Disaster damage refuge spot and path identification method based on satellite remote sensing image

Abstract

The invention relates to the technical field of disaster evacuation and risk avoidance, in particular to a disaster damage refuge spot and path identification method based on satellite remote sensing images, which comprises the following steps: S1, determining candidate refuge spots according to disaster types and real-time position information of users; s2, predicting a corresponding disaster damage condition based on the real-time satellite remote sensing image of each candidate refuge point; s3, based on the safety related information of each remaining candidate refuge point, matching a target refuge point with the highest safety; s4, generating an evacuation guide path from the real-time position of the user to the target refuge spot based on the map information and the real-time satellite remote sensing image of the area where the target refuge spot is located; and S5, guiding the user to arrive at the target refuge point from the real-time position through the evacuation guiding path. According to the method, the optimal refuge point can be matched, the optimal evacuation path can be generated, and the flexibility of disaster avoidance evacuation and the complex scene adaptability of the user are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster evacuation and risk avoidance, and in particular to a method for identifying disaster-damaged refuge points and paths based on satellite remote sensing images. Background Art

[0002] In today's society, natural and man-made disasters, such as earthquakes, floods, and fires, are frequent, posing a significant threat to people's lives and property. When disasters strike, timely and effective evacuation is crucial for minimizing casualties and property losses. The fundamental goal of disaster evacuation is to quickly and safely transfer people in dangerous areas to relatively safe shelters and ensure a smooth evacuation process.

[0003] The selection of shelters and evacuation routes plays a crucial role in disaster evacuation. Appropriate shelters can provide safe havens for evacuees, meeting their basic needs like food, water, and medical care while protecting them from further damage from the disaster. Reasonable evacuation routes are directly related to the speed and efficiency of evacuation, minimizing exposure time and reducing the risk of secondary disasters. For example, after an earthquake, choosing shelters away from potentially collapsed structures like tall buildings and bridges, as well as evacuation routes that avoid potential landslides, mudslides, and other secondary hazards, is crucial for ensuring personal safety.

[0004] However, existing technologies still have the following problems in refuge point selection and evacuation route planning:

[0005] Existing methods for selecting shelters rely on pre-designated, fixed locations. Information on these locations is often static, lacking real-time monitoring and assessment of the actual conditions at the time of a disaster. After a disaster strikes, these fixed shelters may become unsafe or unusable due to damage. For example, during a flood, buildings such as schools and gymnasiums originally designated as shelters may be submerged.

[0006] When it comes to evacuation route selection, existing technologies primarily rely on map information and historical traffic data for route planning, failing to fully consider the real-time changes in road conditions during disasters. Disasters can cause road damage, traffic congestion, bridge failures, and other conditions, rendering previously accessible roads impassable. Existing methods are unable to capture this dynamic information in a timely manner, resulting in generated evacuation routes that may pose safety risks or be impassable. For example, after an earthquake, some roads may be blocked by ground subsidence or collapsed buildings.

[0007] Therefore, how to design a method that can obtain dynamic information of refuge points and evacuation routes in real time and perform refuge point matching and evacuation route generation based on this information is a technical problem that needs to be solved urgently. Summary of the Invention

[0008] In view of the above-mentioned deficiencies of the existing technology, the technical problem to be solved by the present invention is: how to provide a method for identifying disaster-damaged shelters and paths based on satellite remote sensing images, and obtain dynamic information of shelters and evacuation paths through real-time satellite remote sensing images to match the optimal shelters and generate the optimal evacuation paths, thereby improving the flexibility of users' disaster avoidance and evacuation and their ability to adapt to complex scenarios.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0010] Methods for identifying disaster-damaged shelters and paths based on satellite remote sensing images include:

[0011] S1: Determine candidate shelters based on disaster type and user's real-time location information;

[0012] S2: Predict the corresponding disaster damage based on the real-time satellite remote sensing images of each candidate shelter and delete unavailable candidate shelters;

[0013] S3: Match the target refuge with the highest security based on the security-related information of the remaining candidate refuges;

[0014] S4: Generate an evacuation guidance path from the user's real-time location to the target shelter based on map information and real-time satellite remote sensing images of the area where the target shelter is located;

[0015] S5: Guide the user from their real-time location to the target shelter through the evacuation guidance path.

[0016] Preferably, in step S1, preliminary screening shelters are obtained by matching all shelters according to the disaster type, and then candidate shelters whose distance from the user's real-time location does not exceed a distance threshold are matched from the preliminary screening shelters.

[0017] Preferably, in step S2, the processing steps of predicting disaster damage conditions of candidate refuge sites based on real-time satellite remote sensing images include:

[0018] S201: Obtain satellite remote sensing images of refuge points used as training data and perform preprocessing;

[0019] S202: Extracting disaster feature vectors of the refuge area based on the pre-processed satellite remote sensing image; and labeling the satellite remote sensing image of the refuge with true labels of the disaster damage.

[0020] S203: Build a disaster damage prediction model based on a deep learning framework;

[0021] S204: Inputting the disaster feature vector of the refuge area into the disaster damage prediction model and outputting the corresponding disaster damage prediction label;

[0022] S205: Calculating a loss function based on the disaster damage prediction label and the corresponding disaster damage true label, and reversely optimizing the parameters of the disaster damage prediction model;

[0023] S206: Repeat steps S201 to S205 to iteratively train the disaster damage prediction model until the model converges or reaches a preset number of iterations;

[0024] S207: Extract disaster feature vectors from satellite remote sensing images of candidate refuge sites, input them into a trained disaster damage prediction model, and output corresponding disaster damage prediction labels as their disaster damage conditions.

[0025] Preferably, in step S201, the step of pre-processing the satellite remote sensing image of the refuge point includes:

[0026] S2011: Radiometric correction of satellite remote sensing images of refuge sites;

[0027] The formula is:

[0028]

[0029] Where: L x Indicates the radiation value of the corrected satellite remote sensing image; L r Indicates the radiation value of the original satellite remote sensing image; L min and L max Indicates the minimum and maximum values ​​of the original satellite remote sensing image; L t,min and L t,max Indicates the minimum and maximum values ​​of the target radiation value range;

[0030] S2012: Geometric correction of radiometrically corrected satellite remote sensing images;

[0031] The formula is:

[0032]

[0033] Where: (x′, y′) represents the pixel coordinates in the geometrically corrected satellite remote sensing image; (x, y) represents the pixel coordinates in the original satellite remote sensing image; a i 、b i represents the polynomial coefficients, which are solved by ground control points;

[0034] S2013: Perform image enhancement on the geometrically corrected satellite remote sensing image to obtain a preprocessed satellite remote sensing image.

[0035] Preferably, in step S202, the processing step of extracting the disaster feature vector of the refuge area includes:

[0036] S2021: Use the gray-level co-occurrence matrix to extract the texture features of the refuge area based on the pre-processed satellite remote sensing image;

[0037] The formula is:

[0038]

[0039] Where: F c represents the contrast of the refuge area, i.e., the texture feature; i1 and j1 represent the grayscale values ​​of the pixels in the grayscale co-occurrence matrix of the refuge area; (i1-j1) represents the grayscale difference between two adjacent pixels in the refuge area; p(i1-j1) represents the probability of grayscale values ​​i1 and j1 appearing at the same time; N represents the number of grayscale levels;

[0040] S2022: Extract the shape features of the refuge area based on the pre-processed satellite remote sensing image;

[0041] The formula is:

[0042]

[0043] Where: F a Indicates the area of ​​the refuge point, that is, the shape feature; S i represents the area of ​​the i-th pixel in the refuge area; n represents the number of pixels in the refuge area;

[0044] S2023: Extract spectral characteristics of refuge area based on pre-processed satellite remote sensing images;

[0045] The formula is:

[0046]

[0047] Where: F r represents the mean reflectivity of the kth band in the refuge area, i.e., the spectral characteristics; R km represents the reflectivity of the i-th pixel in the refuge area in the k-th band;

[0048] S2023: Concatenate the texture features, shape features, and spectral features of the refuge area to form a disaster feature vector.

[0049] Preferably, in step S3, the corresponding refuge safety index is calculated based on the safety-related information of each candidate refuge point, and then the refuge point with the highest refuge safety index is selected as the target refuge point.

[0050] Preferably, in step S3, the processing step of calculating the refuge safety index of the candidate refuge point includes:

[0051] S301: Calculate the structural safety score based on the building structure type, construction year and usable area of ​​the candidate refuge point;

[0052] The formula is:

[0053] S j =S l ×k n ×k m ;

[0054]

[0055] Where: S j Indicates the structural safety score; S l Represents the building structure type score. If the candidate refuge point is an open space, then S l is 100 points. If it is a frame structure, then S l is 90 points. If it is a brick-concrete structure, then S l is 70 points. If it is a civil structure, S l 50 points; k n represents the age correction factor; k m Indicates the available area correction factor; n d Indicates the current year; n j Indicates the construction year of the candidate refuge site; s m Indicates the available area of ​​the candidate refuge site;

[0056] S302: Calculating a personnel carrying safety score based on the real-time number of refugees and the upper limit of the number of people accommodated at the candidate refuge point;

[0057] The formula is:

[0058]

[0059] Where: S r Indicates the personnel carrying safety score; R r represents the personnel carrying rate of the candidate refuge point; r s Indicates the real-time number of refugees at candidate refuge points; r max Indicates the maximum number of people that can be accommodated at the candidate shelter;

[0060] S303: Calculate the material reserve adequacy score based on the food reserve, drinking water reserve, and medical supply reserve of the candidate shelter;

[0061] The formula is:

[0062] S w =w1·R s +w2·R w +w3·R y ;

[0063]

[0064] Where: S w Indicates the material reserve adequacy score; w1, w2, w3 indicate the set weight coefficients; R s 、R w 、R y They represent the food sufficiency rate, drinking water sufficiency rate and medical supplies sufficiency rate of candidate shelters respectively; n s 、n w 、n y They represent the food reserves, drinking water reserves, and medical supplies reserves of candidate refuge sites respectively; n s,r,t represents the daily food demand per person; t represents the expected number of days of evacuation; n w,r,t The daily drinking water requirement per person; n y,r represents the demand for medical supplies per person;

[0065] S304: When the concentration of harmful gases at the candidate refuge site is less than the safety threshold, the environmental safety score is 100 points; when the concentration of harmful gases at the candidate refuge site is equal to or greater than the safety threshold, the environmental safety score is 0 points;

[0066] S305: Perform weighted summation of the structural safety score, the personnel load safety score, the material reserve adequacy score, and the environmental safety score to obtain a refuge safety index for the candidate refuge point.

[0067] Preferably, in step S4, the processing step of generating an evacuation guidance path includes:

[0068] S401: registering map information and satellite remote sensing images;

[0069] S402: Extracting a road network from the registered map information and constructing a road network topology structure including road nodes and edges;

[0070] S403: extracting real-time road condition information and disaster-affected area information from the registered satellite remote sensing image;

[0071] S404: assigning a dynamic weight to each edge in the road network topology structure based on real-time road condition information and disaster-affected area information;

[0072] S405: Path planning is performed based on the user's real-time location information, the location of the target refuge point, the road network topology, and the dynamic weight of the edge through the A* algorithm to obtain an evacuation guidance path from the user's real-time location to the target refuge point.

[0073] Preferably, in step S404, the process of assigning a dynamic weight to each edge in the road network topology structure includes:

[0074] S4041: Calculate the basic weight w for each edge based on the road grade and speed limit information base ;

[0075] The formula is:

[0076]

[0077] Where: L represents the road grade; V represents the road speed limit; δ is an adjustment coefficient;

[0078] S4042: Adjust the basic weight according to the real-time traffic information to obtain the traffic adjustment weight w lk ;

[0079] The formula is:

[0080] w lk =w base ×(1+α·ρ);

[0081] Where: α is an adjustment coefficient used to control the influence of vehicle density on the weight; ρ represents vehicle density, which is used to measure the congestion of the road;

[0082] S4043: Adjust the road condition adjustment weight w based on the disaster-affected area information lk , get the dynamic weight w;

[0083] The formula is:

[0084]

[0085] Where: S1 indicates that the road is located in the disaster-affected area; S2 indicates that the road is not located in the disaster-affected area; M represents a fixed value.

[0086] Preferably, in step S405, the A* algorithm starts from the user's real-time location node, and each time selects the node with the smallest heuristic function value to expand until it reaches the target refuge point node;

[0087] The calculation formula of the heuristic function value is as follows:

[0088]

[0089] Where: h(n) represents the heuristic function value; d(m,g) represents the Euclidean distance from node m to target node g; β represents the weight coefficient, which is used to balance the influence of distance and road weight; p(m,g) represents the edge set on the path from node m to target node g; w e Represents the dynamic weight of edge e.

[0090] Compared with the existing technology, the method for identifying disaster-damaged shelters and paths based on satellite remote sensing images in the present invention has the following beneficial effects:

[0091] The present invention predicts the corresponding disaster damage based on real-time satellite remote sensing images of each candidate refuge point, and deletes unavailable candidate refuge points. Real-time satellite remote sensing images can accurately reflect the dynamic information of candidate refuge points when a disaster occurs. Predicting disaster damage based on real-time satellite remote sensing images can promptly identify those refuge points that have become unsafe and cannot be used normally due to the impact of the disaster, such as areas that have been flooded, structurally damaged by earthquakes, or buried by landslides. Deleting these unavailable candidate refuge points is conducive to matching users with truly safe and reliable refuge places, reducing the risk of users encountering secondary disasters during the process of going to the refuge point, thereby improving the rationality of refuge point selection during disaster evacuation. At the same time, timely deleting unavailable candidate refuge points avoids invalid calculations for unavailable refuge points, thereby improving the guidance efficiency of users during disaster evacuation.

[0092] After determining the target refuge point, the present invention generates an evacuation guidance path from the user's real-time location to the target refuge point based on map information and real-time satellite remote sensing images of the area where the target refuge point is located. Combined with map information, it can comprehensively consider factors such as the road layout, traffic conditions, and topography of the target area where the user is located. The real-time satellite remote sensing images of the target refuge point provide dynamic information about the actual environment surrounding the refuge point. The evacuation guidance path generated based on these two pieces of information can avoid roads, bridges, and other transportation facilities damaged by the disaster, and choose a more unobstructed and safe route, which is conducive to better guiding users to the target refuge point, thereby improving the safety and success rate of users' disaster evacuation. At the same time, the evacuation path is complex and changeable when a disaster occurs, and road conditions may change at any time. Real-time satellite remote sensing images can capture these changes in a timely manner, such as road blockages and the emergence of new dangerous areas. Based on this dynamic information, the evacuation guidance path can be adjusted and optimized in real time to provide users with the safest and most feasible route, thereby improving the flexibility of users' disaster evacuation and their ability to adapt to complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] In order to make the purpose, technical solutions and advantages of the invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0094] Figure 1 This is a logical block diagram of the method for identifying disaster-damaged shelters and paths based on satellite remote sensing images. DETAILED DESCRIPTION

[0095] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but only represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0096] It should be noted that similar reference numerals and letters denote similar items in the following figures. Therefore, once an item is defined in one figure, it does not require further definition or explanation in subsequent figures. In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" indicate positions or relationships based on the positions or relationships shown in the figures, or the positions or relationships in which the inventive product is typically placed when in use. These terms are intended solely to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation, and are therefore not to be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third," etc., are used solely to distinguish descriptions and are not to be construed as indicating or implying relative importance. Furthermore, terms such as "horizontal" and "vertical" do not imply that a component is absolutely horizontal or overhanging, but rather may be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but rather may be slightly tilted. In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0097] The following is a further detailed description through specific implementation methods:

[0098] Example:

[0099] This embodiment discloses a method for identifying disaster-damaged refuge points and paths based on satellite remote sensing images.

[0100] like Figure 1 As shown in FIG, the method for identifying disaster-damaged shelters and paths based on satellite remote sensing images includes:

[0101] S1: Determine candidate shelters based on disaster type and user's real-time location information;

[0102] In this embodiment, corresponding refuge points are set in advance for each disaster type (such as fire, earthquake, flood, etc.).

[0103] S2: Predict the corresponding disaster damage based on the real-time satellite remote sensing images of each candidate shelter and delete unavailable candidate shelters;

[0104] S3: Match the target shelter with the highest security based on the safety-related information of the remaining candidate shelters (including basic information, real-time number of evacuees, material reserves, and real-time environmental information);

[0105] S4: Generate an evacuation guidance path from the user's real-time location to the target shelter based on map information and real-time satellite remote sensing images of the area where the target shelter is located;

[0106] S5: Guide the user from their real-time location to the target shelter through the evacuation guidance path.

[0107] In this embodiment, during the disaster evacuation process, steps S1 to S5 are continuously repeated to dynamically update the target refuge point and evacuation guidance path. This is because the development of disasters is often dynamic, and the safety status of refuge points may also change accordingly. For example, as an earthquake continues, some buildings that once seemed safe may develop new cracks or structural damage; flood levels may continue to rise, causing previously high-lying areas to be at risk of flooding. By dynamically updating the refuge safety index of candidate refuge points and promptly adjusting the target refuge point, the present invention can better cope with the complexity and uncertainty of disasters and provide users with continuous and reliable safety protection.

[0108] The present invention determines candidate refuge points based on the disaster type and the user's real-time location information. Different disaster types have significantly different requirements for refuge points. For example, in earthquake disasters, open areas such as open squares and parks are more ideal refuge points because they can avoid damage caused by building collapse; in flood disasters, areas such as buildings and hillsides with higher terrain are more suitable as refuge places. By determining candidate refuge points based on the disaster type, potential refuge places that meet the characteristics of the current disaster can be accurately screened, thereby improving the effectiveness and pertinence of disaster evacuation. At the same time, combined with the user's real-time location information, candidate refuge points within a certain range around the user can be quickly determined, avoiding the subsequent recommendation of difficult-to-reach refuge points to the user, thereby improving the effectiveness of the user's disaster evacuation.

[0109] The present invention predicts the corresponding disaster damage based on real-time satellite remote sensing images of each candidate refuge point, and deletes unavailable candidate refuge points. Real-time satellite remote sensing images can accurately reflect the dynamic information of candidate refuge points when a disaster occurs. Predicting disaster damage based on real-time satellite remote sensing images can promptly identify those refuge points that have become unsafe and cannot be used normally due to the impact of the disaster, such as areas that have been flooded, structurally damaged by earthquakes, or buried by landslides. Deleting these unavailable candidate refuge points is conducive to matching users with truly safe and reliable refuge places, reducing the risk of users encountering secondary disasters during the process of going to the refuge point, thereby improving the rationality of refuge point selection during disaster evacuation. At the same time, timely deleting unavailable candidate refuge points avoids invalid calculations for unavailable refuge points, thereby improving the guidance efficiency of users during disaster evacuation.

[0110] After determining the target refuge point, the present invention generates an evacuation guidance path from the user's real-time location to the target refuge point based on map information and real-time satellite remote sensing images of the area where the target refuge point is located. Combined with map information, it can comprehensively consider factors such as the road layout, traffic conditions, and topography of the target area where the user is located. The real-time satellite remote sensing images of the target refuge point provide dynamic information about the actual environment surrounding the refuge point. The evacuation guidance path generated based on these two pieces of information can avoid roads, bridges, and other transportation facilities damaged by the disaster, and choose a more unobstructed and safe route, which is conducive to better guiding users to the target refuge point, thereby improving the safety and success rate of users' disaster evacuation. At the same time, the evacuation path is complex and changeable when a disaster occurs, and road conditions may change at any time. Real-time satellite remote sensing images can capture these changes in a timely manner, such as road blockages and the emergence of new dangerous areas. Based on this dynamic information, the evacuation guidance path can be adjusted and optimized in real time to provide users with the safest and most feasible route, thereby improving the flexibility of users' disaster evacuation and their ability to adapt to complex scenarios.

[0111] In order to better introduce the technical solution of the present invention, this embodiment is described through the following parts.

[0112] 1. Matching candidate shelters

[0113] In this embodiment, preliminary screening shelters are obtained by matching all shelters according to the disaster type, and then candidate shelters whose distances to the user's real-time location do not exceed a distance threshold are obtained from the preliminary screening shelters.

[0114] 2. Predicting Disaster Damage

[0115] In this embodiment, the steps of predicting disaster damage to candidate refuge sites based on real-time satellite remote sensing images include:

[0116] S201: Obtain satellite remote sensing images of refuge points used as training data and perform preprocessing;

[0117] S202: Extracting disaster feature vectors of the refuge area based on the pre-processed satellite remote sensing image; and labeling the satellite remote sensing image of the refuge with true labels of the disaster damage.

[0118] In this embodiment, manual labeling can be used to label the real labels of the disaster damage.

[0119] S203: Build a disaster damage prediction model based on a deep learning framework (convolutional neural network);

[0120] S204: Inputting the disaster feature vector of the refuge area into the disaster damage prediction model and outputting the corresponding disaster damage prediction label;

[0121] S205: Calculating a loss function based on the disaster damage prediction label and the corresponding disaster damage true label, and reversely optimizing the parameters of the disaster damage prediction model;

[0122] In this embodiment, the loss function may adopt a cross entropy loss function.

[0123] S206: Repeat steps S201 to S205 to iteratively train the disaster damage prediction model until the model converges or reaches a preset number of iterations;

[0124] S207: Extract disaster feature vectors from satellite remote sensing images of candidate refuge sites, input them into a trained disaster damage prediction model, and output corresponding disaster damage prediction labels as their disaster damage conditions.

[0125] In this embodiment, a disaster damage threshold (such as 0.8) is preset according to the disaster type and actual situation, and the disaster damage prediction label of each candidate refuge point (a value between 0 and 1) is compared with the preset disaster damage threshold, and unavailable candidate refuge points are deleted to obtain the final list of candidate refuge points.

[0126] Specifically, the steps for preprocessing the satellite remote sensing image of the refuge point include:

[0127] S2011: Radiometric correction of satellite remote sensing images of refuge sites;

[0128] When satellite sensors acquire satellite remote sensing images, they are affected by factors such as the atmosphere and lighting, which may cause errors in the radiation values ​​of the images. Therefore, radiation correction is required.

[0129] The formula is:

[0130]

[0131] Where: L x Indicates the radiation value of the corrected satellite remote sensing image; L r Indicates the radiation value of the original satellite remote sensing image; L min and L max Indicates the minimum and maximum values ​​of the original satellite remote sensing image; L t,min and L t,max Indicates the minimum and maximum values ​​of the target radiation value range;

[0132] S2012: Geometric correction of radiometrically corrected satellite remote sensing images;

[0133] Satellite remote sensing images may be geometrically distorted during acquisition due to factors such as satellite attitude and earth curvature. Therefore, it is necessary to perform geometric correction on the images using ground control points and polynomial geometric correction models.

[0134] The formula is:

[0135]

[0136] Where: (x′, y′) represents the pixel coordinates in the geometrically corrected satellite remote sensing image; (x, y) represents the pixel coordinates in the original satellite remote sensing image; a i 、b i represents the polynomial coefficients, which are solved by ground control points;

[0137] S2013: Perform image enhancement on the geometrically corrected satellite remote sensing image to obtain a preprocessed satellite remote sensing image.

[0138] In this embodiment, the contrast and clarity of the satellite remote sensing image can be enhanced by histogram equalization and sharpening to achieve image enhancement and highlight the key information in the image.

[0139] Specifically, the steps for extracting the disaster feature vector of the refuge area include:

[0140] S2021: Use the Gray Level Co-occurrence Matrix (GLCM) to extract the texture features of the refuge area based on the pre-processed satellite remote sensing image;

[0141] The formula is:

[0142]

[0143] Where: F c represents the contrast of the refuge area, that is, the texture feature; i1 and j1 represent the grayscale values ​​of the pixels in the grayscale co-occurrence matrix of the refuge area (the grayscale co-occurrence matrix describes the spatial dependence of grayscale in satellite remote sensing images); (i1-j1) represents the grayscale difference between two adjacent pixels in the refuge area; p(i1-j1) represents the probability of grayscale values ​​i1 and j1 appearing at the same time; N represents the number of grayscale levels;

[0144] In this embodiment, based on the calculation of the contrast of the refuge point area, statistics such as entropy and energy can be further calculated, and then the contrast, entropy and energy can be combined to form the texture feature of the refuge point area.

[0145] S2022: Extract the shape features of the refuge area based on the pre-processed satellite remote sensing image;

[0146] The formula is:

[0147]

[0148] Where: F a Indicates the area of ​​the refuge point, that is, the shape feature; S i represents the area of ​​the i-th pixel in the refuge area (in discrete images, the area of ​​each pixel is usually considered to be 1); n represents the number of pixels in the refuge area;

[0149] In this embodiment, the perimeter and aspect ratio can be further calculated based on the calculated area of ​​the refuge area, and then the area, perimeter, and aspect ratio are combined to form the shape characteristics of the refuge area. The perimeter can be calculated by obtaining the length of the pixel boundary through a boundary tracking algorithm.

[0150] S2023: Extract spectral characteristics of refuge area based on pre-processed satellite remote sensing images;

[0151] The formula is:

[0152]

[0153] Where: F r represents the mean reflectivity of the kth band in the refuge area, i.e., the spectral characteristics; R km represents the reflectivity of the i-th pixel in the refuge area in the k-th band;

[0154] In this embodiment, based on the calculation of the reflectivity mean of the refuge area, the standard deviation of each band can be further calculated, and the reflectivity mean and the standard deviation can be concatenated to form the spectral characteristics of the refuge area.

[0155] S2023: Concatenate the texture features, shape features, and spectral features of the refuge area to form a disaster feature vector.

[0156] 3. Calculation of the Evacuation Safety Index

[0157] In this embodiment, a corresponding refuge safety index is calculated based on the safety-related information of each candidate refuge point, and then the refuge point with the highest refuge safety index is selected as the target refuge point.

[0158] Safety-related information about candidate shelters includes basic building information, real-time number of evacuees, material reserves, and real-time environmental information;

[0159] Basic building information includes building structure type, usable area, construction year and maximum number of people accommodated;

[0160] Material reserve information includes food reserves, drinking water reserves, and medical supplies reserves;

[0161] Real-time environmental information includes hazardous gas concentrations.

[0162] In this embodiment, basic building information includes structural type (e.g., reinforced concrete, brick and wood), usable area, construction year, and maximum occupancy. Real-time evacuee population information and resource inventory (food, drinking water, and medical supplies) are uploaded by shelter managers. Real-time environmental information includes toxic and hazardous gas concentrations (carbon monoxide, sulfur dioxide, and formaldehyde), acquired through appropriate sensors.

[0163] The processing steps for calculating the refuge safety index of candidate refuge points include:

[0164] S301: Calculate the structural safety score based on the building structure type, construction year and usable area of ​​the candidate refuge point;

[0165] The formula is:

[0166] S j=S l ×k n ×k m ;

[0167] This formula indicates that for every additional year of construction, the age correction coefficient decreases by 0.01; the minimum age correction coefficient is no less than 0.5.

[0168] This formula indicates that for every additional 10,000 square meters of usable area, the usable area correction coefficient increases by 0.2, and the maximum usable area correction coefficient shall not exceed 1.

[0169] Where: S j Indicates the structural safety score; S l Represents the building structure type score. If the candidate refuge point is an open space, then S l is 100 points. If it is a frame structure, then S l is 90 points. If it is a brick-concrete structure, then S l is 70 points. If it is a civil structure, S l 50 points; k n represents the age correction factor. The older the building, the lower the safety. m Represents the available area correction coefficient. The larger the available area, the more stable the refuge space provided, but it also increases the difficulty of management. d Indicates the current year; n j Indicates the construction year of the candidate refuge site; s m Indicates the available area of ​​the candidate refuge site;

[0170] S302: Calculating a personnel carrying safety score based on the real-time number of refugees and the upper limit of the number of people accommodated at the candidate refuge point;

[0171] The formula is:

[0172] This formula means that when the personnel load rate does not exceed 50%, the score is 100 points; as the personnel load rate increases, the score gradually decreases, and when the personnel load rate exceeds 100%, the score is 0 points.

[0173]

[0174] Where: S r Indicates the personnel carrying safety score; R r represents the personnel carrying rate of the candidate refuge point; r s Indicates the real-time number of refugees at candidate refuge points; r max Indicates the maximum number of people that can be accommodated at the candidate shelter;

[0175] S303: Calculate the material reserve adequacy score based on the food reserve, drinking water reserve, and medical supply reserve of the candidate shelter;

[0176] The formula is:

[0177] S w =w1·R s +w2·R w +w3·R y ;

[0178]

[0179] Where: S w Indicates the material reserve adequacy score; w1 (set to 0.4), w2 (set to 0.4), and w3 (set to 0.4) represent the set weight coefficients; R s 、R w 、R y They represent the food sufficiency rate, drinking water sufficiency rate and medical supplies sufficiency rate of candidate shelters respectively; n s 、n w 、n y They represent the food reserves, drinking water reserves, and medical supplies reserves of candidate refuge sites respectively; n s,r,t represents the daily food demand per person; t represents the expected number of days of evacuation; n w,r,t The daily drinking water requirement per person; n y,r represents the demand for medical supplies per person;

[0180] S304: When the concentration of harmful gases at the candidate refuge site is less than the safety threshold, the environmental safety score is 100 points; when the concentration of harmful gases at the candidate refuge site is equal to or greater than the safety threshold, the environmental safety score is 0 points;

[0181] In this embodiment, the environmental safety score can be comprehensively calculated by further combining environmental information such as smoke concentration, temperature, and humidity of the candidate refuge points.

[0182] S305: Perform weighted summation of the structural safety score, the personnel load safety score, the material reserve adequacy score, and the environmental safety score to obtain a refuge safety index for the candidate refuge point.

[0183] In this embodiment, the weight of each evaluation indicator is determined by the hierarchical analysis method or entropy weight method.

[0184] The present invention calculates the refuge safety index of candidate refuge points based on safety-related information, and then selects the refuge point with the highest safety index from all candidate refuge points as the target refuge point. Safety-related information includes basic building information, real-time number of refugees information, material reserve information, and real-time environmental information. By calculating the refuge safety index using safety-related information, these complex safety factors can be quantified so that the safety of each candidate refuge point can be presented in an intuitive numerical form. The quantitative evaluation method avoids the limitations of subjective judgment and provides users with a more objective and accurate basis for refuge point selection, thereby improving the rationality of refuge point selection during disaster evacuation. At the same time, selecting the refuge point with the highest safety index as the target refuge point can maximize the safety of users during the refuge process, thereby improving the safety and reliability of refuge point selection during disaster evacuation.

[0185] 4. Generate evacuation guidance path

[0186] In this embodiment, the steps of generating an evacuation guidance path include:

[0187] S401: registering map information and satellite remote sensing images;

[0188] In this embodiment, since the coordinate systems of the map information and the satellite remote sensing image may be inconsistent, a registration operation is required. Specifically, a feature point-based registration method is used to extract feature points (such as road intersections and landmark buildings) from the map and image, and then calculate the transformation matrix between the feature points to unify the two into the same coordinate system.

[0189] S402: Extracting a road network from the registered map information and constructing a road network topology structure including road nodes and edges; wherein each road node represents an intersection or endpoint of a road, and each edge represents a road segment, and also recording the road grade and speed limit information;

[0190] S403: Using image processing technology (e.g., edge detection) to extract real-time road condition information (e.g., road congestion, whether blocked by obstacles, etc.) and disaster-affected area information (e.g., flooded areas, fire spread areas, etc.) from the registered satellite remote sensing image;

[0191] S404: assigning a dynamic weight to each edge in the road network topology structure based on real-time road condition information and disaster-affected area information;

[0192] S405: Path planning is performed based on the user's real-time location information, the location of the target refuge point, the road network topology, and the dynamic weight of the edge through the A* algorithm to obtain an evacuation guidance path from the user's real-time location to the target refuge point.

[0193] In this embodiment, a cubic spline interpolation method is used to smooth the generated evacuation guidance path, reducing the turning angles of the path and improving the safety of evacuation. If multiple paths are planned, the one with the shortest distance or the shortest time is selected as the evacuation guidance path.

[0194] Specifically, the steps for assigning dynamic weights to each edge in the road network topology include:

[0195] S4041: Calculate the basic weight w for each edge based on the road grade and speed limit information base ;

[0196] The basic weight is inversely proportional to the road grade and directly proportional to the speed limit. The formula is:

[0197]

[0198] Where: L represents the road grade (the higher the grade, the smaller the L value); V represents the road speed limit; δ is an adjustment coefficient;

[0199] S4042: Adjust the basic weight according to the real-time traffic information to obtain the traffic adjustment weight w lk If the road is congested, the weight is increased; if the road is blocked by obstacles, the weight is set to infinity.

[0200] The formula is:

[0201] w lk =w base ×(1+α·ρ);

[0202] Where: α is an adjustment coefficient used to control the influence of vehicle density on the weight; ρ represents vehicle density, which is used to measure the congestion of the road;

[0203] S4043: Adjust the road condition adjustment weight w based on the disaster-affected area information lk , and obtain the dynamic weight w; if the road is located in the disaster-affected area, the weight is greatly increased;

[0204] The formula is:

[0205]

[0206] Where: S1 indicates that the road is located in the disaster-affected area; S2 indicates that the road is not located in the disaster-affected area; M represents a fixed value.

[0207] Specifically, the A* algorithm starts from the user's real-time location node and selects the node with the smallest heuristic function value each time to expand until it reaches the target refuge point node; during the expansion process, the dynamic weight of the road is updated in real time and the heuristic function value is recalculated;

[0208] The calculation formula of the heuristic function value is as follows:

[0209]

[0210] Where: h(n) represents the heuristic function value; d(m,g) represents the Euclidean distance from node m to target node g; β represents the weight coefficient, which is used to balance the influence of distance and road weight; p(m,g) represents the edge set on the path from node m to target node g; w e Represents the dynamic weight of edge e.

[0211] 5. Matching User Type

[0212] In this embodiment, when multiple reachable paths are obtained through A* algorithm planning: the reachable paths are first divided into several path types; the user's personal physical condition information is obtained; the personal physical condition information is input into a trained route type prediction model, and a predicted path type corresponding to the user's personal physical condition information is output; a reachable path corresponding to the predicted path type is matched from all reachable paths; and the reachable path with the lowest arrival difficulty index is selected from the matched reachable paths as the evacuation guidance path;

[0213] If no accessible path corresponding to the predicted path type is matched, the accessible path with the lowest difficulty index is directly selected as the evacuation guidance path.

[0214] The path types of reachable paths include:

[0215] Fast-track routes: Short, barrier-free, and smooth routes that may include stairs or express lanes. Suitable for users in good health, without mobility impairments, who need to evacuate quickly (e.g., young people and those without chronic diseases).

[0216] Safe escape routes: Avoid dangerous areas (such as fire sources and collapse risk areas), have wide paths, emergency lighting, and safe exit signs. Suitable for users with average or poor health who need to avoid high-risk areas (such as the elderly, pregnant women, and those with chronic diseases);

[0217] Accessible paths: Step-free, gently sloping, with handrails or barrier-free facilities, and a smooth surface; suitable for users with limited mobility (such as those using wheelchairs, crutches, or those with physical disabilities);

[0218] Low-intensity physical trails: The trails are flat and the walking distance is moderate. Avoid continuous uphill or long-distance walking. Suitable for users with weak physical strength or who are easily fatigued (such as the elderly, children, and patients in recovery).

[0219] In this embodiment, the user's personal physical condition information includes basic information (age, gender, height, weight, etc.), health monitoring data (heart rate, blood pressure, blood sugar, sleep quality, exercise steps, etc., collected through wearable devices or health apps) and medical records (including disease history, allergy history, surgical history, etc.).

[0220] Among them, the route type prediction model is built and generated based on machine learning algorithms (such as random forest, support vector machine, neural network, etc.).

[0221] In this embodiment, the training of the route type prediction model is achieved through existing means.

[0222] The present invention divides accessible paths into several types, such as fast-pass paths, safe and evacuation paths, barrier-free paths, and low-intensity paths. By clearly defining and categorizing the characteristics of each type of path, it facilitates subsequent precise matching based on different user needs, making evacuation guidance more detailed and targeted, and avoiding users choosing routes that are inappropriate for their circumstances due to unclear path characteristics. Furthermore, different users have different physical conditions. By obtaining a user's personal physical condition information and inputting it into a trained route type prediction model, a predicted path type that is appropriate for the user's personal physical condition information is obtained. This fully accounts for individual user differences and provides users with personalized evacuation guidance solutions, helping to improve their cooperation and execution efficiency during disasters, thereby improving the efficiency of disaster avoidance and evacuation. Furthermore, the trained route type prediction model, based on a large amount of data and algorithms, can comprehensively consider the relationship between a user's personal physical condition information and different path types, effectively predicting the path type that is appropriate for the user's physical condition, thereby improving the accuracy of route type predictions. In addition, matching the accessible paths corresponding to the predicted path types from all accessible paths can accurately screen out paths that meet the user's physical condition and needs, thereby further improving the rationality of accessible path selection during disaster evacuation.

[0223] 6. Evacuation Guidance

[0224] In this embodiment, based on the evacuation guidance path, augmented reality (AR) and virtual reality (VR) are combined to guide users from their real-time location to the target refuge point.

[0225] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the technical solutions. Those skilled in the art should understand that modifications or equivalent replacements of the technical solutions of the present invention that do not depart from the purpose and scope of the technical solutions of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for identifying disaster-damaged shelters and paths based on satellite remote sensing images, characterized in that: include: S1: Determine candidate shelters based on disaster type and user's real-time location information; S2: Predict the corresponding disaster damage based on the real-time satellite remote sensing images of each candidate shelter and delete unavailable candidate shelters; S3: Match the target refuge with the highest security based on the security-related information of the remaining candidate refuges; S4: Generate an evacuation guidance path from the user's real-time location to the target shelter based on map information and real-time satellite remote sensing images of the area where the target shelter is located; S5: Guide the user from their real-time location to the target shelter through the evacuation guidance path.

2. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 1, wherein: In step S1, preliminary screening shelters are obtained from all shelters according to the disaster type, and then candidate shelters whose distance from the user's real-time location does not exceed a distance threshold are obtained from the preliminary screening shelters.

3. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 1, wherein: In step S2, the processing steps for predicting disaster damage conditions of candidate refuge sites based on real-time satellite remote sensing images include: S201: Obtain satellite remote sensing images of refuge points used as training data and perform preprocessing; S202: Extracting disaster feature vectors of the refuge area based on the pre-processed satellite remote sensing image; and labeling the satellite remote sensing image of the refuge with true labels of the disaster damage. S203: Build a disaster damage prediction model based on a deep learning framework; S204: Inputting the disaster feature vector of the refuge area into the disaster damage prediction model and outputting the corresponding disaster damage prediction label; S205: Calculating a loss function based on the disaster damage prediction label and the corresponding disaster damage true label, and reversely optimizing the parameters of the disaster damage prediction model; S206: Repeat steps S201 to S205 to iteratively train the disaster damage prediction model until the model converges or reaches a preset number of iterations; S207: Extract disaster feature vectors from satellite remote sensing images of candidate refuge sites, input them into a trained disaster damage prediction model, and output corresponding disaster damage prediction labels as their disaster damage conditions.

4. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 3, wherein: In step S201, the steps of pre-processing the satellite remote sensing image of the refuge point include: S2011: Radiometric correction of satellite remote sensing images of refuge sites; The formula is: Where: L x Indicates the radiation value of the corrected satellite remote sensing image; L r Indicates the radiation value of the original satellite remote sensing image; L min and L max Indicates the minimum and maximum values ​​of the original satellite remote sensing image; L t,min and L t,max Indicates the minimum and maximum values ​​of the target radiation value range; S2012: Geometric correction of radiometrically corrected satellite remote sensing images; S2013: Perform image enhancement on the geometrically corrected satellite remote sensing image to obtain a preprocessed satellite remote sensing image.

5. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 3, wherein: In step S202, the process of extracting the disaster feature vector of the refuge area includes: S2021: Use the gray-level co-occurrence matrix to extract the texture features of the refuge area based on the pre-processed satellite remote sensing image; The formula is: Where: F c represents the contrast of the refuge area, i.e., the texture feature; i1 and j1 represent the grayscale values ​​of the pixels in the grayscale co-occurrence matrix of the refuge area; (i1-j1) represents the grayscale difference between two adjacent pixels in the refuge area; p(i1-j1) represents the probability of grayscale values ​​i1 and j1 appearing at the same time; N represents the number of grayscale levels; S2022: Extract the shape features of the refuge area based on the pre-processed satellite remote sensing image; The formula is: Where: F a Represents the area of ​​the refuge point, that is, the shape feature; S i represents the area of ​​the i-th pixel in the refuge area; n represents the number of pixels in the refuge area; S2023: Extract spectral characteristics of refuge area based on pre-processed satellite remote sensing images; The formula is: Where: F r represents the mean reflectivity of the kth band in the refuge area, i.e., the spectral characteristics; R km represents the reflectivity of the i-th pixel in the refuge area in the k-th band; S2023: Concatenate the texture features, shape features, and spectral features of the refuge area to form a disaster feature vector.

6. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 1, wherein: In step S3, the corresponding refuge safety index is calculated based on the safety-related information of each candidate refuge point, and then the refuge point with the highest refuge safety index is selected as the target refuge point.

7. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 6, wherein: In step S3, the steps of calculating the refuge safety index of the candidate refuge point include: S301: Calculate the structural safety score based on the building structure type, construction year and usable area of ​​the candidate refuge point; The formula is: S j =S l ×k n ×k m ; Where: S j Indicates the structural safety score; S l Represents the building structure type score. If the candidate refuge point is an open space, then S l is 100 points. If it is a frame structure, then S l is 90 points. If it is a brick-concrete structure, then S l is 70 points. If it is a civil structure, S l 50 points; k n represents the age correction factor; k m Indicates the available area correction factor; n d Indicates the current year; n j Indicates the construction year of the candidate refuge site; s m Indicates the available area of ​​the candidate refuge site; S302: Calculating a personnel carrying safety score based on the real-time number of refugees and the upper limit of the number of people accommodated at the candidate refuge point; The formula is: Where: S r Indicates the personnel carrying safety score; R r represents the personnel carrying rate of the candidate refuge point; r s Indicates the real-time number of refugees at candidate refuge points; r max Indicates the maximum number of people that can be accommodated at the candidate shelter; S303: Calculate the material reserve adequacy score based on the food reserve, drinking water reserve, and medical supply reserve of the candidate shelter; The formula is: S w =w1·R s +w2·R w +w3·R y ; Where: S w Indicates the material reserve adequacy score; w1, w2, w3 indicate the set weight coefficients; R s 、R w 、R y They represent the food sufficiency rate, drinking water sufficiency rate and medical supplies sufficiency rate of candidate shelters respectively; n s 、n w 、n y They represent the food reserves, drinking water reserves, and medical supplies reserves of candidate shelters respectively; n s,r,t represents the daily food demand per person; t represents the expected number of days of evacuation; n w,r,t The daily drinking water requirement per person; n y,r represents the demand for medical supplies per person; S304: When the concentration of harmful gases at the candidate refuge site is less than the safety threshold, the environmental safety score is 100 points; when the concentration of harmful gases at the candidate refuge site is equal to or greater than the safety threshold, the environmental safety score is 0 points; S305: Perform weighted summation of the structural safety score, the personnel load safety score, the material reserve adequacy score, and the environmental safety score to obtain a refuge safety index for the candidate refuge point.

8. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 1, wherein: In step S4, the processing steps of generating an evacuation guidance route include: S401: registering map information and satellite remote sensing images; S402: Extracting a road network from the registered map information and constructing a road network topology structure including road nodes and edges; S403: extracting real-time road condition information and disaster-affected area information from the registered satellite remote sensing image; S404: assigning a dynamic weight to each edge in the road network topology structure based on real-time road condition information and disaster-affected area information; S405: Path planning is performed based on the user's real-time location information, the location of the target refuge point, the road network topology, and the dynamic weight of the edge through the A* algorithm to obtain an evacuation guidance path from the user's real-time location to the target refuge point.

9. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 8, wherein: In step S404, the process of assigning a dynamic weight to each edge in the road network topology structure includes: S4041: Calculate the basic weight w for each edge based on the road grade and speed limit information base ; The formula is: Where: L represents the road grade; V represents the road speed limit; δ is an adjustment coefficient; S4042: Adjust the basic weight according to the real-time traffic information to obtain the traffic adjustment weight w lk ; The formula is: w lk =w base ×(1+a·r); Where: α is an adjustment coefficient used to control the influence of vehicle density on the weight; ρ represents vehicle density, which is used to measure the congestion of the road; S4043: Adjust the road condition adjustment weight w based on the disaster-affected area information lk , get the dynamic weight w; The formula is: Where: S1 indicates that the road is located in the disaster-affected area; S2 indicates that the road is not located in the disaster-affected area; M represents a fixed value.

10. The method for identifying disaster-damaged shelters and paths based on satellite remote sensing images according to claim 9, wherein: In step S405, the A* algorithm starts from the user's real-time location node and selects the node with the smallest heuristic function value each time to expand until it reaches the target refuge point node; The calculation formula of the heuristic function value is as follows: Where: h(n) represents the heuristic function value; d(m,g) represents the Euclidean distance from node m to target node g; β represents the weight coefficient, which is used to balance the influence of distance and road weight; p(m,g) represents the edge set on the path from node m to target node g; w e Represents the dynamic weight of edge e.

Images