Landslide early warning method and system based on image data analysis

Abstract

The invention discloses a landslide early warning method and system based on image data analysis, and relates to the technical field of image data analysis, and the method comprises the steps: defining the stability index of a side slope according to the geological related data of the side slope, dividing the type of the side slope, and determining the stability condition of the analyzed side slope. The overall structure model of the slope is divided into a plurality of areas, the judgment time scale is set, a time length or window is determined, various change factors can be considered, the time window can accurately capture the change condition of image features, and unstable key areas of the slope can be conveniently screened out. And carrying out detailed splitting on the key region to obtain a multi-dimensional part of each region, and generating image content. Image contents are integrated according to position space correlation among the key areas, and early warning of the landslide phenomenon of the slope is realized by combining a slope stability index. The adaptability and accuracy of landslide early warning are improved, and safe operation of the slope is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of image data analysis, and particularly to a landslide warning method and system based on image data analysis. Background Art

[0002] With the rapid development of remote sensing technology, unmanned aerial vehicle technology, and image recognition technology, the ability to acquire and process landslide-related image data has been significantly improved. When a landslide disaster occurs or is about to occur, the instability of the surface soil and rocks and the imbalance of the slope body are often manifested as specific morphological, color, texture, and other features in the image. By extracting and analyzing these features, macroscopic and rapid identification and warning of landslide disasters can be achieved.

[0003] In the prior art, traditional landslide warning is often achieved by analyzing the forces acting on the slope. However, the force conditions of the slope are not easily directly, real-time, and comprehensively collected, resulting in poor adaptability and accuracy of landslide warning and unable to effectively ensure the safety of the slope.

[0004] Therefore, how to improve the adaptability and accuracy of landslide warning is a technical problem to be solved at present. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem of poor adaptability and accuracy of landslide warning existing in the prior art due to the difficulty in collecting the force conditions of the slope, and to propose a landslide warning method based on image data analysis. The method includes:

[0006] Defining a stability index of the slope according to the geological-related data of the slope, and dividing the slope types according to the stability index;

[0007] Constructing an overall structure model of the slope through the overall structure image and geological-related data of the slope, splitting the overall structure model of the slope into multiple regions, setting a judgment time scale, generating a time series image of each region, and screening out key regions;

[0008] Splitting the key regions to obtain multi-dimensional parts of the key regions, extracting image features according to the images of the multi-dimensional parts of the key regions, and generating image content;

[0009] Integrating the image content according to the spatial association of the positions of the key regions with each other, and then combining with the slope stability index to realize early warning of the landslide phenomenon of the slope.

[0010] In some embodiments of the present application, defining a stability index of the slope according to the geological-related data of the slope includes:

[0011] Calculating multiple stable description parameters by virtue of the geological-related data of the slope;

[0012] Collect historical dangerous accident records of the slope, extract accident parameters under each type of accident from the historical dangerous accident records, analyze the correlation between each type of accident and the stability description parameters, and thus screen out the dangerous accident types;

[0013] Define the stability index of the slope according to the stability description parameters and the accident parameters of the screened dangerous accident types;

[0014]

[0015] Among them, S is the stability index of the slope, n1 and n2 are the numbers of stability description parameters and the screened dangerous accident types respectively, are the combined weights of the i 1 th stability description parameter and the combined weights of the i 2 th dangerous accident type respectively, is the size of the i 1 th stability description parameter, is the degree of influence on the slope stability determined by multiple accident parameters under the i 2 th dangerous accident type, and k 1 is a preset constant.

[0016] In some embodiments of the present application, an overall structure model of the slope is constructed through the overall structure image of the slope, including,

[0017] The collected multi-source overall structure images of the slope and geological-related data are classified into multiple image data sets according to different data sources;

[0018] Analyze the existence of slope feature points in each image data set, select one image data set as the registration reference, load multiple image data sets and geological-related data in the registration tool, and perform spatial registration on multiple image data sets and geological-related data by virtue of the registration reference, so as to form an overall structure model of the slope.

[0019] In some embodiments of the present application, analyzing the existence of slope feature points in each image data set and selecting one image data set as the registration reference includes,

[0020] The slope feature points include one or more of terrain mutation points, geological structure lines, natural landmark points and artificial landmark points;

[0021] Divide the overall area of the slope into multiple parts evenly, and analyze the density and type quantity of slope feature points such as terrain mutation points, geological structure lines, natural landmark points and artificial landmark points under each part;

[0022] Distinguish the qualified part and the unqualified part by the density and type quantity of slope feature points, calculate the first quality of the qualified part based on the variability of slope feature points of each type, and generate the second quality according to the respective quantities of the qualified part and the unqualified part;

[0023] Determine the third quality of each image dataset by integrating the first quality and the second quality, and screen out an image dataset as the registration benchmark according to the third quality.

[0024] In some embodiments of the present application, set a judgment time scale to generate time series images of each region, including,

[0025] Collect the climate information of the slope, obtain the first time scale interval according to the slope type and climate information of the slope, statistically obtain the second time scale interval of the deformation rate and displacement rate of the slope, and use the intersection time interval of the first time scale interval and the second time scale interval as the time scale;

[0026] Capture images of each region through the time scale, and generate time series images of each region in chronological order.

[0027] In some embodiments of the present application, screen out key regions, including,

[0028] Extract the regional edge features on the time series images of each region to obtain the regional edge features under each frame of image, and compare the regional edge features under multiple frames of images on the time series images to obtain the change situation of the regional edge features;

[0029] Quantify the change situation of the regional edge features, and after quantification, evaluate and integrate to generate a regional edge change index, and screen out key regions by virtue of the regional edge change index.

[0030] In some embodiments of the present application, split the key region to obtain the multi-dimensional part of the key region, including,

[0031] Identify multiple dimensional contents in the key region. The multiple dimensional contents include cracks, vegetation coverage, key parts of the slope, and groundwater conditions, and layer the key region according to the dimensional contents to obtain the multi-dimensional part of the key region.

[0032] In some embodiments of the present application, integrate the image content according to the spatial association between key regions, and then combine with the slope stability index to realize early warning of the landslide phenomenon of the slope, including,

[0033] Mark the positions of the multi-dimensional parts of each key region on the overall structure model of the slope, and conduct multi-dimensional part association between key regions, evaluate the mutual influence under the same dimensional part between different key regions, and generate a dimensional index under each dimensional part;

[0034] The landslide warning value is obtained based on the dimensional index and the slope stability index, and the landslide warning value is used to describe the degree of danger and emergency of the occurrence of the landslide phenomenon of the slope;

[0035]

[0036] Among them, Lw is the landslide warning value, σ is the warning conversion coefficient, m is the number of dimensional parts, β j is the stability weight of the j-th dimensional part, C j is the dimensional index of the j-th dimensional part, max(β j C j is the maximum value in β j C j S is the slope stability index, k 2 is the slope stability constant, S→k 2 represents the slope stability constant mapped from the slope stability index.

[0037] Correspondingly, the present application also provides a landslide warning system based on image data analysis, which includes,

[0038] The first module is used to define the slope stability index according to the geological related data of the slope and divide the slope types according to the stability index;

[0039] The second module is used to construct an overall structure model of the slope through the overall structure image and geological related data of the slope, split the overall structure model of the slope into multiple regions, set the judgment time scale, generate the time series image of each region, and screen out the key regions;

[0040] The third module is used to split the key region to obtain the multi-dimensional parts of the key region, extract image features according to the images of the multi-dimensional parts of the key region, and generate image content;

[0041] The fourth module is used to integrate the image content according to the spatial association of the positions of the key regions with each other, and then combine the slope stability index to realize the warning of the landslide phenomenon of the slope.

[0042] By applying the above technical solutions, the stability index of the slope is defined based on the geological relevant data of the slope, and the slope types are divided, so as to determine the stability of the analyzed slope. The overall structure model of the slope is split into multiple regions, and the judgment time scale is set to determine a time length or window that can take into account various changing factors, so that this time window can accurately capture the changes in image features, facilitating the screening of the key regions where the slope is unstable. The key regions are detailedly split to obtain the multi-dimensional parts of each region, and the image content is generated. The image content is integrated according to the spatial association between the key regions, and combined with the slope stability index to realize the early warning of the landslide phenomenon of the slope. The adaptability and accuracy of the landslide early warning are improved, ensuring the safe operation of the slope. Description of the Drawings

[0043] Figure 1 It is a schematic structural diagram of the landslide early warning method based on image data analysis proposed by the present invention;

[0044] Figure 2 It is a schematic structural diagram of the landslide early warning system based on image data analysis proposed by the present invention. Detailed Embodiments

[0045] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0046] Refer to Figure 1 , the landslide early warning method based on image data analysis includes the following steps:

[0047] Step S101, define the stability index of the slope according to the geological relevant data of the slope, and divide the slope types according to the stability index.

[0048] In this embodiment, the geological relevant data includes detailed data such as geological structure, rock and soil layer distribution, groundwater conditions, and historical landslide records of the slope area collected through means such as geological exploration, drilling and sampling, and geophysical exploration. These data are the basis for defining the stability index. According to the degree of stability, it is divided into stable slopes, unstable slopes, and limit equilibrium state slopes.

[0049] The following is a detailed description of these three slope types:

[0050] 1. Stable Slope

[0051] Definition:

[0052] A stable slope refers to a slope where the rock and soil mass can maintain a relatively stable state under natural or artificial conditions and will not undergo obvious sliding or deformation.

[0053] Features:

[0054] The conditions such as the slope gradient, the nature and structure of the rock and soil of the slope are relatively favorable and can resist the action of external forces.

[0055] There may be some small cracks or depressions on the surface of the slope, but the overall stability is good and it will not pose a threat to the surrounding environment and human activities.

[0056] Examples:

[0057] In mountainous or hilly areas, highway slopes, railway slopes, etc. that have been reasonably designed and constructed often belong to stable slopes. These slopes can maintain good stability during long-term use, ensuring the safety and smoothness of traffic.

[0058] 2. Unstable slopes

[0059] Definition:

[0060] An unstable slope refers to a slope whose rock and soil mass is prone to sliding or deformation under the action of external forces, posing a threat to the surrounding environment and human activities.

[0061] Features:

[0062] The slope gradient of the slope is relatively steep, the nature of the rock and soil is poor, or there are obvious unfavorable conditions such as weak surfaces and joint surfaces.

[0063] Larger cracks, uplifts or settlements may appear on the surface of the slope, indicating that the internal stress state of the slope has changed.

[0064] Under the action of external forces such as rainfall and earthquake, unstable slopes are prone to sliding or collapse, causing damage to the buildings and roads below.

[0065] Examples:

[0066] In some mountainous or hilly areas with complex geological conditions, due to factors such as poor rock and soil properties and steep slopes, unstable slopes are likely to occur. These slopes need to take reinforcement measures such as setting up retaining walls and anti-slide piles to improve their stability.

[0067] 3. Slopes in the limit equilibrium state

[0068] Definition:

[0069] A slope in the limit equilibrium state refers to a slope whose rock and soil mass is in a critical state of about to slide or deform under the action of external forces. The stability of this kind of slope is relatively poor and its change dynamics need to be closely monitored.

[0070] Features:

[0071] The slope gradient, geotechnical properties, structure and other conditions of the slope are relatively unfavorable, but it has not reached the state of complete instability.

[0072] Some signs indicating sliding may appear on the slope surface, such as crack expansion, soil loosening, etc.

[0073] Under the action of external forces, the slope in the limit equilibrium state is prone to local sliding or deformation, which has a certain impact on the surrounding environment and human activities.

[0074] In some embodiments of the present application, the stability index of the slope is defined according to the geological relevant data of the slope, including,

[0075] Calculating a plurality of stability description parameters based on the geological relevant data of the slope;

[0076] Collecting the historical dangerous accident records of the slope, extracting the accident parameters under each type of accident from the historical dangerous accident records, and analyzing the correlation between each type of accident and the stability description parameters, so as to screen the dangerous accident types;

[0077] Defining the stability index of the slope according to the stability description parameters and the accident parameters of the screened dangerous accident types;

[0078]

[0079] Wherein, S is the stability index of the slope, n1 and n2 are respectively the number of stability description parameters and the number of screened dangerous accident types, are respectively the combined weight of the i 1 th stability description parameter, the combined weight of the i 2 th dangerous accident type, is the size of the i 1 th stability description parameter, is the degree of influence on the slope stability determined by multiple accident parameters under the i 2 th dangerous accident type, k 1 is a preset constant.

[0080] In this embodiment, the stability description parameters are calculated based on the basic parameters of the geological relevant data of the slope. For example, the slope safety factor, slope coefficient, anti-sliding force of the potential landslide surface, parameters related to the permeability coefficient and hydrogeological conditions of the slope, etc. The slope safety factor is usually expressed as the ratio of the anti-sliding force of the slope to the sliding force. By calculating this coefficient, it is possible to evaluate whether the slope is stable under the current conditions and the degree of stability.

[0081] In this embodiment, if a landslide has occurred on the slope in the past, this usually means that there are some factors of instability accidents to a certain extent on the slope. Other dangerous accidents such as debris flows and mountain collapses may also affect the slope stability. Here, the correlation between each type of accident and the stability description parameters is analyzed to screen the accident types, and the correlation can be calculated by the Pearson correlation coefficient. The accident parameters include the scale, frequency, duration, and the degree of damage to the slope structure of the accident, etc., and comprehensively generate the degree of influence on the slope stability.

[0082] In this embodiment, represents the correction of the sum of the stability description parameters by the dangerous accident type, and k 1 is to balance the magnitude of the correction function.

[0083] Step S102, construct an overall structure model of the slope through the overall structure image and geological related data of the slope, split the overall structure model of the slope into multiple regions, set the judgment time scale, generate the time series images of each region, and screen out the key regions.

[0084] In this embodiment, use remote sensing technology, UAV aerial photography or ground measurement and other means to obtain the overall structure image of the slope. Use GIS or 3D modeling software to construct the overall structure model of the slope according to the image data.

[0085] In some embodiments of the present application, constructing an overall structure model of the slope through the overall structure image of the slope includes,

[0086] Collect the multi-source overall structure images and geological related data of the slope, and divide them into multiple image data sets according to different data sources;

[0087] Analyze the existence of slope feature points in each image data set, select one image data set as the registration reference, load multiple image data sets and geological related data in the registration tool, and perform spatial registration on multiple image data sets and geological related data by virtue of the registration reference, so as to form the overall structure model of the slope.

[0088] In this embodiment, for image preprocessing, preprocessing operations such as denoising, enhancement, registration, and correction are performed on the overall structure image of the slope to improve the quality and accuracy of the image. This helps to ensure that features such as terrain, landform, and distribution of rock and soil layers in the image are clearly distinguishable. For geological data preprocessing, information such as collected geological exploration reports, rock and soil layer parameters, and geological structures is sorted out and analyzed to ensure its accuracy and integrity. There may be many sources of overall structure image data, and image data calibration is required. The geological data and the image data are fused so that the geological information can be accurately corresponding to the corresponding positions on the image. Registration tools include GIS software, remote sensing image processing software, etc. The registered geological data and image data are superimposed to form an image containing geological information. During the superimposition process, methods such as transparency adjustment and color mapping can be used to highlight geological features.

[0089] In some embodiments of the present application, the presence of slope feature points in each image dataset is analyzed, and one image dataset is selected as the registration benchmark, including

[0090] The slope feature points include one or more of terrain mutation points, geological structure lines, natural landmark points, and artificial landmark points;

[0091] The overall area of the slope is evenly divided into multiple parts, and the density and type quantity of slope feature points such as terrain mutation points, geological structure lines, natural landmark points, and artificial landmark points under each part are analyzed;

[0092] The qualified parts and unqualified parts are distinguished by the density and type quantity of slope feature points, and the first quality of the qualified parts is calculated based on the variability of each type of slope feature points, and the second quality is generated according to the respective quantities of the qualified parts and unqualified parts;

[0093] The third quality of each image dataset is determined by integrating the first quality and the second quality, and one image dataset is selected as the registration benchmark according to the third quality.

[0094] In this embodiment, the relevant situations of slope feature points in each part of each image dataset are analyzed, and thus the image dataset with good quality is used as the registration benchmark. The feature points on the slope structure refer to those points that have obvious signs or unique properties in terms of terrain, landform, or geological structure. These feature points can be identified in multi-source images or multi-source geological data, and thus can be used as the basis for registration. The following are some common slope structure feature points:

[0095] Terrain mutation points:

[0096] Such as positions with obvious terrain changes such as ridge lines, valley lines, and toe lines.

[0097] These points usually appear as the intersections or turning points of lines on the topographic map.

[0098] Geological structure lines:

[0099] The outcrop positions on the earth's surface of geological structures such as fault lines, joint lines, etc.

[0100] These lines are clearly represented in multi-source geological data (such as geological maps, remote sensing images).

[0101] Artificial buildings or markers:

[0102] Artificially constructed objects such as roads, bridges, houses, steles, etc.

[0103] These objects usually have clear shapes and positions in remote sensing images and may also be marked on geological maps.

[0104] Natural markers:

[0105] Markers in nature such as unique trees, rocks, water bodies, etc.

[0106] These markers are recognizable in multi-source images but may be affected by factors such as seasons and weather.

[0107] In this embodiment, the qualified part and the unqualified part are distinguished by the density and the number of types of slope feature points. The part with high density and many types of feature points is taken as the qualified part. The first quality of the qualified part is calculated by virtue of the variability of each type of slope feature point. The change degrees of different feature points are different, and the first quality of each qualified part is analyzed by combining the quantity and density.

[0108] The second quality is generated according to the respective quantities of the qualified part and the unqualified part. The ratio of the two is used as the second quality. The first quality describes the quality of the qualified part, and the second quality describes the quality of the overall slope. The third quality is obtained by combining the two.

[0109] In some embodiments of the present application, a judgment time scale is set, and a time series image of each region is generated, including

[0110] Collect the climate information of the slope, obtain the first time scale interval according to the slope type and climate information of the slope, statistically obtain the second time scale interval of the deformation rate and displacement rate of the slope, and use the intersection time interval of the first time scale interval and the second time scale interval as the time scale;

[0111] Capture the images of each region through the time scale, and generate the time series image of each region in chronological order.

[0112] In this embodiment, in order to capture the dynamic changes of the slope, it is necessary to analyze the changes in the slope image features, so a time span (time scale) needs to be determined. Collect historical climate data of the area where the slope is located, including rainfall, temperature, humidity, wind direction and wind speed, etc. Analyze the influence degree of different climate conditions on the stability of different types of slopes, calculate an average value, and thus determine the first time scale interval. Use professional monitoring equipment and technologies, such as GPS, radar interferometry (InSAR), etc., to regularly measure the deformation rate and displacement rate of the slope. The faster the change, the shorter the time scale. Capture images of each area according to the time scale, set a period, and regularly obtain image data.

[0113] In some embodiments of the present application, key areas are screened out, including

[0114] Extract the regional edge features on the time series images of each area to obtain the regional edge features under each frame of image. Compare the regional edge features under multiple frames of images on the time series images to obtain the change situation of the regional edge features;

[0115] Quantify the change situation of the regional edge features. After quantification, evaluate and integrate to generate a regional edge change index, and screen out key areas based on the regional edge change index.

[0116] In this embodiment, based on the above time scale, image features (edge features) are extracted. The change situation of the regional edge features includes indicators such as the displacement amount, deformation amount or change rate of the edge features, etc., to quantify the deformation situation of the area. According to the change situation of the edge features, select the areas with larger deformation, displacement or change rate as key areas. These areas are areas with a higher risk of slope landslide and need to be focused on and monitored.

[0117] Step S103: Split the key area to obtain the multi-dimensional parts of the key area, and extract image features according to the images of the multi-dimensional parts of the key area to generate image content.

[0118] In this embodiment, the splitting of the key area here refers to specific layering. Each layer can reflect the changes of the slope from one dimension, and these changes can be obtained by analyzing the changes in image features.

[0119] I. Analysis of crack area images

[0120] Image content features:

[0121] Crack length and width:

[0122] On the image, the length of the crack appears as continuous or discontinuous dark lines, and its extension direction may be consistent with or intersect the slope direction of the slope.

[0123] The width of the crack is reflected by the thickness or the change in light and shade of the line. A wider crack may appear as an obvious black or dark banded area.

[0124] Crack shape and direction:

[0125] The shape of the crack may appear as a straight line, a curve or an irregular shape, which depends on the stress state inside the slope and the properties of the soil mass.

[0126] The direction of the crack is an important basis for judging the potential sliding direction of the landslide and is usually related to the slope direction of the slope or the direction of the potential slip surface.

[0127] Crack edge characteristics:

[0128] The edge of the crack may be clear and regular, or it may be blurred and irregular, which reflects the process of crack formation and the stability of the slope soil mass.

[0129] Analysis key points:

[0130] Observe the changes in the length, width and shape of the crack to judge whether the crack is expanding or tending to be stable.

[0131] Analyze the relationship between the direction of the crack and the slope direction of the slope to evaluate the potential sliding direction of the landslide.

[0132] Check the characteristics of the crack edge to judge the stability of the slope soil mass and the landslide risk.

[0133] II. Image analysis of the slope top and toe

[0134] Image content characteristics:

[0135] Morphological changes of the slope top:

[0136] In the image, the subsidence, uplift or deformation of the slope top may be manifested as the unevenness of the ground surface, the inclination or death of the vegetation, etc.

[0137] Debris accumulation at the slope toe:

[0138] The debris accumulation at the slope toe may appear as the accumulation of gravel, soil or vegetation, and these accumulations may be evidence that the landslide has occurred or is about to occur.

[0139] Analysis key points:

[0140] Observe the morphological changes of the slope top to judge whether there are signs of subsidence, uplift or deformation of the slope.

[0141] Check whether there is debris accumulation at the slope toe to evaluate the possibility and scale of the landslide occurrence.

[0142] III. Image analysis of the vegetation-covered area on the slope

[0143] Image content features:

[0144] Vegetation distribution and density:

[0145] On the image, the distribution and density of vegetation are reflected by the coverage and density of the green area.

[0146] Vegetation growth status:

[0147] The growth status of vegetation may be manifested as the inclination of trees, the bending of the main trunk towards the slope bottom, the withering of leaves, etc.

[0148] Analysis key points:

[0149] Observe the distribution and density of vegetation to determine whether the stability of the slope is affected by vegetation.

[0150] Analyze the growth status of vegetation to evaluate whether there are signs of sliding or deformation inside the slope.

[0151] IV. Image analysis of drainage facilities on the slope

[0152] Image content features:

[0153] Drainage facility status:

[0154] On the image, the status of the drainage facility may be manifested as unobstructed, blocked or damaged, etc.

[0155] Water accumulation situation:

[0156] Water accumulation may appear as bright areas or puddles on the slope, indicating the failure of the drainage facility or the existence of a sliding surface inside the slope.

[0157] Analysis key points:

[0158] Observe the status of the drainage facility to determine whether it is unobstructed and functioning properly.

[0159] Check whether there is water accumulation on the slope and evaluate the risk of landslide occurrence.

[0160] In some embodiments of the present application, the key area is split to obtain multi-dimensional parts of the key area, including,

[0161] Identify multiple-dimensional contents in the key area. The multiple-dimensional contents include cracks, vegetation coverage, key parts of the slope, and groundwater conditions. Layer the key area according to the dimensional contents to obtain multi-dimensional parts of the key area.

[0162] In this embodiment, the dimension may also include other dimensions that can reflect the changes of the slope. One dimension is equivalent to one level. Each key area may have multiple dimensions, and each dimension needs to be analyzed.

[0163] Step S104, integrate the image content according to the positional and spatial associations between the key regions, and then combine with the slope stability index to achieve early warning of landslide phenomena on the slope.

[0164] In some embodiments of the present application, integrating the image content according to the positional and spatial associations between the key regions, and then combining with the slope stability index to achieve early warning of landslide phenomena on the slope, includes,

[0165] Mark the multi-dimensional partial positions of each key region on the overall structural model of the slope, and perform multi-dimensional partial associations between the key regions, evaluate the mutual influence under the same dimensional part between different key regions, and generate a dimensional index for each dimensional part;

[0166] Obtain a landslide warning value based on the dimensional index and the slope stability index, and describe the degree of danger and urgency of the occurrence of landslide phenomena on the slope through the landslide warning value;

[0167]

[0168] Wherein, Lw is the landslide warning value, σ is the warning conversion coefficient, m is the number of dimensional parts, β j is the stability weight of the j-th dimensional part, C j is the dimensional index of the j-th dimensional part, max(β j C j ) is the maximum value in β j C j , S is the slope stability index, k 2 is the slope stability constant, S→k 2 represents the slope stability constant mapped from the slope stability index.

[0169] In this embodiment, the image content is the information directly representing slope stability transformed from image features, such as crack size, width, shape, vegetation coverage, etc.

[0170] In this embodiment, perform correlation analysis between key regions on the marked multi-dimensional parts. This correlation analysis not only considers the interaction between different dimensional parts within the same key region, but also evaluates the mutual influence under the same dimensional part between different key regions. For example, whether the expansion trends of adjacent cracks will affect each other, or whether the change in vegetation coverage affects the overall stability of the slope. The dimensional index for each dimensional part is the slope stability index synthesized from the content on the same dimension under all key regions.

[0171] In this embodiment, represents the correction of the sum of the slope stability index and the dimension with the greatest influence on the dimension, (S→k 2) indicates the balance of the slope stability index with respect to the magnitude of the correction function.

[0172] Correspondingly, the present application also provides a landslide warning system based on image data analysis, as Figure 2 shown, which includes

[0173] A first module for defining a slope stability index based on geological data of the slope and classifying the slope types according to the stability index;

[0174] A second module for constructing an overall structure model of the slope through the overall structure image and geological data of the slope, splitting the overall structure model of the slope into multiple regions, setting a determination time scale, generating time series images of each region, and screening out key regions;

[0175] A third module for splitting the key regions to obtain multi-dimensional parts of the key regions, extracting image features based on the images of the multi-dimensional parts of the key regions, and generating image content;

[0176] A fourth module for integrating the image content according to the spatial association of the key regions with each other, and then combining with the slope stability index to realize early warning of landslide phenomena of the slope.

[0177] By applying the above technical solutions, a slope stability index is defined based on geological data of the slope, and the slope types are classified to determine the stability of the analyzed slope. The overall structure model of the slope is split into multiple regions, and a determination time scale is set to determine a time length or window, which can take into account various changing factors, so that this time window can accurately capture the changes in image features and facilitate screening out the key regions where the slope is unstable. The key regions are split in detail to obtain multi-dimensional parts of each region and generate image content.

[0178] The image content is integrated according to the spatial association of the key regions with each other, and then combined with the slope stability index to realize early warning of landslide phenomena of the slope. The adaptability and accuracy of landslide warning are improved, and the safe operation of the slope is ensured.

[0179] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented through hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present invention.

[0180] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily essential for implementing the present invention.

[0181] Those skilled in the art can understand that the modules in the system in the implementation scenario can be distributed in the system of the implementation scenario according to the description of the implementation scenario, or can be correspondingly changed and located in one or more systems different from this implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0182] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent replacements or changes, and should be covered within the protection scope of the present invention.

Claims

1. A landslide early warning method based on image data analysis, characterized in that: include, Define the stability index of the slope according to the geological data of the slope, and classify the slope types according to the stability index; The overall structural model of the slope is constructed through the overall structural image of the slope and geological related data. The overall structural model of the slope is divided into multiple areas, the judgment time scale is set, the time series image of each area is generated, and the key areas are screened out; The key area is split to obtain the multi-dimensional part of the key area, and the image features are extracted according to the image of the multi-dimensional part of the key area to generate the image content; The image contents are integrated according to the spatial relationship between the key areas, and then combined with the slope stability index to achieve early warning of slope landslides.

2. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: The slope stability index is defined based on the geological data of the slope, including: Several stability description parameters are calculated based on the geological data of the slope; Collect the historical records of dangerous accidents of the slope, extract the accident parameters of each type of accident from the historical records of dangerous accidents, analyze the correlation between each type of accident and the stability description parameters, and thus screen the types of dangerous accidents; Define the stability index of the slope according to the stability description parameters and the accident parameters of the dangerous accident types after screening; Among them, S is the stability index of the slope, n1 and n2 are the number of stability description parameters and the number of dangerous accident types after screening, respectively. are the combined weight of the i1th stable description parameter and the i2th dangerous accident type, respectively. is the size of the i1th stable description parameter, is the degree of influence on slope stability determined by multiple accident parameters under the i2th dangerous accident type, and k1 is a preset constant.

3. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: The overall structural model of the slope is constructed through the overall structural image of the slope, including: The multi-source overall structural images and geological related data of the slope are collected and divided into multiple image data sets according to different data sources; Analyze the existence of slope feature points in each image dataset, select an image dataset as the registration benchmark, load multiple image datasets and geological related data in the registration tool, and spatially register multiple image datasets and geological related data based on the registration benchmark to form an overall structural model of the slope.

4. The landslide early warning method based on image data analysis according to claim 3 is characterized in that: Analyze the existence of slope feature points in each image dataset and select one image dataset as the registration benchmark. include, The slope characteristic points include one or more of terrain mutation points, geological structure lines, natural landmarks and artificial landmarks; The whole slope area is evenly divided into multiple parts, and the density and number of slope characteristic points of terrain mutation points, geological structure lines, natural landmarks and artificial landmarks under each part are analyzed; The qualified part and the unqualified part are distinguished by the density and the number of types of the slope feature points, the first quality of the qualified part is calculated by the variability of each type of slope feature points, and the second quality is generated according to the number of the qualified part and the unqualified part respectively; The third quality of each image data set is determined by combining the first quality and the second quality, and an image data set is selected according to the third quality as a registration reference.

5. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: Set the judgment time scale and generate time series images for each area, including: Collecting climate information of the slope, obtaining a first time scale interval according to the slope type and climate information of the slope, obtaining a second time scale interval by statistically analyzing the deformation rate and displacement rate of the slope, and taking the intersection time interval of the first time scale interval and the second time scale interval as the time scale; The image of each area is captured by time scale, and a time series image of each area is generated in time sequence.

6. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: Screen out key areas, including, Extracting regional edge features on the time series images of each region to obtain regional edge features under each frame of the image, and comparing regional edge features under multiple frames of the time series images to obtain changes in regional edge features; The changes in regional edge characteristics are quantified, and after quantification, the regional edge change indicators are evaluated and integrated to generate regional edge change indicators, and key areas are screened out based on the regional edge change indicators.

7. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: Split the key area to obtain the multi-dimensional parts of the key area. include, Multiple dimensional contents in the key areas are identified, including cracks, vegetation coverage, key parts of the slope and groundwater conditions. The key areas are stratified according to the dimensional contents to obtain the multidimensional parts of the key areas.

8. The landslide early warning method based on image data analysis according to claim 1 is characterized in that: The image content is integrated according to the spatial relationship between the key areas, and then combined with the slope stability index to achieve early warning of slope landslide phenomena, including: The multi-dimensional part position of each key area is marked on the overall structural model of the slope, and the multi-dimensional parts between the key areas are associated, the mutual influence of the same dimensional part between different key areas is evaluated, and the dimensional index under each dimensional part is generated; The landslide warning value is obtained according to the dimension index and the slope stability index, and the landslide warning value is used to describe the emergency degree of the landslide phenomenon on the slope; Where Lw is the landslide warning value, σ is the warning conversion coefficient, m is the number of dimensional parts, β j is the stable weight of the j-th dimension, C j is the dimension index of the j-th dimension part, max(β j C j ) is β j C j The maximum value of , S is the slope stability index, k2 is the slope stability constant, and S→k2 represents the slope stability constant obtained by mapping the slope stability index.

9. The landslide early warning system based on image data analysis is characterized by: include, The first module is used to define the stability index of the slope according to the geological data of the slope, and classify the slope types according to the stability index; The second module is used to construct the overall structural model of the slope through the overall structural image of the slope and geological related data, split the overall structural model of the slope into multiple areas, set the judgment time scale, generate time series images of each area, and screen out key areas; The third module is used to split the key area to obtain the multi-dimensional part of the key area, extract image features according to the image of the multi-dimensional part of the key area, and generate image content; The fourth module is used to integrate the image content according to the spatial relationship between the key areas, and then combine it with the slope stability index to achieve early warning of slope landslide phenomena.

Images