Building disaster assessment method and device, storage medium and electronic equipment

Abstract

The invention provides a building disaster assessment method and device, a storage medium and electronic equipment. According to the method, an electronic device obtains pre-disaster texture features and spectral features of a building area and post-disaster texture features and spectral features; and processing the pre-disaster texture features and the spectral features and the post-disaster texture features and the spectral features through a pre-trained disaster assessment model to obtain the disaster degree of the building area. Therefore, through utilization of the pre-disaster and post-disaster texture and spectral feature changes of the building area, the identification precision of the building disaster degree is improved.

Description

Technical Field

[0001] The present application relates to the field of image recognition, and more specifically, to a method, device, storage medium and electronic device for building disaster assessment. Background Art

[0002] In recent years, earthquakes have occurred frequently around the world, and earthquake disasters with great destructive power have caused huge losses to human life and property. In addition to paying attention to the disaster mechanism of earthquake disasters themselves and conducting pre-disaster monitoring and early warning, it is also very important to conduct post-earthquake disaster assessment and obtain post-earthquake disaster information in a timely and effective manner. The main means of obtaining traditional earthquake damage information is on-site field investigation. Although this method has high credibility in obtaining information, it often has shortcomings such as large workload and low efficiency when carried out in large areas. In view of this, disaster assessment technology based on image recognition has been proposed in related technologies, aiming to improve the efficiency and coverage of disaster assessment through automated image processing methods. These technologies can quickly capture and analyze image information in the earthquake zone, providing a key basis for disaster management decisions.

[0003] The study found that the current disaster assessment methods based on image recognition are mostly evaluated from the perspective of image texture, but in practice, it is found that different types of disasters have different effects on the texture changes of buildings before and after the disaster. This means that not all disasters will cause significant changes in the texture characteristics of buildings. Summary of the invention

[0004] In order to overcome at least one of the deficiencies in the prior art, the present application provides a building disaster assessment method, device, storage medium and electronic device, specifically including:

[0005] In a first aspect, the present application provides a method for building disaster assessment, the method comprising:

[0006] Obtain the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster;

[0007] The pre-disaster texture features, spectral features and post-disaster texture features, spectral features are processed by a pre-trained disaster assessment model to obtain the disaster extent of the building area.

[0008] In combination with the optional implementation manner of the first aspect, the spectral characteristics include spectral characteristics of at least one preset band;

[0009] The spectral characteristics of each preset band include at least one of the mean, standard deviation, maximum difference value and brightness value of the band;

[0010] The spectral feature includes at least one of a HOG feature, a LBP feature, a SSIM feature, and a GLCM feature.

[0011] In combination with an optional implementation manner of the first aspect, the building area is located in an area to be evaluated, and the area to be evaluated includes a plurality of the building areas, and the method further includes:

[0012] Determining the disaster marking color of each building area according to the degree of disaster damage of the building area;

[0013] In the image of the area to be assessed, each of the building areas is covered with a corresponding disaster marking color.

[0014] In combination with an optional implementation manner of the first aspect, the building area is located in the area to be evaluated, and the area to be evaluated includes a plurality of the building areas, and the obtaining of the texture features and spectral features of the building area after the disaster includes:

[0015] Determining the location of each of the building areas based on the pre-disaster image of the area to be assessed;

[0016] Determining a post-disaster image of each of the building areas from the post-disaster image of the area to be assessed according to the position of each of the building areas;

[0017] According to the post-disaster image of each of the building areas, the post-disaster texture features of each of the building areas are obtained;

[0018] Determining a post-disaster spectral image of each of the building areas from the post-disaster spectral image of the area to be assessed according to the position of each of the building areas;

[0019] According to the post-disaster spectral image of each of the building areas, the post-disaster spectral characteristics of each of the building areas are obtained.

[0020] In conjunction with the optional implementation manner of the first aspect, determining the location of each of the building areas according to the pre-disaster image of the area to be assessed includes:

[0021] Performing object-level segmentation on the pre-disaster image of the area to be evaluated to obtain an object segmentation result of the pre-disaster image of the area to be evaluated;

[0022] Performing pixel-level segmentation on the pre-disaster image of the area to be evaluated to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated;

[0023] The object segmentation result is fused with the pixel segmentation result to obtain the position of each building area.

[0024] In combination with an optional implementation manner of the first aspect, performing pixel-level segmentation on the pre-disaster image of the area to be evaluated to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated includes:

[0025] The pre-disaster image of the area to be evaluated is processed by a U-Net derivative model to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated, wherein the feature map input to each decoding layer in the U-Net derivative model is a fusion feature map between a decoding feature map output by an upper-level decoding layer and an encoding feature map output by an encoding layer at the same level, and the decoding feature map and the encoding feature map have the same size;

[0026] Each encoding layer of the U-Net derived model downsamples the input feature map through convolution kernels of different sizes and fuses them into an encoded feature map;

[0027] Each decoding layer of the U-Net derived model upsamples the input features through convolution kernels of different sizes and fuses them into a decoding feature map.

[0028] In combination with the optional implementation manner of the first aspect, performing object-level segmentation on the pre-disaster image of the area to be evaluated to obtain an object segmentation result of the pre-disaster image of the area to be evaluated includes:

[0029] The pre-disaster image of the area to be evaluated is processed by the SAM model to obtain an object segmentation result of the pre-disaster image of the area to be evaluated.

[0030] In a second aspect, the present application further provides a building disaster assessment device, the device comprising:

[0031] A feature extraction module is used to obtain the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster;

[0032] The disaster assessment module is used to process the pre-disaster texture features, spectral features and post-disaster texture features, spectral features through a pre-trained disaster assessment model to obtain the disaster extent of the building area.

[0033] In a third aspect, the present application further provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the building disaster assessment method is implemented.

[0034] In a fourth aspect, the present application further provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the building disaster assessment method is implemented.

[0035] Compared with the prior art, this application has the following beneficial effects:

[0036] The present application provides a method, device, storage medium and electronic device for building disaster assessment. In the method, the electronic device obtains the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster; the pre-disaster texture features and spectral features and the post-disaster texture features and spectral features are processed by a pre-trained disaster assessment model to obtain the disaster degree of the building area. In this way, by utilizing the changes in the texture and spectral features of the building area before and after the disaster, the recognition accuracy of the building disaster degree is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A flow chart of a building disaster assessment method provided in an embodiment of the present application;

[0039] Figure 2 A structural diagram of the U-Net derivative model provided in an embodiment of the present application;

[0040] Figure 3 A disaster marking effect diagram of the area to be assessed provided in an embodiment of the present application;

[0041] Figure 4 A schematic diagram of the structure of a building disaster assessment device provided in an embodiment of the present application;

[0042] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0044] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0045] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0046] In the description of the present application, it should be noted that the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. In addition, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0047] Based on the above statement, as introduced in the background technology, it is found in practice that different types of disasters have different effects on the texture changes of buildings before and after the disaster. This means that not all disaster situations will cause significant changes in the texture characteristics of buildings.

[0048] For example, when a building suffers internal structural damage, such as hidden cracks or reduced material strength, its appearance may remain unchanged, but in spectral images, these damages may cause changes in reflectivity or thermal radiation characteristics. Similarly, if a disaster causes changes in the chemical composition of building materials, such as smoke or chemical leaks after a fire, these changes may not be immediately apparent in visual images, but will be reflected in differences in spectral characteristics. In addition, small deformations or local damage to buildings may not be easily detected in ordinary remote sensing images, but these subtle changes can be revealed through spectral analysis. For example, underground structural changes caused by earthquakes may affect the spectral characteristics of surface buildings, and such changes are not easy to identify in optical remote sensing images.

[0049] Based on the discovery of the above technical problems, the inventors have proposed the following technical solutions to solve or improve the above problems through creative work. It should be noted that the defects in the solutions in the above prior art are the results obtained by the inventors after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed in the embodiments of the present application for the above problems below should all be the contributions made by the inventors to the present application in the process of invention and creation, and should not be understood as technical contents known to those skilled in the art.

[0050] In view of this, this embodiment provides a method for assessing building disasters. In the method, an electronic device obtains the texture features and spectral features of a building area before a disaster and the texture features and spectral features after a disaster; the texture features and spectral features before a disaster and the texture features and spectral features after a disaster are processed by a pre-trained disaster assessment model to obtain the degree of damage to the building area. In this way, by utilizing the changes in the texture and spectral features of the building area before and after a disaster, the recognition accuracy of the degree of building disasters is improved.

[0051] It should be noted that the electronic device for implementing the above-mentioned building disaster assessment method may be, but is not limited to, a mobile terminal, a tablet computer, a laptop computer, a desktop computer, and a server, etc. Among them, the server may be a single server or a server group. The server group may be centralized or distributed (for example, the server may be a distributed system). In some embodiments, the server may be local or remote relative to the user terminal. In some embodiments, the server may be implemented on a cloud platform; as an example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud (Community Cloud), a distributed cloud, an inter-cloud (Inter-Cloud), a multi-cloud (Multi-Cloud), etc., or any combination thereof. In some embodiments, the server may be implemented on an electronic device having one or more components.

[0052] To make the solution provided by this embodiment clearer, it is assumed that a server is used as an electronic device to implement the method, and combined with Figure 1 Each step of the method is described in detail. However, it should be understood that the operations of the flowchart may not be implemented in order, and steps without logical contextual relationships may be reversed in order or implemented simultaneously. In addition, those skilled in the art may add one or more other operations to the flowchart, or remove one or more operations from the flowchart under the guidance of the content of this application. Figure 1 As shown, the method includes:

[0053] S1, obtaining the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster.

[0054] In this embodiment, the spectral characteristics include spectral characteristics of at least one preset band; the spectral characteristics of each band include at least one of the mean, standard deviation, maximum difference value and brightness value of the band. It should be noted that the selection of the preset band is related to the type of disaster. For example, there may be some differences in the bands selected for earthquake disasters and flood disasters.

[0055] The spectral feature includes at least one of HOG feature, LBP feature, SSIM feature and GLCM feature. The above four features are described in detail below.

[0056] HOG (Histogram of Oriented Gradients) features:

[0057] HOG feature is a feature used for image object detection, a descriptor used in image processing and computer vision, and is particularly suitable for target detection and image classification tasks. Specifically, the features of the local image area are represented by calculating the gradient histogram in each direction in the image. When calculating the HOG feature, the image is divided into windows of several directions and sizes, the histogram of the gradient direction is calculated for each window, and then these histograms are combined to form a vector. This feature has a good effect in edge detection and object recognition.

[0058] LBP (Local Binary Pattern) features:

[0059] LBP feature is a feature that describes the local texture of an image. It compares each pixel in the image with its neighboring pixels. If the gray value of the current pixel is greater than that of its neighbor, it is marked as 1, otherwise it is marked as 0. The result of this comparison forms a binary number, namely the LBP value. LBP feature can effectively capture the texture information of the image and is widely used in face recognition, object recognition and other fields.

[0060] SSIM (Structural Similarity Index) features:

[0061] The SSIM feature is an indicator for measuring image similarity, and is particularly suitable for comparing the structural similarity of images. This feature takes into account three aspects: image brightness, contrast, and structure, and can more accurately evaluate the similarity of two images. SSIM has important applications in image processing, video quality assessment, image matching and fusion, etc.

[0062] GLCM (Gray Level Co-occurrence Matrix) features:

[0063] GLCM features are features extracted based on the gray-level co-occurrence matrix. The gray-level co-occurrence matrix (GLCM) is constructed to describe the co-occurrence relationship of pixels with different gray values ​​in different directions and distances in the image. Therefore, the GLCM feature can capture the spatial dependence of the image, such as the directionality and contrast of the texture. By analyzing the GLCM, multiple texture features can be extracted, such as energy, entropy, contrast, etc.

[0064] Based on the introduction of each feature in the above embodiment, it should be noted that the buildings in the post-disaster image will be integrated with the surrounding environment due to collapse, making it impossible to distinguish the area where the buildings are located from the post-disaster image. It should also be noted that the building area is located in the area to be evaluated, and the area to be evaluated includes multiple building areas. Therefore, this embodiment provides the following optional implementation methods of step S1:

[0065] S1-1, determine the location of each building area based on the pre-disaster image of the area to be assessed.

[0066] It should be understood that the pre-disaster image and the post-disaster image have the same size. Therefore, the post-disaster image of each building area can be determined from the post-disaster image by using the position of each building area obtained from the pre-disaster image. In addition, it is found in practice that since the pre-disaster image in this embodiment is a remote sensing image, the building appears particularly small in the image, and a single image segmentation method is difficult to achieve a better segmentation effect. Therefore, this embodiment improves the segmentation accuracy by fusing the segmentation results of multiple segmentation models.

[0067] In an optional implementation manner, the server performs object-level segmentation on the pre-disaster images of the area to be evaluated to obtain object segmentation results of the pre-disaster images of the area to be evaluated; performs pixel-level segmentation on the pre-disaster images of the area to be evaluated to obtain pixel segmentation results of the pre-disaster images of the area to be evaluated; and fuses the object segmentation results with the pixel segmentation results to obtain the location of each building area.

[0068] The server processes the pre-disaster images of the area to be evaluated through the SAM model to obtain object segmentation results of the pre-disaster images of the area to be evaluated.

[0069] It should be understood that SAM, as an advanced deep learning image segmentation model, has excellent performance in the classification and segmentation tasks of complex and diverse images. SAM only needs to be fine-tuned for a small number of use cases for specific purposes to show high applicability in image segmentation tasks. The SAM model structure mainly includes three components: Image Encoder, Promp Encoder, and Masked Autoencoder.

[0070] The image encoder is a Vision Transformer (ViT) network pre-trained with MAE (Masked Autoencoder) to generate an Image Embedding representation for the image to be segmented.

[0071] The prompt encoder is used to embed prompts such as points, boxes, and text.

[0072] The mask decoder adopts the design of Transformer decoder and modifies it to efficiently map image embedding, hint embedding and output token to masks and calculate the effective probability of target object mask.

[0073] Therefore, the above component functions are used to implement the segmentation of the input image and obtain a higher segmentation result. SAM has two segmentation modes: interactive segmentation and fully automatic segmentation. Interactive segmentation is performed by the user by inputting information such as borders, points or text, and only the specified area is segmented. Therefore, this mode is not suitable for remote sensing image object segmentation tasks. Fully automatic segmentation segments all objects in the entire image by setting segmentation parameters.

[0074] In addition, the pre-disaster images of the area to be evaluated are processed by the U-Net derivative model to obtain the pixel segmentation results of the pre-disaster images of the area to be evaluated, wherein the feature map input to each decoding layer in the U-Net derivative model is a fusion feature map between the decoding feature map output by the upper decoding layer and the encoding feature map output by the same-level encoding layer, and the decoding feature map and the encoding feature map have the same size;

[0075] Each encoding layer of the U-Net derivative model downsamples the input feature map through convolution kernels of different sizes and fuses them into the encoded feature map;

[0076] Each decoding layer of the U-Net derivative model upsamples the input features through convolution kernels of different sizes and fuses them into a decoding feature map.

[0077] It should be noted that the convolution kernels in conventional U-Net models all use the "valid" mode, which makes the sizes of the corresponding feature maps in the decoding path and the encoding path different. The feature map in the encoding path needs to be cropped to the same size as the feature map in the decoding path and then fused. Finally, the output image size of the model is smaller than the input image. Therefore, there is a problem of feature loss and inability to accurately locate each pixel in the input image.

[0078] like Figure 2 As shown in the figure, in the U-Net derivative model, the original main framework of the conventional U-Net model is maintained, but unlike the conventional U-Net model, the operation of each convolution kernel in the U-Net derivative model uses the "same" method, so that the feature map corresponding to the decoding path and the encoding path in the model are the same size and can be directly fused. The output result of the obtained network has the same size as the input, which can achieve accurate positioning of the input image pixel by pixel.

[0079] In addition, see Figure 2, convolutional neural networks are widely used in the field of image processing because of their receptive field characteristics. Among them, the receptive field is the size of the area where a feature is mapped to the input space in a convolutional neural network. Different convolution kernel sizes can extract different receptive field features, and the range of the receptive field expands with the increase in the number of convolution layers. For example, the receptive field after two layers of 3×3 convolution operations is 5×5, and the receptive field after three layers of 3×3 convolution operations is 7×7. The convolution kernels used in the convolution operations in the conventional U-Net model are all 3×3. In contrast, the U-Net derivative model splices 3×3 and 5×5 convolution operations together. In this way, on the one hand, the width of the network can be increased, and on the other hand, it can extract the features of multi-scale receptive fields, fully explore the feature relationship of the neighborhood space of the pixel, and extract relevant information.

[0080] Since the U-net derivative model uses 3×3 and 5×5 convolution kernels on the same feature map, the 3×3 convolution kernel has 9 parameters, while the 5×5 convolution kernel has 25 parameters. The number of parameters is higher than that of U-Net, so it takes longer to train the network, and overfitting should be avoided during model training. In addition, the stochastic gradient descent method SGD is used as a parameter optimization algorithm in the conventional U-Net model, but it is found in practice that it is difficult for SGD to select a suitable hyperparameter learning rate. It requires continuous manual attempts to select, and the parameters are updated using the same learning rate, which is easy to converge to the local optimum. Therefore, during the training of the U-Net derivative model, the optimization algorithm used is Adam. In addition to having a small memory requirement, Adam can calculate the learning rate according to different parameters as the parameters are updated during the model training process, so that there is no need for manual intervention to find a suitable learning rate.

[0081] In this way, the object segmentation result is obtained by the SAM model, and the pixel segmentation result is obtained by the U-Net derivative model; then, the segmentation results of the two are fused. Specifically, the geometric features of each segmented object in the object segmentation result are extracted, such as length, convexity, firmness, roundness, shape factor, and rectangle fit. Among them, convexity represents the convexity of the object shape, which is defined as the ratio of the area of ​​the object's convex hull (the smallest convex polygon that can contain all the points of the object) to the actual area of ​​the object; firmness is the ratio of the actual area of ​​the object to the area of ​​its convex hull; roundness describes the degree to which the object shape is close to a circle; the shape factor is a measure of the complexity of the object shape, usually defined as the ratio of the object area to the square of the perimeter; rectangle fit is a measure of the degree of similarity between an object and a rectangle.

[0082] The key to fusing pixel segmentation results with pixel segmentation results is to set thresholds for the area of ​​the segmented object, the aspect ratio of the minimum circumscribed rectangle, and the probability that each pixel in the pixel segmentation result is a building. By setting this set of thresholds, the system can screen out high-confidence candidate building objects. In specific implementation, it is possible to determine whether to confirm that the object is a building by comparing the characteristics of the segmented object (such as the aspect ratio) and the probability that the pixel of the segmented object is a building, combined with the area size of the segmented object, and then obtain the location of each building area. In this way, the fusion process not only takes into account the information at the object level, but also includes the fine segmentation results at the pixel level, thereby achieving more accurate recognition of buildings.

[0083] Based on the position of each building area obtained in the above embodiment, step S1 further includes:

[0084] S1-2, determining a post-disaster image of each building area from the post-disaster images of the area to be evaluated according to the position of each building area.

[0085] S1-3, obtaining the post-disaster texture features of each building area based on the post-disaster images of each building area.

[0086] S1-4, determining a post-disaster spectral image of each building area from the post-disaster spectral image of the area to be assessed according to the location of each building area.

[0087] S1-5, obtaining the spectral characteristics of each building area after the disaster based on the spectral image of each building area after the disaster.

[0088] Similarly, the position of the building area is used to obtain the texture features of the building area before the disaster from the pre-disaster image; and the spectral features before the disaster are obtained from the pre-disaster spectral image.

[0089] Continue to see Figure 1 , the method provided in this embodiment also includes:

[0090] S2, the pre-disaster texture features, spectral features and post-disaster texture features, spectral features are processed through the pre-trained disaster assessment model to obtain the disaster extent of the building area.

[0091] For the above disaster assessment model, it was found in practice that the parameters of the network model are mainly concentrated in the convolution kernel in the convolution layer and the connection weights between the fully connected layer, where the convolution layer extracts the features of the image and the fully connected layer is used for feature integration and classification. Increasing the number of layers of the convolutional neural network helps to improve the feature extraction ability of the model, but when the network is deepened, the parameters of the convolution kernel increase, and a large amount of data is required to complete the training. It is difficult to achieve the global optimality of the parameters under small data sets. Considering that the sample data of building damage assessment in this experiment is small, the convolution layer of the convolutional neural network is set to 2, which can avoid overfitting while learning the deeper features of the pre-disaster and post-disaster images, and obtain a disaster assessment model for the degree of building damage.

[0092] In addition, in this embodiment, since the building area is located in the area to be evaluated, the area to be evaluated includes multiple building areas. For example, in order to intuitively display the disaster situation of the area to be evaluated, the server also determines the disaster marking color of the building area according to the degree of disaster of each building area; and covers each building area with the corresponding disaster marking color in the image of the area to be evaluated.

[0093] For example, the degree of damage includes undamaged, slightly damaged, severely damaged and completely damaged, and is marked with different colors. Figure 3 As shown, in the image of the area to be assessed, the building areas in the image are covered with colors that match the degree of damage, so that users can intuitively feel which areas in the area to be assessed are more severely affected by the disaster and which areas are less severely affected by the disaster.

[0094] Based on the same inventive concept as the building disaster assessment method provided in this embodiment, this embodiment also provides a building disaster assessment device, which includes at least one software function module that can be stored in a memory or fixed in an electronic device in the form of software. The processor in the electronic device is used to execute the executable module stored in the memory. For example, the software function module and computer program included in the building disaster assessment device. Please refer to Figure 4 , functionally speaking, the device may include:

[0095] A feature extraction module 11 is used to obtain the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster;

[0096] The disaster assessment module 12 is used to process the pre-disaster texture features, spectral features and post-disaster texture features, spectral features through a pre-trained disaster assessment model to obtain the disaster extent of the building area.

[0097] In this embodiment, the feature extraction module 11 is used to implement Figure 1 In step S1, the disaster assessment module 12 is used to implement Figure 1Therefore, for the detailed description of each of the above modules, please refer to the specific implementation of the corresponding step. In addition, in view of the same inventive concept as the above-mentioned building disaster assessment method, the building disaster assessment device can also implement other steps or sub-steps of the method through the above modules or other modules, which will not be described in detail in this embodiment.

[0098] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0099] It should also be understood that if the above implementation is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application.

[0100] Therefore, this embodiment further provides a storage medium, which is a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by a processor, the building disaster assessment method provided in this embodiment is implemented. Among them, the storage medium can be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc., which can store program codes.

[0101] This embodiment provides an electronic device for implementing a building disaster assessment method. Figure 5 As shown, the electronic device may include a processor 22 and a memory 21. In addition, the memory 21 stores a computer program, and the processor implements the building disaster assessment method provided in this embodiment by reading and executing the computer program corresponding to the above implementation in the memory 21.

[0102] Continue to see Figure 5 The electronic device further includes a communication unit 23. The memory 21, the processor 22 and the communication unit 23 are electrically connected to each other directly or indirectly through a system bus 24 to achieve data transmission or interaction.

[0103] The memory 21 may be an information recording device based on any electronic, magnetic, optical or other physical principle, used to record execution instructions, data, etc. In some embodiments, the memory 21 may be, but is not limited to, a volatile memory, a non-volatile memory, a storage drive, etc.

[0104] In some embodiments, the volatile memory may be a random access memory (RAM); in some embodiments, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), a flash memory, etc.; in some embodiments, the storage drive may be a disk drive, a solid-state drive, any type of storage disk (such as a CD, a DVD, etc.), or a similar storage medium, or a combination thereof, etc.

[0105] The communication unit 23 is used to send and receive data through a network. In some embodiments, the network may include a wired network, a wireless network, a fiber optic network, a telecommunication network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a wide area network (WAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, or a near field communication (NFC) network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include a wired or wireless network access point, such as a base station and / or a network switching node, through which one or more components of the service request processing system may be connected to the network to exchange data and / or information.

[0106] The processor 22 may be an integrated circuit chip having a signal processing capability, and the processor may include one or more processing cores (e.g., a single-core processor or a multi-core processor). By way of example only, the processor may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), or a microprocessor, or any combination thereof.

[0107] Understandably, Figure 5 The structure shown is for illustration only. The electronic device may also have Figure 5 More or fewer components than shown, or with Figure 5 Different configurations are shown. Figure 5 The components shown may be implemented in hardware, software or a combination thereof.

[0108] It should be understood that the apparatus and method disclosed in the above-mentioned embodiments can also be implemented in other ways. The apparatus embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the apparatus, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0109] The above are only various implementations of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A building disaster assessment method, characterized in that: The method comprises: Obtain the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster; The pre-disaster texture features, spectral features and post-disaster texture features, spectral features are processed by a pre-trained disaster assessment model to obtain the disaster extent of the building area.

2. The building disaster assessment method according to claim 1, characterized in that: The spectral characteristics include spectral characteristics of at least one preset band; The spectral characteristics of each preset band include at least one of the mean, standard deviation, maximum difference value and brightness value of the band; The spectral feature includes at least one of a HOG feature, a LBP feature, a SSIM feature, and a GLCM feature.

3. The building disaster assessment method according to claim 1, characterized in that: The building area is located in an area to be evaluated, and the area to be evaluated includes a plurality of the building areas. The method further includes: Determining the disaster marking color of each building area according to the degree of disaster suffered by the building area; In the image of the area to be assessed, each of the building areas is covered with a corresponding disaster marking color.

4. The building disaster assessment method according to claim 1, characterized in that: The building area is located in the area to be evaluated, and the area to be evaluated includes multiple building areas. The obtaining of texture features and spectral features of the building area after the disaster includes: Determining the location of each of the building areas based on the pre-disaster image of the area to be assessed; Determining a post-disaster image of each of the building areas from the post-disaster image of the area to be assessed according to the position of each of the building areas; According to the post-disaster image of each of the building areas, the post-disaster texture features of each of the building areas are obtained; Determining a post-disaster spectral image of each of the building areas from the post-disaster spectral image of the area to be assessed according to the position of each of the building areas; According to the post-disaster spectral image of each of the building areas, the post-disaster spectral characteristics of each of the building areas are obtained.

5. The building disaster assessment method according to claim 4, characterized in that: Determining the location of each building area according to the pre-disaster image of the area to be assessed includes: Performing object-level segmentation on the pre-disaster image of the area to be evaluated to obtain an object segmentation result of the pre-disaster image of the area to be evaluated; Performing pixel-level segmentation on the pre-disaster image of the area to be evaluated to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated; The object segmentation result is fused with the pixel segmentation result to obtain the position of each building area.

6. The building disaster assessment method according to claim 5, characterized in that: The pre-disaster image of the area to be evaluated is segmented at the pixel level to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated, including: The pre-disaster image of the area to be evaluated is processed by a U-Net derivative model to obtain a pixel segmentation result of the pre-disaster image of the area to be evaluated, wherein the feature map input to each decoding layer in the U-Net derivative model is a fusion feature map between a decoding feature map output by an upper-level decoding layer and an encoding feature map output by an encoding layer at the same level, and the decoding feature map and the encoding feature map have the same size; Each encoding layer of the U-Net derived model downsamples the input feature map through convolution kernels of different sizes and fuses them into an encoded feature map; Each decoding layer of the U-Net derived model upsamples the input features through convolution kernels of different sizes and fuses them into a decoding feature map.

7. The building disaster assessment method according to claim 5, characterized in that: The performing object-level segmentation on the pre-disaster image of the area to be evaluated to obtain the object segmentation result of the pre-disaster image of the area to be evaluated includes: The pre-disaster image of the area to be evaluated is processed by the SAM model to obtain an object segmentation result of the pre-disaster image of the area to be evaluated.

8. A building disaster assessment device, characterized in that: The device comprises: A feature extraction module is used to obtain the texture features and spectral features of the building area before the disaster and the texture features and spectral features after the disaster; The disaster assessment module is used to process the pre-disaster texture features, spectral features and post-disaster texture features, spectral features through a pre-trained disaster assessment model to obtain the disaster extent of the building area.

9. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the building disaster assessment method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the building disaster assessment method according to any one of claims 1 to 7 is implemented.

Images