CN116561536B - Landslide hidden danger identification method, terminal equipment and medium - Google Patents
Landslide hidden danger identification method, terminal equipment and medium Download PDFInfo
- Publication number
- CN116561536B CN116561536B CN202310844794.9A CN202310844794A CN116561536B CN 116561536 B CN116561536 B CN 116561536B CN 202310844794 A CN202310844794 A CN 202310844794A CN 116561536 B CN116561536 B CN 116561536B
- Authority
- CN
- China
- Prior art keywords
- landslide
- hidden danger
- prediction model
- factors
- input end
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000012549 training Methods 0.000 claims abstract description 92
- 231100001267 hazard identification Toxicity 0.000 claims abstract description 16
- 238000011176 pooling Methods 0.000 claims description 56
- 238000012545 processing Methods 0.000 claims description 56
- 238000004590 computer program Methods 0.000 claims description 22
- 238000004364 calculation method Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000008707 rearrangement Effects 0.000 claims description 12
- 230000000877 morphologic effect Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000012790 confirmation Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 238000009960 carding Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A50/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE in human health protection, e.g. against extreme weather
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Biophysics (AREA)
- Marketing (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Educational Administration (AREA)
- Primary Health Care (AREA)
- Pit Excavations, Shoring, Fill Or Stabilisation Of Slopes (AREA)
Abstract
The application is suitable for the technical field of geological disaster identification, and provides a landslide hazard identification method, terminal equipment and medium, wherein multichannel image data are obtained by acquiring remote sensing images of identified areas and carrying out channel superposition on the remote sensing images; extracting training samples from the multi-channel image data to obtain a training sample set; extracting landslide hidden danger characteristics to obtain a plurality of candidate characteristic factors, and determining the most relevant characteristic factors corresponding to a training sample set; constructing an initial landslide potential prediction model, and training the initial landslide potential prediction model by using a training sample set to obtain a final landslide potential prediction model; predicting multi-channel image data corresponding to the area to be identified by utilizing a final landslide potential hazard prediction model to obtain a ground surface point with landslide potential hazard; and identifying landslide hidden danger of the ground surface point based on the landslide hidden danger rule set. The method can improve the accuracy of identifying the hidden danger of landslide.
Description
Technical Field
The application belongs to the technical field of geological disaster identification, and particularly relates to a landslide hazard identification method, terminal equipment and medium.
Background
Landslide hazard identification refers to prediction of the position, range and occurrence probability of the landslide hazard under different spatial scales. Landslides are highly concealed, abrupt and destructive. Aiming at the type confirmation of landslide hidden danger, especially the early type confirmation of the landslide hidden danger in a large area, the method has great significance on the research work such as landslide disaster risk evaluation, early warning, mechanism exploration and the like, and is an effective way for realizing the active disaster prevention and treatment from the passive disaster avoidance and relief.
The existing landslide hazard identification method can be classified into two types, namely knowledge driving and data driving. The knowledge driving method aims at carrying out hidden danger category confirmation based on geological expert knowledge, and comprises the following steps of manual ground investigation: the hidden danger is identified through professional field investigation, and the accuracy is higher, but inefficiency, and is difficult to discern high-order, hidden danger, leads to discernment insufficiency. Visual interpretation: the slope is visually judged on the remote sensing image through the knowledge and experience of the professional, so that the accuracy is high, the experience of the professional is very depended, and the time and the labor are wasted. Characteristic threshold: and setting one or more segmentation thresholds according to the manual experience and the data characteristics to identify landslide. The application range is smaller because the recognition standard is a threshold value of the specific characteristics of the specific region.
The data driving method aims at analyzing the association relation between data through different data processing algorithms based on the data, so as to generate an empirical model to realize landslide hidden danger identification, and comprises machine learning: machine learning uses computer algorithms to analyze and predict information by learning from training data. Specifically, the landslide hidden danger identification is realized by learning hidden danger identification samples, and the machine learning method has the advantages of high efficiency and high precision, but the feature selection and super-parameter debugging workload is larger. Deep learning: the deep learning essence is a deeper machine learning method, and by combining low-level features to form more abstract high-level features, the features can be automatically extracted and selected so as to realize hidden danger type confirmation, and the method is suitable for hidden danger confirmation of a large scene, but is limited by the number of landslide hidden danger samples, so that the problem of model overfitting is easily caused. In general, knowledge driving methods are based on expert domain knowledge, which is highly interpretable but not highly automated. The data driving method takes data as driving, has the characteristics of high efficiency, high precision and the like, is limited by the number and the quality of samples, and has the problems of over fitting, poor interpretation, low reliability and the like.
In summary, on one hand, the method can realize large-scale and full-period landslide hidden danger identification by utilizing various monitoring means such as optical remote sensing, interferometric radar (InSAR), geophysical prospecting and the like, but the method is easy to miss judgment and misjudgment. Wherein, the missed judgment means that part of hidden dangers are not identified, and the misjudgment means that other non-hidden dangers such as farmland, bare land and the like are identified as hidden dangers. On the other hand, the knowledge-based driving method and the data-based driving method are not well integrated, so that the interpretation and the accuracy of the existing landslide hazard identification are difficult to balance. Therefore, a method capable of accurately identifying landslide hazards is needed.
Disclosure of Invention
The application provides a landslide hazard identification method, terminal equipment and medium, which can solve the problem of low identification accuracy of the existing landslide hazard.
In a first aspect, the present application provides a method for identifying a landslide hazard, including:
acquiring a remote sensing image of the identified area, and carrying out channel superposition on the remote sensing image to obtain multi-channel image data of the identified area;
extracting training samples from the multi-channel image data to obtain a training sample set; the training sample represents a sequence formed by arranging image attribute values of different channels, and the training sample set comprises a plurality of landslide hidden danger samples and a plurality of non-landslide hidden danger samples;
Extracting landslide hidden danger characteristics of a training sample set to obtain a plurality of candidate characteristic factors, and determining the most relevant characteristic factors corresponding to the training sample set from the plurality of candidate characteristic factors; the characteristic factors represent influence factors for generating landslide hidden danger, and the most relevant characteristic factors represent candidate characteristic factors with the greatest importance degree;
constructing an initial landslide potential prediction model, and training the initial landslide potential prediction model by using a training sample set to obtain a final landslide potential prediction model; the landslide potential hazard prediction model comprises a first branch and a second branch, wherein the first branch is used for processing remote sensing images, and the second branch is used for processing most relevant influence factors;
and predicting the multichannel image data corresponding to the area to be identified by using the final landslide potential hazard prediction model to obtain the ground surface point position with the landslide potential hazard in the area to be identified.
Identifying landslide hidden danger of the ground surface point location based on a pre-constructed landslide hidden danger rule set; landslide hazards are in the deformation area, the historical deformation damage area or the potential unstable slope.
Optionally, the landslide hazard features include morphological features, deformation features, situational features, and other features; the situation features comprise disaster factors of landslide and disaster ranges of the landslide.
Optionally, determining the most relevant feature factor corresponding to the training sample set from the plurality of candidate feature factors includes:
obtaining a correlation degree coefficient among a plurality of candidate feature factorsThe method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Candidate feature factors->And->Candidate feature factors->Correlation coefficient between->，/>，Representing the total number of candidate feature factors, +.>Representing variance->Indicate->Candidate feature factors->And->Candidate feature factors->Covariance between;
by calculation formula
Obtaining the variance expansion coefficient of each candidate characteristic factorThe method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Candidate feature factors->Coefficient of expansion of variance,/">Indicating removal->Is->The certainty factor is obtained when the other candidate feature factors are subjected to linear regression, and the variance expansion coefficient is used for measuring the co-linearity degree inside each candidate feature factor;
and respectively eliminating candidate characteristic factors with the correlation degree coefficient larger than a preset correlation degree threshold value and candidate characteristic factors with the variance expansion coefficient larger than a preset variance expansion threshold value from the candidate characteristic factors to obtain a plurality of correlation characteristic factors.
Optionally, determining the most relevant feature factor corresponding to the training sample set from the plurality of candidate feature factors further includes:
By calculation formula
Obtaining a landslide hidden danger characteristic pair training sample setInformation gain of->The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Characteristic of landslide hazard->For training sample set->Information gain of->Is any one of morphological characteristics, deformation characteristics, situational characteristics and other characteristics, +.>Representing the characteristic of landslide hazard->Dividing training sample set->The number of sub-sets obtained>Indicate->The number of training samples in the subset, +.>，/>Representing training sample set +.>Total number of training samples, +.>Indicate->Information entropy of the individual subsets, +.>Indicate->Total number of categories of relevant feature factors in the respective subset,/->Indicate->The category of the relevant characteristic factors is +.>Probability of occurrence in the individual subsets;
and taking the landslide hidden danger feature corresponding to the maximum information gain as the most relevant landslide hidden danger feature, and taking the relevant feature factor corresponding to the most relevant landslide hidden danger feature as the most relevant feature factor.
Optionally, the first branch includes a first convolution layer, a second convolution layer, and a first pooling layer, and the second branch includes a third convolution layer, a fourth convolution layer, and a second pooling layer; the output end of the first convolution layer is respectively connected with the input end of the second convolution layer and the input end of the first pooling layer, the second convolution layer outputs a first grid matrix corresponding to the remote sensing image calculated by the convolution operator, the first pooling layer outputs a second grid matrix corresponding to the remote sensing image calculated by the maximum value in the receptive field, the output end of the third convolution layer is respectively connected with the input end of the fourth convolution layer and the input end of the second pooling layer, the fourth convolution layer outputs a third grid matrix corresponding to the most relevant influence factor calculated by the convolution operator, and the second pooling layer outputs a fourth grid matrix corresponding to the most relevant influence factor calculated by the maximum value in the receptive field.
Optionally, the landslide hidden danger prediction model further includes a first processing unit, a second processing unit, a channel rearrangement module, a third pooling layer, a fourth pooling layer, a fifth pooling layer, a third processing unit, a fifth convolution layer, a sixth convolution layer, a fourth processing unit and a CBAM module; the input end of the first processing unit is respectively connected with the output end of the second convolution layer and the output end of the first pooling layer in the branch I, the output end of the first processing unit is connected with the input end of the channel rearrangement module, the input end of the second processing unit is respectively connected with the output end of the fourth convolution layer and the output end of the second pooling layer in the branch II, the output end of the second processing unit is connected with the input end of the channel rearrangement module, the output end of the channel rearrangement module is respectively connected with the input end of the third pooling layer, the input end of the fifth pooling layer and the input end of the fifth convolution layer, the output end of the third pooling layer is connected with the input end of the fourth processing unit, and the input end of the fourth pooling layer is connected with the input end of the fourth processing unit; the output end of the fifth pooling layer is connected with the input end of the third processing unit, the output end of the fifth convolution layer is connected with the input end of the third processing unit, the output end of the third processing unit is connected with the input end of the fifth pooling layer, the output end of the sixth convolution layer is connected with the input end of the fourth processing unit, the input end of the CBAM module is connected with the output end of the fourth processing unit, and the CBAM module outputs a landslide hidden danger area prediction result.
Optionally, the landslide hazard prediction model further comprises a U-net network for identifying the disaster-bearing body.
Optionally, training the initial landslide hazard prediction model by using a training sample set to obtain a final landslide hazard prediction model, including:
step i, through a calculation formula
Obtaining an initial landslide hidden danger prediction modelPredictive loss value of (2)The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Representing weight parameters->Represents the total number of training samples, +.>Indicate->True tag of individual training samples, +.>，/>Output label for representing initial landslide hazard prediction model>Probability of->Output of prediction model for initial landslide hazard>Predictive value of the label of the individual training samples, +.>Output of prediction model for initial landslide hazard>Probability of predicted value of label of each training sample;
step ii, if the predicted loss value is smaller than or equal to a preset loss threshold value, taking the initial landslide potential hazard prediction model as a final landslide potential hazard prediction model; and (d) if not, carrying out back propagation on the initial landslide hazard prediction model by using the prediction loss value, and returning to the step (i).
Optionally, predicting the multichannel image data corresponding to the area to be identified by using a final landslide potential prediction model to obtain the ground surface point position with the landslide potential in the area to be identified, including:
Dividing the multichannel image data to be identified; wherein the multi-channel image data is divided into a plurality of sub-images;
inputting each sub-image in the plurality of sub-images into a final landslide potential prediction model to obtain a prediction value of the landslide potential existing in the sub-region corresponding to the sub-image; the sub-region represents a portion of the region to be identified;
and taking the earth surface points corresponding to all the subareas with the predicted values larger than the preset threshold value as the earth surface points with landslide hidden danger.
In a second aspect, the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, where the processor implements the method for identifying a landslide hazard described above when executing the computer program.
In a third aspect, the present application provides a computer readable storage medium, where a computer program is stored, and when the computer program is executed by a processor, the method for identifying a landslide hazard is implemented.
The scheme of the application has the following beneficial effects:
according to the landslide hazard identification method, the remote sensing images of the identified areas are acquired, the remote sensing images are subjected to channel superposition, and then the samples are extracted, so that high-precision training samples can be obtained, and the landslide hazard identification accuracy is improved; by extracting the landslide hidden danger characteristics of the training samples, the most relevant characteristic factors are determined, so that data redundancy can be reduced, interference of a plurality of characteristic factors on the identification of the landslide hidden danger is avoided, workload is reduced, and accuracy of the identification of the landslide hidden danger is improved; the constructed landslide potential hazard prediction model comprehensively considers the remote sensing image and the most relevant influence factors, and can obtain a more accurate landslide potential hazard area, so that the identification accuracy of the landslide potential hazard is improved; based on the landslide hidden danger rule set, the landslide hidden danger areas are classified, the landslide hidden danger of the area to be identified is identified, the advantages of knowledge driving and data driving are fully combined, the interpretability of the result is improved, and the identification accuracy of the landslide hidden danger is improved.
Other advantageous effects of the present application will be described in detail in the detailed description section which follows.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for identifying landslide hazard according to an embodiment of the present application;
FIG. 2a is a schematic structural diagram of branch one of a landslide hazard prediction model according to an embodiment of the present application;
FIG. 2b is a schematic structural diagram of a second branch of the landslide hazard prediction model according to an embodiment of the present application;
FIG. 3 is a schematic structural diagram of a landslide hazard prediction model according to another embodiment of the present application;
FIG. 4 is a schematic diagram of a CBAM module according to an embodiment of the present application;
FIG. 5 is a flowchart for constructing a rule set of landslide hazards according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of a terminal device according to an embodiment of the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
It should be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It should also be understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
As used in the present description and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".
Furthermore, the terms "first," "second," "third," and the like in the description of the present specification and in the appended claims, are used for distinguishing between descriptions and not necessarily for indicating or implying a relative importance.
Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.
Aiming at the problem of low accuracy of the existing landslide hazard identification method, the application provides the landslide hazard identification method, the terminal equipment and the medium, by acquiring the remote sensing image of the identified area, carrying out channel superposition on the remote sensing image and then carrying out sample extraction on the remote sensing image, a high-precision training sample can be obtained, thereby being beneficial to improving the accuracy of landslide hazard identification; by extracting the landslide hidden danger characteristics of the training samples, the most relevant characteristic factors are determined, so that data redundancy can be reduced, interference of a plurality of characteristic factors on the identification of the landslide hidden danger is avoided, workload is reduced, and accuracy of the identification of the landslide hidden danger is improved; the constructed landslide potential hazard prediction model comprehensively considers the remote sensing image and the most relevant influence factors, and can obtain a more accurate landslide potential hazard area, so that the identification accuracy of the landslide potential hazard is improved; based on the landslide hidden danger rule set, the landslide hidden danger areas are classified, the landslide hidden danger of the area to be identified is identified, the advantages of knowledge driving and data driving are fully combined, the interpretability of the result is improved, and the identification accuracy of the landslide hidden danger is improved.
As shown in fig. 1, the method for identifying landslide hazard provided by the application comprises the following steps:
and 11, acquiring a remote sensing image of the identified area, and carrying out channel superposition on the remote sensing image to obtain multi-channel image data of the identified area.
The identified areas represent areas that have been examined by an expert.
In an embodiment of the present application, the remote sensing image includes a high resolution optical remote sensing image, a Sentinel 1 radar (Sentinel-1 SAR) image, digital elevation data (DEM, digital Elevation Model), normalized vegetation index (NDVI, normalized Difference Vegetation Index), and earth surface coverage type data. In one embodiment of the present application, the resolution of the high resolution optical remote sensing image is set to 0.8m, and the other data is set to 30m.
It should be noted that, a common channel stacking method may be used to perform channel stacking on the remote sensing image, for example: color space conversion, gradient operators, anisotropic filtering, and the like, the specific manner of which is not limited herein.
And step 12, extracting training samples from the multi-channel image data to obtain a training sample set.
The training sample represents a sequence formed by arranging image attribute values of different channels, and the training sample set comprises a plurality of landslide hidden danger samples and a plurality of non-landslide hidden danger samples. In an embodiment of the present application, assuming an infrared channel and a visible light channel, a certain training sample is expressed as follows: training sample 1= { sequence of infrared channel pixel values (0.1,0.3,0.2), sequence of visible light channel pixel values (0.5,0.4,0.6), label: landslide hazard samples }.
It should be noted that, in practical application, the proportion of the landslide hidden danger sample is far smaller than that of the non-landslide hidden danger sample, which may cause unbalance of the training sample set, so in some embodiments of the present application, the training sample set needs to be balanced, specifically, the landslide hidden danger sample is taken as a positive sample, the non-landslide hidden danger sample is taken as a negative sample, and 1:1 to balance the training sample set, common balancing methods are downsampling the samples and oversampling the samples.
In the embodiment of the application, positive sample screening, negative sample optimization and sample expansion are also performed after the training sample set is obtained.
The positive sample screening procedure was as follows:
the original landslide disaster point data are screened to identify positive samples consistent with priori knowledge, and part of landslide hidden danger samples are inconsistent with actual positions due to the fact that landslide hidden danger recording positions are recorded incorrectly and even recorded in a flat land area, and according to the landslide hidden danger knowledge, landslide is generally inoculated at a slope of 10-45 degrees, which is contrary to the distribution rule of the landslide, so that corresponding data can be deleted according to the priori knowledge.
The process of negative sample optimization is as follows:
and generating a negative sample by taking the priori knowledge as constraint, wherein the negative sample is a region where the landslide hidden danger does not occur, and screening out the region which is not suitable for the landslide hidden danger according to the priori knowledge, so as to generate the negative sample in the landslide hidden danger region.
The procedure for sample expansion was as follows:
the simulation data is generated according to knowledge to expand samples, and the simulation data is generated according to knowledge and prior methods because of insufficient number of landslide hazard samples, and common simulation data generation methods comprise sample augmentation and generation of countermeasure networks (GAN, generative adversarial network). Firstly, positive and negative samples are obtained in a conventional mode, and the data volume is expanded through sample enhancement strategies such as overturning, rotating and the like, so that more analog data can be generated based on GAN, and sample expansion is realized.
And 13, extracting landslide hidden danger characteristics of the training sample set to obtain a plurality of candidate feature factors, and determining the most relevant feature factors corresponding to the training sample set from the plurality of candidate feature factors.
In an embodiment of the present application, the landslide hazard features include morphological features, deformation features, situational features, and other features, where situational features include disaster factors of the landslide and disaster ranges of the landslide. For example, morphological characteristics can be obtained by investigation on disaster bodies through high-resolution optical remote sensing images, so that geological background and earth surface coverage changes of disasters are researched and formed, and possible disaster forming conditions are revealed; the deformation characteristics aim at judging the sliding scale, the activity stage and the development trend of the disaster body according to the surface deformation state of the geologic body; the situation features are used for analyzing disaster causing situations such as disaster causing factors of hidden danger, threat range after disaster occurrence and the like; other features are the existence of intermediate achievements and other knowledge in the recognition process.
The characteristic factors represent influence factors for generating landslide hazards, and the most relevant characteristic factors represent candidate characteristic factors with the greatest importance degree. It should be noted that, different landslide hidden danger features correspond to different feature factors, and in the embodiment of the application, morphological features correspond to feature factors such as mean, maximum difference, standard deviation, brightness, etc.; the deformation characteristics correspond to characteristic factors such as deformation quantity, deformation rate and the like; the situational characteristics correspond to characteristic factors such as average annual rainfall, earthquake density, land utilization type, road density and the like; other characteristics correspond to characteristic factors such as susceptibility factors, disaster-bearing body distribution diagrams, disaster point distribution diagrams and the like.
And 14, constructing an initial landslide potential prediction model, and training the initial landslide potential prediction model by using a training sample set to obtain a final landslide potential prediction model.
The landslide potential prediction model comprises a first branch and a second branch, wherein the first branch is used for processing remote sensing images, and the second branch is used for processing most relevant influence factors. Specifically, as shown in fig. 2a and 2b, the first branch includes a first convolution layer a1, a second convolution layer a2, and a first pooling layer a3. The second branch comprises a third convolution layer b1, a fourth convolution layer b2 and a second pooling layer b3.
The output end of the first convolution layer a1 is connected with the input end of the second convolution layer a2 and the input end of the first pooling layer a3 respectively. The second convolution layer a2 outputs a first grid matrix corresponding to the remote sensing image calculated by the convolution operator, and the first pooling layer a3 outputs a second grid matrix corresponding to the remote sensing image calculated by the maximum value in the receptive field.
The output end of the third convolution layer b1 is connected to the input end of the fourth convolution layer b2 and the input end of the second pooling layer b3, respectively. The fourth convolution layer b2 outputs a third grid matrix corresponding to the most relevant influence factor calculated by the convolution operator, and the second pooling layer b3 outputs a fourth grid matrix corresponding to the most relevant influence factor calculated by the maximum value in the receptive field.
The receptive field represents a unit mapping range when a convolution layer or a pooling layer carries out convolution calculation, and maximum value calculation is a mode of pooling calculation, and represents that the maximum value in the receptive field is extracted as a mapping element value in the pooling process.
As shown in fig. 3, the landslide hazard prediction model further includes a first processing unit 311, a second processing unit 312, a channel rearrangement module 313, a third pooling layer 314, a fourth pooling layer 315, a fifth pooling layer 316, a third processing unit 317, a fifth convolution layer 318, a sixth convolution layer 319, a fourth processing unit 320, and a CBAM module 321. The input end of the first processing unit 311 is respectively connected with the output end of the second convolution layer and the output end of the first pooling layer in the branch I, and the output end of the first processing unit 311 is connected with the input end of the channel rearrangement module 313; the input end of the second processing unit 312 is respectively connected with the output end of the fourth convolution layer and the output end of the second pooling layer in the branch two, and the output end of the second processing unit is connected with the input end of the channel rearrangement module 313; the output end of the channel rearrangement module 313 is respectively connected to the input end of the third pooling layer 314, the input end of the fifth pooling layer 316 and the input end of the fifth convolution layer 318; the output end of the third pooling layer 314 is connected to the input end of the fourth processing unit 320; the input end of the fourth pooling layer 315 is connected to the output end of the third processing unit 317, and the output end of the fourth pooling layer 315 is connected to the input end of the fourth processing unit 320; the output end of the fifth pooling layer 316 is connected to the input end of the third processing unit 317; the output end of the fifth convolution layer 318 is connected to the input end of the third processing unit 317; the output end of the third processing unit 317 is connected to the input end of the fifth pooling layer 316; the output of the sixth convolution layer 319 is connected to the input of the fourth processing unit 320; the input end of the spatial attention and channel attention (CBAM, convolutional Block Attention Module) module 321 is connected to the output end of the fourth processing unit 320, and the CBAM module 321 outputs a landslide hazard area prediction result.
As shown in fig. 4, the CBAM module includes a channel attention module 41 for multiplying channel weights and a spatial attention module 42 for multiplying spatial weights.
In the embodiment of the application, the landslide hazard prediction model further comprises a U-net network for identifying the disaster-bearing body.
And 15, predicting the multichannel image data corresponding to the area to be identified by using the final landslide potential hazard prediction model to obtain the ground surface point position with the landslide potential hazard in the area to be identified.
And step 16, identifying landslide hidden danger of the ground surface point location in a pre-constructed landslide hidden danger rule set.
Landslide hazards are in the deformation area, the historical deformation damage area or the potential unstable slope.
As shown in fig. 5, the steps of constructing the landslide hazard rule set include:
step 51, mechanism cognition.
Specifically, the landslide hidden trouble is divided into an ongoing deformation area, a historical deformation damage area and a potential unstable slope. An area of deformation refers to an area or region that is currently undergoing deformation and has obvious signs and characteristics of deformation; the historic deformation damage areas mainly refer to slopes with obvious damage such as ancient landslide bodies, shattered mountain bodies, aging deformation bodies (such as large dumping deformation bodies common to valleys in western mountain areas), large loose piles with various causes and the like; the potential unstable slope is basically stable or understable under the natural working condition, deformation damage does not occur historically, no obvious deformation sign exists currently, but sudden instability damage and disaster forming slope can occur under strong disturbance such as heavy rainfall, earthquake or ergonomic activity.
And step 52, identifying hidden danger features.
Specifically, the landslide hidden danger features are divided into four types of morphological features, deformation features, situational features and other features, and the landslide hidden danger features of different types are summarized.
And step 53, establishing a factor system.
Step 54, knowledge description.
The description is based on the distribution characteristics of landslide hazards in different characteristic factors. If the deformation rate is high, landslide hidden danger is usually formed, and disaster-bearing bodies exist around the landslide hidden danger.
And step 55, constructing rules.
And (5) according to the landslide hazard knowledge obtained in the step 54, inducing relevant rules and formally expressing. Exemplary:
landslide hazard rule one: the hidden danger of landslide is mainly distributed between 10-45 degrees, the deformation rate is greater than 10 mm/year, and the direction is downward along the slope body, so that the potential is a deformation zone.
Landslide hazard rule II: disaster-bearing bodies such as roads, buildings, farmlands and the like exist around the landslide hidden danger. Landslide is likely to occur in areas closer to faults.
Landslide hazard rule three: landslide is mainly located in the middle-high easily-developed area.
Landslide hazard rule four: in areas with earthquake intensity greater than 7 degrees, slopes with gradients greater than 25 degrees are prone to landslide under an earthquake.
The following is an exemplary description of the process of determining, from among the plurality of candidate feature factors, the most relevant feature factor corresponding to the training sample set in step 13 (extracting the landslide hidden danger feature of the training sample set to obtain a plurality of candidate feature factors, and determining, from the plurality of candidate feature factors, the most relevant feature factor corresponding to the training sample set), where the steps include steps 13.1-13.5:
Step 13.1, by calculation formula
Obtaining a correlation degree coefficient among a plurality of candidate feature factors。
Wherein,indicate->Candidate feature factors->And->Candidate feature factors->Correlation coefficient between->，/>，/>Representing the total number of candidate feature factors, +.>The variance is represented as a function of the variance,indicate->Candidate feature factors->And->Candidate feature factors->Covariance between.
Step 13.2, through the calculation formula
Obtaining the variance expansion coefficient of each candidate characteristic factor。
Wherein,indicate->Candidate feature factors->Coefficient of expansion of variance,/">Indicating removal->A kind of electronic deviceAnd (3) obtaining a certainty coefficient when carrying out linear regression on other candidate characteristic factors, wherein the variance expansion coefficient is used for measuring the co-linearity degree inside each candidate characteristic factor.
And 13.3, respectively eliminating candidate feature factors with the correlation degree coefficient larger than a preset correlation degree threshold value and candidate feature factors with the variance expansion coefficient larger than a preset variance expansion threshold value from the candidate feature factors to obtain a plurality of correlation feature factors.
Step 13.4, through the calculation formula
Obtaining a landslide hidden danger characteristic pair training sample setInformation gain of->。
Wherein,characteristic of landslide hazard- >For training sample set->Information gain of->Is morphological, deformation and situational characteristicsAnd any of the other features ∈four, +.>Representing the characteristic of landslide hazard->Dividing training sample set->The number of sub-sets obtained>Indicate->The number of training samples in the individual sub-sets,，/>representing training sample set +.>Total number of training samples, +.>Indicate->Information entropy of the individual subsets, +.>Indicate->Total number of categories of relevant feature factors in the respective subset,/->Indicate->The category of the relevant characteristic factors is +.>Probability of occurrence in the individual subsets;
and 13.5, taking the landslide hidden danger feature corresponding to the maximum information gain as the most relevant landslide hidden danger feature, and taking the relevant feature factor corresponding to the most relevant landslide hidden danger feature as the most relevant feature factor.
The following describes an exemplary process of training the initial landslide hazard prediction model by using the training sample set in step 14 (constructing the initial landslide hazard prediction model and training the initial landslide hazard prediction model by using the training sample set to obtain the final landslide hazard prediction model), and the process of obtaining the final landslide hazard prediction model, including steps i-ii:
Step i, through a calculation formula
Obtaining a prediction loss value of an initial landslide potential prediction model。
Wherein,representing weight parameters->Represents the total number of training samples, +.>Indicate->True tag of individual training samples, +.>，/>Output label for representing initial landslide hazard prediction model>Is a function of the probability of (1),output of the initial landslide hazard prediction model is represented by +.>Predictive value of the label of the individual training samples, +.>Output of the initial landslide hazard prediction model is represented by +.>Probability of predicted value of label of each training sample;
step ii, if the predicted loss value is smaller than or equal to a preset loss threshold value, taking the initial landslide potential hazard prediction model as a final landslide potential hazard prediction model; and (d) if not, carrying out back propagation on the initial landslide hazard prediction model by using the prediction loss value, and returning to the step (i).
The following describes the procedure of step 15 (predicting the multi-channel image data corresponding to the area to be identified by using the final landslide hazard prediction model to obtain the ground surface point location with the landslide hazard in the area to be identified), which includes steps 15.1-15.3:
and 15.1, dividing the multi-channel image data to be identified.
The multi-channel image data is divided into a plurality of sub-images.
And 15.2, inputting each sub-image in the plurality of sub-images into a final landslide potential hazard prediction model to obtain a prediction value of the landslide potential hazard existing in the sub-region corresponding to the sub-image.
The sub-region represents a portion of the region to be identified.
And 15.3, taking the surface points corresponding to all the subareas with the predicted values larger than the preset threshold value as the surface points with landslide hidden danger.
For example, in an embodiment of the present application, the preset threshold may be set to 0.8, that is, the ground point corresponding to all sub-regions with predictive values greater than 0.8 is used as the ground point with landslide hazard
As shown in fig. 6, an embodiment of the present application provides a terminal device, and as shown in fig. 6, a terminal device D10 of the embodiment includes: at least one processor D100 (only one processor is shown in fig. 6), a memory D101 and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, the processor D100 implementing the steps in any of the various method embodiments described above when executing the computer program D102.
Specifically, when the processor D100 executes the computer program D102, the remote sensing image of the identified area is obtained, the remote sensing image is subjected to channel superposition to obtain multi-channel image data of the identified area, then a training sample is extracted from the multi-channel image data to obtain a training sample set, the landslide hidden danger characteristics of the training sample set are extracted to obtain a plurality of candidate characteristic factors, the most relevant characteristic factors corresponding to the training sample set are identified from the plurality of candidate characteristic factors, then an initial landslide hidden danger prediction model is constructed, the initial landslide hidden danger prediction model is trained by using the training sample set to obtain a final landslide hidden danger prediction model, the multi-channel image data corresponding to the area to be identified is predicted by using the final landslide hidden danger prediction model to obtain the ground surface point with the landslide hidden danger in the area to be identified, and finally the landslide hidden danger of the ground surface point is identified based on the pre-constructed landslide hidden danger rule set. The remote sensing image of the identified area is acquired, the remote sensing image is subjected to channel superposition, and then the sample is extracted, so that a high-precision training sample can be obtained, and the accuracy of landslide hidden danger identification is improved; by extracting the landslide hidden danger characteristics of the training samples, the most relevant characteristic factors are determined, so that data redundancy can be reduced, interference of a plurality of characteristic factors on the identification of the landslide hidden danger is avoided, workload is reduced, and accuracy of the identification of the landslide hidden danger is improved; the constructed landslide potential hazard prediction model comprehensively considers the remote sensing image and the most relevant influence factors, and can obtain a more accurate landslide potential hazard area, so that the identification accuracy of the landslide potential hazard is improved; based on the landslide hidden danger rule set, the landslide hidden danger areas are classified, the landslide hidden danger of the area to be identified is identified, the advantages of knowledge driving and data driving are fully combined, the interpretability of the result is improved, and the identification accuracy of the landslide hidden danger is improved.
The processor D100 may be a central processing unit (CPU, central Processing Unit), the processor D100 may also be other general purpose processors, digital signal processors (DSP, digital Signal Processor), application specific integrated circuits (ASIC, application Specific Integrated Circuit), off-the-shelf programmable gate arrays (FPGA, field-Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory D101 may in some embodiments be an internal storage unit of the terminal device D10, for example a hard disk or a memory of the terminal device D10. The memory D101 may also be an external storage device of the terminal device D10 in other embodiments, for example, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the terminal device D10. Further, the memory D101 may also include both an internal storage unit and an external storage device of the terminal device D10. The memory D101 is used for storing an operating system, an application program, a boot loader (BootLoader), data, other programs, etc., such as program codes of the computer program. The memory D101 may also be used to temporarily store data that has been output or is to be output.
Embodiments of the present application also provide a computer readable storage medium storing a computer program which, when executed by a processor, implements steps for implementing the various method embodiments described above.
Embodiments of the present application provide a computer program product enabling a terminal device to carry out the steps of the method embodiments described above when the computer program product is run on the terminal device.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiments, and may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to a terminal device, a recording medium, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Such as a U-disk, removable hard disk, magnetic or optical disk, etc. In some jurisdictions, computer readable media may not be electrical carrier signals and telecommunication signals, in accordance with legislation and patent practice.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus/network device and method may be implemented in other manners. For example, the apparatus/network device embodiments described above are merely illustrative, e.g., the division of the modules or units is merely a logical functional division, and there may be additional divisions in actual implementation, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The landslide hazard identification method provided by the application has the following advantages:
1, a knowledge-embedded landslide hazard full-period identification method is created.
2, carrying out comprehensive landslide hazard identification rule carding, and constructing a knowledge-embedded landslide hazard full-period identification method.
And 3, the interpretability and the accuracy of landslide hazard identification can be effectively improved, and the problem of misjudgment and omission of landslide hazard identification is solved.
While the foregoing is directed to the preferred embodiments of the present application, it will be appreciated by those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present application, and such modifications and adaptations are intended to be comprehended within the scope of the present application.
Claims (9)
1. The landslide hazard identification method is characterized by comprising the following steps of:
Acquiring a remote sensing image of an identified area, and carrying out channel superposition on the remote sensing image to obtain multi-channel image data of the identified area;
extracting training samples from the multi-channel image data to obtain a training sample set; the training sample set comprises a plurality of landslide hidden danger samples and a plurality of non-landslide hidden danger samples;
extracting landslide hidden danger characteristics of the training sample set to obtain a plurality of candidate feature factors, and determining the most relevant feature factors corresponding to the training sample set from the plurality of candidate feature factors; the characteristic factors represent influence factors for generating landslide hazards, and the most relevant characteristic factors represent candidate characteristic factors with the greatest importance degree;
constructing an initial landslide potential prediction model, and training the initial landslide potential prediction model by utilizing the training sample set to obtain a final landslide potential prediction model; the landslide hidden danger prediction model comprises a first branch and a second branch, wherein the first branch is used for processing the remote sensing image, the second branch is used for processing the most relevant characteristic factors, the first branch comprises a first convolution layer, a second convolution layer and a first pooling layer, and the second branch comprises a third convolution layer, a fourth convolution layer and a second pooling layer; the output end of the first convolution layer is respectively connected with the input end of the second convolution layer and the input end of the first pooling layer, the second convolution layer outputs a first grid matrix corresponding to the remote sensing image calculated by a convolution operator, the first pooling layer outputs a second grid matrix corresponding to the remote sensing image calculated by the maximum value in the receptive field, the output end of the third convolution layer is respectively connected with the input end of the fourth convolution layer and the input end of the second pooling layer, the fourth convolution layer outputs a third grid matrix corresponding to the most relevant characteristic factor calculated by the convolution operator, and the second pooling layer outputs a fourth grid matrix corresponding to the most relevant characteristic factor calculated by the maximum value in the receptive field;
Predicting multi-channel image data corresponding to an area to be identified by utilizing the final landslide potential hazard prediction model to obtain surface points with landslide potential hazards in the area to be identified;
identifying landslide hidden danger of the ground surface point location based on a pre-constructed landslide hidden danger rule set; the landslide hidden danger is a deformation area, a historical deformation damage area or a potential unstable slope.
2. The method of claim 1, wherein the landslide hazard feature comprises a morphological feature, a deformation feature, a situational feature, and other features; wherein the situational characteristics include disaster factors of landslide and disaster range of landslide.
3. The method of identifying as in claim 2, wherein identifying the most relevant feature factor corresponding to the training sample set from the plurality of candidate feature factors comprises:
by calculation formula
Obtaining a correlation degree coefficient among the candidate feature factorsThe method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Candidate feature factors->And->Candidate feature factors->Correlation coefficient between->，/>，/>Representing the total number of candidate feature factors, +.>Representing variance->Indicate- >Candidate feature factors->And->Candidate feature factors->Covariance between;
by calculation formula
Obtaining the variance expansion coefficient of each candidate characteristic factorThe method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Indicate->Candidate feature factorsCoefficient of expansion of variance,/">Indicating removal->Is->The certainty factor is obtained when the other candidate feature factors are subjected to linear regression, and the variance expansion coefficient is used for measuring the co-linearity degree inside each candidate feature factor;
and respectively eliminating candidate characteristic factors with the correlation degree coefficient larger than a preset correlation degree threshold value and candidate characteristic factors with the variance expansion coefficient larger than a preset variance expansion threshold value from the candidate characteristic factors to obtain a plurality of correlation characteristic factors.
4. The method of identifying of claim 3, wherein the determining the most relevant feature factor corresponding to the training sample set from the plurality of candidate feature factors further comprises:
by calculation formula
Obtaining a landslide hidden danger characteristic pair training sample setInformation gain of->The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Characteristic of landslide hazard->For training sample set->Information gain of->Is any one of morphological characteristics, deformation characteristics, situational characteristics and other characteristics, +. >Representing the characteristic of landslide hazard->Dividing training sample set->The number of sub-collections obtained in the process,indicate->The number of training samples in the subset, +.>，/>Representing training sample set +.>Total number of training samples, +.>Indicate->Information entropy of the individual subsets, +.>Indicate->Total number of categories of relevant feature factors in the respective subset,/->Indicate->The category of the relevant characteristic factors is +.>Probability of occurrence in the individual subsets;
and taking the landslide hidden danger feature corresponding to the maximum information gain as the most relevant landslide hidden danger feature, and taking the relevant feature factor corresponding to the most relevant landslide hidden danger feature as the most relevant feature factor.
5. The identification method of claim 1, wherein the landslide hazard prediction model further comprises a first processing unit, a second processing unit, a channel rearrangement module, a third pooling layer, a fourth pooling layer, a fifth pooling layer, a third processing unit, a fifth convolution layer, a sixth convolution layer, a fourth processing unit, and a CBAM module;
the input end of the first processing unit is respectively connected with the output end of the second convolution layer in the branch I and the output end of the first pooling layer, the output end of the first processing unit is connected with the input end of the channel rearrangement module, the input end of the second processing unit is respectively connected with the output end of the fourth convolution layer in the branch II and the output end of the second pooling layer, the output end of the second processing unit is connected with the input end of the channel rearrangement module, the output end of the channel rearrangement module is respectively connected with the input end of the third pooling layer, the input end of the fifth pooling layer and the input end of the fifth convolution layer, the output end of the third pooling layer is connected with the input end of the fourth processing unit, the input end of the fourth pooling layer is connected with the output end of the third processing unit, and the output end of the fourth pooling layer is connected with the input end of the fourth processing unit; the output end of the fifth pooling layer is connected with the input end of the third processing unit, the output end of the fifth convolution layer is connected with the input end of the third processing unit, the output end of the third processing unit is connected with the input end of the fifth pooling layer, the output end of the sixth convolution layer is connected with the input end of the fourth processing unit, the input end of the CBAM module is connected with the output end of the fourth processing unit, and the CBAM module outputs a landslide hidden danger area prediction result.
6. The identification method according to claim 5, wherein the landslide hazard prediction model further comprises a U-net network for identifying disaster-bearing bodies;
training the initial landslide potential prediction model by using the training sample set to obtain a final landslide potential prediction model, wherein the training sample set comprises the following steps:
step i, through a calculation formula
Obtaining a predicted loss value of the initial landslide hazard prediction modelThe method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Representing weight parameters->Represents the total number of training samples, +.>Indicate->True tag of individual training samples, +.>，/>Output label for representing initial landslide hazard prediction model>Probability of->Representing the initial landslide potential prediction model inputExtra>Predictive value of the label of the individual training samples, +.>Output of the initial landslide hazard prediction model is represented by +.>Probability of predicted value of label of each training sample;
step ii, if the predicted loss value is smaller than or equal to a preset loss threshold value, using the initial landslide potential hazard prediction model as the final landslide potential hazard prediction model; and (c) if not, carrying out back propagation on the initial landslide hidden danger prediction model by using the prediction loss value, and returning to the step (i).
7. The method for identifying according to claim 6, wherein predicting the multichannel image data corresponding to the area to be identified using the final landslide hazard prediction model to obtain the ground surface point where the landslide hazard exists in the area to be identified includes:
dividing the multichannel image data corresponding to the area to be identified; wherein the multi-channel image data is divided into a plurality of sub-images;
inputting each sub-image in the plurality of sub-images into the final landslide potential prediction model to obtain a prediction value of the landslide potential in the sub-region corresponding to the sub-image; the sub-region represents a portion of the region to be identified;
and taking the earth surface points corresponding to all the subareas with the predicted values larger than the preset threshold value as the earth surface points with landslide hidden danger.
8. A terminal device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the method for identifying landslide hazards according to any one of claims 1 to 7 when executing the computer program.
9. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the method of identifying landslide hazards of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310844794.9A CN116561536B (en) | 2023-07-11 | 2023-07-11 | Landslide hidden danger identification method, terminal equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310844794.9A CN116561536B (en) | 2023-07-11 | 2023-07-11 | Landslide hidden danger identification method, terminal equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116561536A CN116561536A (en) | 2023-08-08 |
| CN116561536B true CN116561536B (en) | 2023-11-21 |
Family
ID=87503997
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310844794.9A Active CN116561536B (en) | 2023-07-11 | 2023-07-11 | Landslide hidden danger identification method, terminal equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116561536B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117292275A (en) * | 2023-11-24 | 2023-12-26 | 江西省国土空间调查规划研究院 | Landslide prediction method and equipment based on land utilization change |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20160061445A (en) * | 2014-11-17 | 2016-06-01 | 서울대학교산학협력단 | Method and System of Construction of Landslide Hazard Map During Earthquakes Considering Geometrical Amplification Characteristics of Slope |
| CN109165424A (en) * | 2018-08-03 | 2019-01-08 | 四川理工学院 | A kind of landslide assessment of easy generation method based on domestic GF-1 satellite data |
| CN111222466A (en) * | 2020-01-08 | 2020-06-02 | 武汉大学 | Remote sensing image landslide automatic detection method based on three-dimensional space-channel attention mechanism |
| WO2020244261A1 (en) * | 2019-06-05 | 2020-12-10 | 中国科学院长春光学精密机械与物理研究所 | Scene recognition system for high-resolution remote sensing image, and model generation method |
| CN113866764A (en) * | 2021-08-13 | 2021-12-31 | 成都理工大学 | Landslide susceptibility improvement evaluation method based on InSAR and LR-IOE model |
| CN114186641A (en) * | 2021-12-16 | 2022-03-15 | 长安大学 | Landslide susceptibility evaluation method based on deep learning |
| CN114882361A (en) * | 2022-05-13 | 2022-08-09 | 东南大学 | Deep learning forest overground biological estimation method based on multi-source remote sensing fusion |
| CN115631162A (en) * | 2022-10-21 | 2023-01-20 | 云南省地质环境监测院(云南省环境地质研究院) | Landslide hidden danger identification method, system, medium and equipment |
| CN115830469A (en) * | 2022-11-25 | 2023-03-21 | 中国科学院空天信息创新研究院 | Multi-mode feature fusion based landslide and surrounding ground object identification method and system |
| CN116108758A (en) * | 2023-04-10 | 2023-05-12 | 中南大学 | Landslide susceptibility evaluation method |
| CN116152666A (en) * | 2023-03-17 | 2023-05-23 | 中南大学 | Cross-domain remote sensing image self-adaptive learning method considering ground feature weather heterogeneity |
| CN116257742A (en) * | 2023-05-12 | 2023-06-13 | 中大智能科技股份有限公司 | Landslide displacement dynamic prediction method and device, electronic equipment and storage medium |
-
2023
- 2023-07-11 CN CN202310844794.9A patent/CN116561536B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20160061445A (en) * | 2014-11-17 | 2016-06-01 | 서울대학교산학협력단 | Method and System of Construction of Landslide Hazard Map During Earthquakes Considering Geometrical Amplification Characteristics of Slope |
| CN109165424A (en) * | 2018-08-03 | 2019-01-08 | 四川理工学院 | A kind of landslide assessment of easy generation method based on domestic GF-1 satellite data |
| WO2020244261A1 (en) * | 2019-06-05 | 2020-12-10 | 中国科学院长春光学精密机械与物理研究所 | Scene recognition system for high-resolution remote sensing image, and model generation method |
| CN111222466A (en) * | 2020-01-08 | 2020-06-02 | 武汉大学 | Remote sensing image landslide automatic detection method based on three-dimensional space-channel attention mechanism |
| CN113866764A (en) * | 2021-08-13 | 2021-12-31 | 成都理工大学 | Landslide susceptibility improvement evaluation method based on InSAR and LR-IOE model |
| CN114186641A (en) * | 2021-12-16 | 2022-03-15 | 长安大学 | Landslide susceptibility evaluation method based on deep learning |
| CN114882361A (en) * | 2022-05-13 | 2022-08-09 | 东南大学 | Deep learning forest overground biological estimation method based on multi-source remote sensing fusion |
| CN115631162A (en) * | 2022-10-21 | 2023-01-20 | 云南省地质环境监测院(云南省环境地质研究院) | Landslide hidden danger identification method, system, medium and equipment |
| CN115830469A (en) * | 2022-11-25 | 2023-03-21 | 中国科学院空天信息创新研究院 | Multi-mode feature fusion based landslide and surrounding ground object identification method and system |
| CN116152666A (en) * | 2023-03-17 | 2023-05-23 | 中南大学 | Cross-domain remote sensing image self-adaptive learning method considering ground feature weather heterogeneity |
| CN116108758A (en) * | 2023-04-10 | 2023-05-12 | 中南大学 | Landslide susceptibility evaluation method |
| CN116257742A (en) * | 2023-05-12 | 2023-06-13 | 中大智能科技股份有限公司 | Landslide displacement dynamic prediction method and device, electronic equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| PSO-RBF神经网络在历史滑坡发育地区滑坡危险性预测中的应用;张铎;;地理与地理信息科学(04);全文 * |
| 地理因子支持下的滑坡隐患点空间分布特征及识别研究;张福浩;朱月月;赵习枝;张杨;石丽红;刘晓东;;武汉大学学报(信息科学版)(08);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116561536A (en) | 2023-08-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Goetz et al. | Evaluating machine learning and statistical prediction techniques for landslide susceptibility modeling | |
| Tonini et al. | Rockfall detection from terrestrial LiDAR point clouds: A clustering approach using R | |
| Chen et al. | A practical trial of landslide detection from single-temporal Landsat8 images using contour-based proposals and random forest: A case study of national Nepal | |
| Jena et al. | Earthquake hazard and risk assessment using machine learning approaches at Palu, Indonesia | |
| Wang et al. | Change detection-based co-seismic landslide mapping through extended morphological profiles and ensemble strategy | |
| Su et al. | Deep convolutional neural network–based pixel-wise landslide inventory mapping | |
| WO2008118977A1 (en) | Data analysis process | |
| Wan | Entropy-based particle swarm optimization with clustering analysis on landslide susceptibility mapping | |
| CN116561536B (en) | Landslide hidden danger identification method, terminal equipment and medium | |
| US11676375B2 (en) | System and process for integrative computational soil mapping | |
| Bai et al. | Object-based building damage assessment methodology using only post event ALOS-2/PALSAR-2 dual polarimetric SAR intensity images | |
| Cui et al. | A benchmark evaluation of similarity measures for multitemporal SAR image change detection | |
| Gašparović | Urban growth pattern detection and analysis | |
| Fekete et al. | Tree segmentation and change detection of large urban areas based on airborne LiDAR | |
| CN110532974B (en) | High-resolution remote sensing geological disaster information automatic extraction method based on deep learning | |
| Dmitriev et al. | Spectral texture classification of high-resolution satellite images for the state forest inventory in Russia | |
| Weidner et al. | An algorithm for measuring landslide deformation in terrestrial lidar point clouds using trees | |
| CN113688172B (en) | Landslide susceptibility evaluation model training method, landslide susceptibility evaluation device and medium | |
| Loebel et al. | Extracting glacier calving fronts by deep learning: The benefit of multispectral, topographic, and textural input features | |
| Wang et al. | Machine learning powered high-resolution co-seismic landslide detection | |
| Feizizadeh et al. | Systematic evaluation of fuzzy operators for object-based landslide mapping | |
| CN114882020B (en) | Product defect detection method, device, equipment and computer readable medium | |
| CN115331096B (en) | Mining subsidence area identification method, system, storage medium and electronic equipment | |
| WO2023022587A1 (en) | Geographic data processing methods and systems for assessing geohazard risk | |
| Chen et al. | Urban damage estimation using statistical processing of satellite images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |