy of landslide hazard identification for slope-cutting construction, and thus facilitates timely disaster emergency measures.

[0274] Example 4

[0275] Please see Figure 6 , Figure 6 This is a schematic diagram of another landslide hazard identification device based on remote sensing imagery for slope-cutting construction disclosed in an embodiment of the present invention. Figure 6 As shown, the landslide hazard identification device based on remote sensing imagery for building houses on slopes may include:

[0276] Memory 401 storing executable program code;

[0277] Processor 402 coupled to memory 401;

[0278] The processor 402 calls the executable program code stored in the memory 401 to execute some or all of the steps in the method for identifying landslide hazards in slope-cutting house construction based on remote sensing images as described in Embodiment 1 or Embodiment 2 of the present invention.

[0279] Example 5

[0280] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in the method for identifying landslide hazards in slope-cutting house construction based on remote sensing images, as described in Embodiment 1 or Embodiment 2 of this invention.

[0281] Example 6

[0282] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform some or all of the steps in the method for identifying landslide hazards in slope-cutting construction based on remote sensing images described in Embodiment 1 or Embodiment 2.

[0283] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0284] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0285] Finally, it should be noted that the method and device for identifying landslide hazards in slope-cutting construction based on remote sensing images disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, characterized in that, The method includes: Acquire remote sensing image data and SAR image data corresponding to the target area; Analyzing the remote sensing image data, the landslide hazard location identification results corresponding to the target area are obtained; the landslide hazard location identification results include the location information of several landslide hazard points related to slope cutting for building construction; Based on PS-InSAR technology, the SAR image data is analyzed to obtain the deformation region identification result corresponding to the target area; the deformation region identification result includes several target deformation regions; Based on the landslide hazard location identification results and the deformation area identification results, the landslide hazard identification results corresponding to the target area are determined; the landslide hazard identification results include landslide hazard information corresponding to several target hazard areas located within the target area; The analysis of the remote sensing image data to obtain the landslide hazard location identification result corresponding to the target area includes: The remote sensing image data is preprocessed to obtain preprocessed remote sensing image data; Based on a pre-defined intelligent patch extraction algorithm, the preprocessed remote sensing image data is subjected to intelligent patch extraction to obtain patch extraction results. Based on the trained hazardous building identification model, the hazardous building identification operation is performed on the preprocessed remote sensing image data to obtain the hazardous building identification result; wherein, the hazardous building identification model is a U-Net model improved by the transfer learning method; Based on the results of the extraction of the land parcels and the results of the identification of the hazardous buildings, several landslide hazard points within the target area and the location information corresponding to each landslide hazard point are determined. The method of performing intelligent patch extraction on the preprocessed remote sensing image data based on a pre-defined patch extraction algorithm to obtain patch extraction results includes: A segmentation operation is performed on the preprocessed remote sensing image data to obtain the patch segmentation results corresponding to the preprocessed remote sensing image data; the patch segmentation results include multiple patches; Based on a Gabor filter, a texture feature extraction operation is performed on the patch segmentation result to obtain a texture feature extraction result; the texture feature extraction result includes texture feature information corresponding to each patch. For each of the aforementioned patches, based on the determined normal texture feature distribution, the deviation value of the texture feature information corresponding to the patch with respect to the normal texture feature distribution is calculated; From all the said patches, abnormal patches whose deviation values ​​are greater than or equal to a preset deviation threshold are selected as the abnormal patch selection results; The image patch extraction results include the abnormal image patch screening results.

2. The method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery according to claim 1, characterized in that, The SAR image data includes multiple SAR images; The step of analyzing the SAR image data based on PS-InSAR technology to obtain the deformation region identification result corresponding to the target region includes: Calculate the spatiotemporal baseline of each SAR image in the SAR image data, and determine the main image from the SAR image data based on the spatiotemporal baseline; the spatiotemporal baseline includes a temporal baseline and a spatial baseline; Obtain DEM data that matches the SAR image data, and perform registration operations on other SAR images and the DEM data based on the main image to obtain the auxiliary image and target DEM data corresponding to the SAR image data. Based on the main image, all the auxiliary images, and the target DEM data, a differential interferometric phase sequence corresponding to the SAR image data is generated; Based on the differential interferometric phase sequence, estimate the deformation phase sequence corresponding to each target PS point in the SAR image data; Based on the deformation phase sequence corresponding to each target PS point, the deformation information corresponding to the target PS point is calculated; the deformation information includes deformation amount and deformation rate. Based on the deformation information corresponding to each target PS point, the target deformation region corresponding to the target region is identified to obtain the deformation region identification result; wherein, each target deformation region in the deformation region identification result is one of the deformation aggregation region, rapid deformation region and accelerated deformation region.

3. The method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery according to claim 2, characterized in that, The step of identifying the target deformation region corresponding to the target region based on the deformation information corresponding to each target PS point, and obtaining the deformation region identification result, includes: From all the target PS points, select the target PS points whose deformation rate is greater than or equal to the preset deformation rate threshold, and use them as the first candidate PS points; Based on spatial location conditions, a clustering operation is performed on all the first candidate PS points to obtain clustering results; the clustering results include multiple candidate deformation regions; Based on the deformation rate corresponding to each target PS point, the clustering result, and the obtained deformation rate noise level, the fast deformation region corresponding to the target region is determined.

4. The method for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery according to claim 3, characterized in that, The step of determining the fast deformation region corresponding to the target region based on the deformation rate corresponding to each target PS point, the clustering result, and the obtained deformation rate noise level includes: For each candidate deformation region, the proportion of the first candidate PS point corresponding to the candidate deformation region in the candidate deformation region is calculated. Based on the obtained deformation rate noise level, the deformation rate corresponding to each target PS point is corrected to obtain the target deformation rate corresponding to each target PS point; For each candidate deformation region, target PS points whose target deformation rate is greater than or equal to the preset deformation rate threshold are selected from all the target PS points corresponding to the candidate deformation region and used as second candidate PS points. The second number proportion of the second candidate PS points corresponding to the candidate deformation region in the candidate deformation region is then calculated. For each candidate deformation region, when the first quantity percentage is greater than or equal to the first preset quantity percentage and the second quantity percentage is greater than or equal to the second preset quantity percentage, the candidate deformation region is determined as one of the fast deformation regions corresponding to the target region.

5. The method for identifying landslide hazards based on slope-cutting construction using remote sensing imagery according to any one of claims 1-4, characterized in that, The step of determining the landslide hazard identification result corresponding to the target area based on the landslide hazard location identification result and the deformation area identification result includes: The environmental impact factors corresponding to the target area are obtained; the environmental impact factors include one or more combinations of rainfall information, geological information, and topographic information; the geological information includes lithology and / or structure; the topographic information includes one or more combinations of slope, slope height, and slope aspect. Based on the environmental influencing factors, the landslide hazard location identification results, and the deformation area identification results, the target hazard areas corresponding to the target areas are determined, and the landslide hazard prediction results corresponding to each target hazard area are analyzed; the landslide hazard prediction results include the probability of landslide occurrence. Based on the landslide hazard prediction results, the hazard impact range corresponding to each target hazard area is determined; The landslide hazard information corresponding to each target hazard area includes the landslide hazard prediction results and the hazard impact range.

6. A device for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, characterized in that, The device includes: The acquisition module is used to acquire remote sensing image data corresponding to the target area and SAR image data corresponding to the target area; The analysis module is used to analyze the remote sensing image data to obtain the landslide hazard location identification results corresponding to the target area; the landslide hazard location identification results include the location information of several landslide hazard points related to slope cutting for building construction; The analysis module is also used to analyze the SAR image data based on PS-InSAR technology to obtain the deformation region identification result corresponding to the target area; the deformation region identification result includes several target deformation regions; The determination module is used to determine the landslide hazard identification result corresponding to the target area based on the landslide hazard location identification result and the deformation area identification result; the landslide hazard identification result includes landslide hazard information corresponding to several target hazard areas located within the target area; The specific method by which the analysis module analyzes the remote sensing image data to obtain the landslide hazard location identification result corresponding to the target area includes: The remote sensing image data is preprocessed to obtain preprocessed remote sensing image data; Based on a pre-defined intelligent patch extraction algorithm, the preprocessed remote sensing image data is subjected to intelligent patch extraction to obtain patch extraction results. Based on the trained hazardous building identification model, the hazardous building identification operation is performed on the preprocessed remote sensing image data to obtain the hazardous building identification result; wherein, the hazardous building identification model is a U-Net model improved by the transfer learning method; Based on the results of the extraction of the land parcels and the results of the identification of the hazardous buildings, several landslide hazard points within the target area and the location information corresponding to each landslide hazard point are determined. The analysis module performs intelligent patch extraction on the preprocessed remote sensing image data based on a pre-defined patch extraction algorithm, and the specific methods for obtaining the patch extraction results include: A segmentation operation is performed on the preprocessed remote sensing image data to obtain the patch segmentation results corresponding to the preprocessed remote sensing image data; the patch segmentation results include multiple patches; Based on a Gabor filter, a texture feature extraction operation is performed on the patch segmentation result to obtain a texture feature extraction result; the texture feature extraction result includes texture feature information corresponding to each patch. For each of the aforementioned patches, based on the determined normal texture feature distribution, the deviation value of the texture feature information corresponding to the patch with respect to the normal texture feature distribution is calculated; From all the said patches, abnormal patches whose deviation values ​​are greater than or equal to a preset deviation threshold are selected as the abnormal patch selection results; The image patch extraction results include the abnormal image patch screening results.

7. A device for identifying landslide hazards in slope-cutting housing construction based on remote sensing imagery, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for identifying landslide hazards based on slope cutting for building construction as described in any one of claims 1-5.

8. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the method for identifying landslide hazards based on slope cutting for house construction as described in any one of claims 1-5.

Images