Urban change monitoring analysis early warning method and system based on remote sensing surface feature data

Abstract

The embodiment of the invention provides a city change monitoring analysis early warning method and system based on remote sensing ground feature data. The method is applied to the technical field of remote-sensing image change monitoring. The method comprises the steps of obtaining front and back time sequence images of urban ground features; inputting the front and back time sequence images of the urban ground features into a pre-trained urban ground feature change monitoring model to obtain an urban ground feature change result; wherein the urban surface feature change monitoring model is constructed based on a cGAN network model, and comprises a generator and a discriminator; the urban surface feature change monitoring model is trained through the following steps: acquiring front and back time sequence images of urban surface features; marking the change areas of the time sequence images before and after the urban surface features to obtain corresponding change label maps, and taking the time sequence images before and after the urban surface features and the corresponding change label maps as training samples; and alternately training a generator and a discriminator by using the training sample to obtain an urban ground feature change monitoring model. Therefore, monitoring and early warning of urban ground feature changes are realized.

Description

Technical Field

[0001] The present disclosure relates to the field of remote sensing image change monitoring, and in particular to a method and system for monitoring, analyzing and warning urban changes based on remote sensing object classification. Background Art

[0002] With the acceleration of urbanization, the changes in urban features (such as buildings, vegetation, roads, water bodies, etc.) are becoming increasingly significant, and these changes have had a profound impact on urban planning, resource management, and environmental protection. Therefore, accurate and timely monitoring and analysis of changes in urban features are of great significance to the sustainable development of cities. Remote sensing technology, as a non-contact, large-scale, and rapid technical means of acquiring surface information, plays an increasingly important role in urban change monitoring. In recent years, with the continuous advancement of satellite remote sensing technology, key indicators such as resolution, spectral range, and temporal resolution of remote sensing images have been significantly improved. This makes remote sensing technology show great potential in urban feature classification, change monitoring, and quantitative analysis. At the same time, the rapid development of artificial intelligence technologies such as computer vision, machine learning, and deep learning has also provided strong support for the automated processing and information extraction of remote sensing data.

[0003] The method of urban change monitoring, analysis and early warning based on remote sensing land feature classification came into being against the background of the rapid development of remote sensing technology and artificial intelligence technology. This method has the advantages of rich data sources, high classification accuracy, strong real-time monitoring, accurate quantitative analysis and effective early warning mechanism. It is of great significance to promote the sustainable development of cities.

[0004] Among the methods of urban change monitoring, analysis and early warning, traditional on-site surveys are often time-consuming and labor-intensive, and can no longer meet the needs of rapid changes and efficient management; existing remote sensing recognition technology only supports target recognition of high-resolution remote sensing images, lacks the association of different periods of the same target, and does not have a means of automatically monitoring changes in the same area or target over a long period of time; deep convolutional neural networks (CNNs) perform feature learning and target recognition on remote sensing images. Based on the CNN algorithm, end-to-end target detection and recognition can be achieved, and important land features in the city can be extracted, including buildings, water bodies, vegetation, roads and other land features; classified change monitoring algorithms are used to compare changes in land features and accurately perceive changes in urban land features. CNN can learn local features of the data, but in generation tasks, relying solely on CNN may not be able to capture global dependencies and complex data distributions. Summary of the invention

[0005] The present invention provides a method and system for monitoring, analyzing and warning of urban changes based on remote sensing object data, which solves the technical problems of monitoring, analyzing and warning of urban object changes.

[0006] According to a first aspect of the present disclosure, a method for monitoring, analyzing and warning urban changes based on remote sensing ground object data is provided. The method comprises: Obtaining before and after time series images of urban features; The before and after time series images of urban features are input into the pre-trained urban feature change monitoring model to obtain the urban feature change results; The pre-trained urban feature change monitoring model is built based on the cGAN model, including: Generator and Discriminator; The pre-trained urban feature change monitoring model is trained through the following steps: Obtaining before and after time series images of urban features; The changed areas of the before and after time series images of urban features are marked to obtain the corresponding change label map, and the before and after time series images of urban features and the corresponding change label map are used as training samples; The generator and discriminator are trained alternately using training samples to obtain a model for monitoring urban land feature changes.

[0007] As in the above aspects and any possible implementation, an implementation is further provided, wherein the generator includes a U-Net network and a BiSeNet network; wherein, The U-Net network consists of an encoder and a decoder and introduces skip connections; The BiSeNet network introduces a feature fusion module and an attention refinement module.

[0008] As in the above aspects and any possible implementation, an implementation is further provided, wherein the discriminator includes an LcGAN discriminator and a PatchGAN discriminator; wherein, The LcGAN discriminator evaluates the image globally in combination with the object category labels; The PatchGAN discriminator evaluates local patches of pixels in an image.

[0009] As in the above aspects and any possible implementation, an implementation is further provided, where the overall objective function of the urban feature change monitoring model is: , in, is the objective function of the cGAN model, is the objective function for a pixel pair, is the weight coefficient; where Said Includes a space penalty term.

[0010] As in the above aspects and any possible implementation, an implementation is further provided, where the objective function of the cGAN model includes a gradient penalty term to further stabilize the training process and prevent mode collapse.

[0011] As in the above aspects and any possible implementation, an implementation is further provided, wherein alternately training the generator and the discriminator using the training sample set includes: First fix the generator and train the discriminator, then fix the discriminator and train the generator.

[0012] As in the above aspects and any possible implementation manner, an implementation manner is further provided, and the method further includes: Interpolation analysis algorithm is used to monitor and warn of changes in urban features.

[0013] According to a second aspect of the present disclosure, a city change monitoring, analysis and early warning system based on remote sensing ground object data is provided. The system comprises: Acquisition module, used to obtain before and after time series images of urban features; The monitoring and early warning module is used to input the before and after time series images of urban features into the pre-trained urban feature change monitoring model to obtain the urban feature change results, wherein the pre-trained urban feature change monitoring model is constructed based on the cGAN model, including: a generator and a discriminator; the pre-trained urban feature change monitoring model is trained through the following steps: Obtaining before and after time series images of urban features; The changed areas of the before and after time series images of urban features are marked to obtain the corresponding change label map, and the before and after time series images of urban features and the corresponding change label map are used as training samples; The generator and discriminator are trained alternately using training samples to obtain a model for monitoring urban land feature changes.

[0014] According to a third aspect of the present disclosure, an electronic device is provided, which includes a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the program, the method described above is implemented.

[0015] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method according to the first aspect and / or the second aspect of the present disclosure is implemented.

[0016] Based on the cGAN model, a model for monitoring urban landform changes is constructed. The acquired time-series images of urban landforms are input into the model to obtain the results of urban landform changes. The interpolation analysis algorithm is used for the results of urban landform changes to monitor and warn of urban landform changes. The present disclosure realizes the monitoring and warning of urban landform changes by constructing a model for monitoring urban landform changes.

[0017] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein: Figure 1 A flow chart of a method for monitoring, analyzing and warning urban changes based on remote sensing object data according to an embodiment of the present disclosure is shown; Figure 2 The figure shows a structure diagram of a generator and a discriminator of an urban feature change monitoring model according to an embodiment of the present disclosure; Figure 3 A block diagram of a city change monitoring, analysis and early warning system based on remote sensing object data according to an embodiment of the present disclosure is shown; Figure 4 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0020] In addition, the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0021] In the present invention, a city feature change monitoring model is constructed based on the cGAN model. By continuously training the city feature change monitoring model, the monitoring accuracy is improved. The acquired before and after time series images of the city features are input into the pre-trained city feature change monitoring model to obtain the city feature change results. The interpolation analysis algorithm is used for the city feature change results, thereby realizing the monitoring and early warning of urban changes.

[0022] Figure 1 FIG. 1 is a flow chart of a method 100 for monitoring, analyzing and warning urban changes based on remote sensing object data according to an embodiment of the present disclosure. Figure 1 As shown, a method for monitoring, analyzing and warning urban changes based on remote sensing ground object data may include the following steps: S101, obtaining before and after time series images of urban features.

[0023] In some embodiments, high-resolution images of land objects with timestamps are collected to ensure that the data set contains images taken in different seasons and weather conditions. Data sources may include satellite images, images taken by drones, and images taken on the ground, covering various types of land objects such as urban buildings, farmlands, water bodies, forests, etc.

[0024] In some embodiments, after acquiring the before and after time-series images of the urban features, a pre-processing operation is performed on the before and after time-series images of the urban features.

[0025] In some embodiments, the preprocessing may be a noise reduction process. For example, due to sensor factors, periodic noise may appear in some acquired remote sensing images, and Fourier transform may be used for filtering.

[0026] In some embodiments, preprocessing may be thin cloud and shadow processing. For example, due to weather reasons, thin clouds appearing in some remote sensing images are weakened; due to the solar altitude angle, some images may have mountain shadows, which can be eliminated using the ratio method.

[0027] In some embodiments, the preprocessing may be radiation calibration, for example, to eliminate the error of the sensor itself and determine the accurate radiation value at the sensor entrance, the pixel brightness value (DN) of the recorded original remote sensing image is converted into the reflectivity of the outer surface of the atmosphere (or called the radiation brightness value); Furthermore, the formula used for radiation calibration is as follows: , in, It is the spectral radiance of a certain band; is the incremental correction factor, To correct the deviation, It is the pixel brightness value of the original remote sensing image.

[0028] In some embodiments, the preprocessing can be atmospheric correction, rough geometric correction and orthorectification. Atmospheric correction can eliminate the influence of substances such as water vapor, oxygen, carbon dioxide, methane and ozone in the atmosphere on the reflection of ground objects, and eliminate the influence of atmospheric molecules and aerosol scattering; rough geometric correction is to correct the geometric distortion of the original remote sensing image; orthorectification is to use existing geographic reference data (original remote sensing images, topographic maps and control points, etc.) and digital elevation model data to correct the original remote sensing image, so as to eliminate or reduce the image deformation caused by terrain undulations, so that the original remote sensing image has accurate ground coordinates and projection information.

[0029] In some embodiments, the preprocessing can be matching fusion, which is to match and fuse the panchromatic bands and multispectral bands of different satellites to form an image. The image retains the high-resolution information of the panchromatic data and integrates the characteristics of the multispectral data. The resulting image has both high spatial resolution and texture features, and rich spectral information, thereby achieving rich image information, good visual effects and high quality.

[0030] In some embodiments, preprocessing can be image enhancement, which can make the ground object information contained in the remote sensing image more readable and the target of interest more prominent; image enhancement can also classify the grayscale image according to the grayscale value of the pixel, and then assign different colors to the classification, so that the original grayscale image becomes a pseudo-color image, thereby achieving the purpose of image enhancement; two or more single-band images can also be spatially aligned and arithmetic operations can be performed to achieve image enhancement.

[0031] In some embodiments, the preprocessing may be image mosaicking, which is to first perform geometric correction on each image, plan them into a unified coordinate system, then crop them, remove the overlapping parts, and then assemble the cropped images to form an image of a certain range; after the image mosaicking, the image is processed for color uniformity to eliminate the inconsistency in tones between different images.

[0032] In some embodiments, the preprocessing may be image cropping, where the mosaicked image is cropped according to the boundary range to obtain a remote sensing image within the required range; For example, we first obtain remote sensing images of different objects in a certain area in Qingdao, and then perform preprocessing operations such as image denoising, image enhancement, geometric correction, image mosaicking and image cropping on these remote sensing images, and finally obtain the remote sensing images we need.

[0033] S102, inputting the before and after time series images of the urban features into a pre-trained urban feature change monitoring model to obtain the urban feature change result.

[0034] In some embodiments, the urban feature change monitoring model is constructed based on the cGAN model, including a generator and a discriminator.

[0035] In some embodiments, the urban feature change monitoring model is trained by the following steps: Obtaining before and after time series images of urban features; The changed areas of the before and after time series images of urban features are marked to obtain the corresponding change label map, and the before and after time series images of urban features and the corresponding change label map are used as training samples.

[0036] In some embodiments, the above-mentioned preprocessing operation is applied to the acquired before and after time series images of urban features to enhance the diversity of data, thereby improving the generalization ability and robustness of the model; Specifically, considering the limitation of CPU memory capacity, an image segmentation algorithm was used to cut each urban feature time-series image into regional images with a pixel size of 256x256, and normalized them. Finally, 3176 before-and-after time-series image pairs of urban features were obtained. Professional computer vision annotation software was used to annotate the change areas of the before-and-after time-series image pairs of urban features in detail, and the corresponding change label maps were obtained. The before-and-after time-series images of urban features and the corresponding change label maps were used as training samples.

[0037] In some embodiments, the generator and the discriminator are alternately trained using training samples to obtain an urban land feature change monitoring model.

[0038] In some embodiments, conditional information (such as category labels, text descriptions, etc.) is introduced on the basis of the GAN model, so that the GAN model can generate data with specific attributes, forming a cGAN model (also known as a conditional generative adversarial network); Furthermore, the optimal generator of the GAN model It can be expressed as: , in, G represents the generator of the GAN model, D represents the discriminator of the GAN model, Represents the objective function of the discriminator of the GAN model.

[0039] In some embodiments, the training of the GAN model is as shown in the form of formula (1), first fix the generator of the GAN model, train the discriminator of the GAN model to maximize its objective function value, thereby maximizing the difference between the generated data and the real data; then fix the discriminator of the GAN model, train the generator of the GAN model to minimize the probability distribution divergence between the predicted data calculated by the discriminator of the GAN model and the real data, that is, try to generate data that makes the discriminator of the GAN model lose its discrimination ability. Through iterative training, the generator and discriminator of the GAN model compete with each other until the discriminator of the GAN model cannot distinguish between true and false.

[0040] In some embodiments, based on the cGAN model, a model for monitoring urban landform changes is constructed, and the model for monitoring urban landform changes includes a generator G1 and a discriminator D1; wherein, Generator G1 includes a U-Net network and a BiSeNet network. The U-Net network consists of an encoder and a decoder and introduces jump connections; the BiSeNet network introduces a feature fusion module and an attention refinement module.

[0041] In some embodiments, the U-Net network is as follows: Figure 2 As shown in (a), the main feature is that it has a symmetrical encoder-decoder structure. The encoder is the downsampling path of the U-Net network, which mainly uses continuous convolutional pooling layers to extract feature information from the image and gradually maps the feature information to high dimensions. The highest dimension of the entire network includes rich feature information in the entire image; the decoder is the upsampling path of the U-Net network. Unlike the Fully Convolutional Networks (FCN), the U-Net model maps high-dimensional features to low dimensions again through deconvolution. In the mapping process, in order to enhance the accuracy of segmentation, the images with the same dimension in the contraction network with the same dimension will be fused. Since the dimension will become twice the original dimension during the fusion process, it is necessary to convolve again at this time to ensure that the dimension after processing is the same as the dimension before the fusion operation so that it can be fused with the image with the same dimension after another deconvolution, until the image can be output with the same dimension as the original image. For example, after the original remote sensing image is input into the U-Net network, the encoder part first gradually extracts high-level features of the image through convolution and pooling operations, and the size of the feature map continues to decrease. Then, in the decoder part, deconvolution and feature fusion are used to gradually restore the image resolution, combining low-level details and high-level semantic information, and finally outputting a predicted image with the same or similar size as the input image.

[0042] In some embodiments, the BiSeNet network is as follows Figure 2As shown in (b), the BiSeNet network consists of a spatial path (SpatialPatch, SP) and a context path (Context Patch, CP). The spatial path with a small step size is used to retain the spatial location information, thereby generating a high-resolution feature map; the semantic path with a fast downsampling rate is used to obtain an objective receptive field. In addition, BiSeNet introduces a feature fusion module (Feature Fusion Module, FFM) and an attention refinement module (Attention Refinement Module, ARM) to further improve the segmentation performance; For example, first, the original remote sensing image enters the two branches of the BiSeNet network. One branch is the spatial path, which quickly downsamples to retain spatial detail information, and extracts shallow features of the image through a series of convolution operations, focusing mainly on spatial information such as boundaries and textures in the image; the other branch is the context path, which gradually extracts high-level semantic features of the image through a deeper network structure to understand the category information of different objects in the image. Then, the features extracted by the two branches are fused. The fused features contain both rich spatial details and strong semantic information. Finally, the fused features are output through output layers such as classifiers to output the prediction results of the image.

[0043] In some embodiments, the discriminator D1 includes a LcGAN discriminator and a PatchGAN discriminator, wherein the LcGAN discriminator evaluates the image globally in combination with the object category label; the PatchGAN discriminator evaluates the local pixel blocks in the image, that is, divides the image into multiple fixed-size blocks (Patches), judges the authenticity of each Patch separately, and finally takes the average value as the final output of the PatchGAN discriminator.

[0044] In some embodiments, the PatchGAN discriminator no longer outputs a single judgment of whether the entire image is true or false, but instead outputs a matrix in which each element represents the true or false of the corresponding patch in the input image. This ensures that each small block of the generated image has real urban features. Figure 2 As shown in (c), the PatchGAN discriminator is in full convolutional form. After the image passes through various convolutional layers, the output is mapped to an N×N matrix.

[0045] In some embodiments, the LcGAN discriminator improves the quality of the generated image by introducing conditional information, taking the location or marking information of the urban features as conditional input, so that it can perceive the specific location of the urban features in the local area, and can use a convolutional layer in the discriminator to combine the location information, and enhance the model's positioning ability for the urban features through conditional input; Exemplarily, after the high-resolution image training set x with a pixel size of 256×256 and the random noise z are input into the generator, the generated sample G1(x,z) generated according to the original image features is obtained, and then the generated sample G1(x,z) and the real sample y are input into the discriminator D1, where: The PatchGAN discriminator maps the input to an N×N patch feature map, and then uses the mean of the feature map as the output of the discriminator. From the patch feature map, we can trace back to a certain position in the original image, and we can see the impact of this position on the final output result. The LcGAN discriminator extracts image features from a global perspective, including the overall distribution of urban objects, their relationship with surrounding objects, and object category label information. Since the LcGAN discriminator takes object category labels into consideration, it checks whether the changes in urban objects in the generated image are consistent with the category characteristics of urban objects. If the changes in the generated image are inconsistent with the category characteristics of urban objects, the LcGAN discriminator will reduce the probability of the generated image being judged as a real image.

[0046] In some embodiments, the urban land feature change monitoring model is trained, and the specific training steps are as follows: 1) Set the objective function Overall objective function As shown in formula (2): , in, is the objective function of the cGAN model, is the objective function for a pixel pair, is the weight coefficient; where As shown in formula (3): , in, represents the spatial penalty term, represents mathematical expectation, and y represents real sample data; Furthermore, in order to solve the problem of imbalanced positive and negative samples caused by sparse distribution of road networks in high-resolution images, the traditional A space penalty term has been added to , the calculation formula of the spatial penalty term is shown in formula (4): , If the pixel If it is not a road, it is set to 1; if the pixel If it is a road, Set as ; As shown in formula (5):

[0047] In some embodiments, conventional Using JS divergence to evaluate the gap between real samples and generated samples may lead to problems such as "gradient disappearance" and "mode collapse". Therefore, the Wasserstein distance is used instead of the traditional JS divergence. As shown in formula (6): , in, is the penalty coefficient, Represents the output result of generator G1, express about gradient.

[0048] 2) Objective function optimization The training set is input into the generator and discriminator of the urban land feature change monitoring model, and the objective function is calculated according to formula (2): , and optimize the value of the objective function, and update the network parameters of the generator and the discriminator respectively, that is, first fix G1, maximize D1, update the parameters of D1, then fix D1, minimize G1, and update the parameters of G1, the optimal generator of the urban feature change monitoring model It can be expressed as:

[0049] 3) Iterative training Repeat the training process until the optimization result is close to the Nash equilibrium or the loss function of the generator and the discriminator converges, and the optimal generator is obtained. .

[0050] In some embodiments, an alternating training strategy is adopted to train the generator after multiple iterations of the discriminator, effectively avoiding the mode collapse problem; Furthermore, the discriminator is usually iterated multiple times (e.g., 5 times) and then the generator is trained once to ensure that the discriminator can distinguish between real and generated images well.

[0051] In some embodiments, the urban feature change monitoring model is trained on a training set to obtain a model weight for model evaluation; Specifically, during the training process, the initial learning rate is set to 0.005, the optimizer uses SGD, the weight decay is 0.0002, the momentum is 0.9, and the update strategy is Ploy; where power is a hyperparameter used to control the speed at which the learning rate decreases. When power=1, the Poly update strategy is equivalent to the linear learning rate decrease strategy. When power<1, the speed at which the learning rate decreases will slow down. When power>1, the speed at which the learning rate decreases will accelerate. The batch size is set to 8, and the total number of training times is 100 epochs. The training ends when the loop iterates until the epoch reaches 100.

[0052] In some embodiments, precision, recall, F1 score, and intersection over union (IOU) are selected as evaluation indicators. Cross-validation can also be used to evaluate the performance of the urban land feature change monitoring model on different data sets. In addition to quantitative indicators, subjective evaluation can also be performed by comparing the visually generated change images with the real change images to ensure the effectiveness of the model in actual scenarios.

[0053] In some embodiments, the before and after time series images of urban features are input into a pre-trained urban feature change monitoring model to obtain urban feature change results, and an interpolation analysis algorithm is used on the urban feature change results to monitor and warn of changes in urban features.

[0054] In some embodiments, the obtained urban change result data is used as known sample points, and the urban land feature changes are predicted and extended using interpolation algorithms. Changes of different land feature types can also be classified and monitored, such as separately monitoring changes in buildings, water bodies, and vegetation, so as to enable more targeted management and decision-making. For example, taking buildings as an example, the buildings are analyzed, including adding new buildings, reducing buildings, increasing building areas, and reducing building areas; the data of multiple image periods are compared to obtain comparative analysis results, which are intuitively displayed on the map.

[0055] In some embodiments, according to the characteristics and needs of changes in urban features, early warning indicators and thresholds are set, and when the interpolation analysis result exceeds the early warning threshold, an early warning signal is issued; For example, a certain area in Qingdao City is divided into different grids. The decision tree algorithm is used to determine that the area of ​​the best grid division is 1 square kilometer. The warning threshold is set based on the quantitative evaluation results, and an early warning model is constructed. When the monitored urban changes exceed the warning threshold, the early warning mechanism is automatically triggered. The data is interpreted by the background for comparative analysis, and the changes within the grid are identified and early warnings are issued.

[0056] According to the embodiments of the present disclosure, by constructing an urban feature change monitoring model, the before and after time series images of urban features are input into the trained urban feature change monitoring model to obtain the urban feature change results, thereby realizing the monitoring and early warning of urban feature changes.

[0057] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0058] The above is an introduction to the method embodiment. The following is a further explanation of the scheme disclosed in the present invention through an apparatus embodiment.

[0059] Figure 3 FIG. 3 is a block diagram of an urban change monitoring, analysis and early warning system 300 based on remote sensing object data according to an embodiment of the present disclosure. Figure 3 As shown, the device 300 includes: The acquisition module 301 is used to acquire the before and after time-series images of urban features.

[0060] The monitoring and early warning module 302 is used to input the before and after time series images of the urban features into the pre-trained urban feature change monitoring model to obtain the urban feature change results; wherein, The pre-trained urban feature change monitoring model is built based on the cGAN model, including: Generator and Discriminator; The pre-trained urban land feature change monitoring model is trained by the following steps: Obtaining before and after time series images of urban features; Annotate the change area of ​​the before and after time series images of the urban features to obtain a corresponding change label map, and use the before and after time series images of the urban features and the corresponding change label map as training samples; The generator and discriminator are trained alternately using training samples to obtain a model for monitoring urban land feature changes.

[0061] In some embodiments, module 302 is specifically configured to: The generator includes U-Net network and BiSeNet network; among them, The U-Net network consists of an encoder and a decoder and introduces skip connections; The BiSeNet network introduces a feature fusion module and an attention refinement module.

[0062] In some embodiments, module 302 is further configured to: The discriminator includes LcGAN discriminator and PatchGAN discriminator; among them, The LcGAN discriminator evaluates the image globally in combination with the object category labels; The PatchGAN discriminator evaluates local patches of pixels in the image.

[0063] In some embodiments, module 302 is further configured to: The overall objective function of the urban land feature change monitoring model is: , in, is the objective function of the cGAN model, is the objective function for a pixel pair, is the weight coefficient; where Includes a space penalty term.

[0064] In some embodiments, module 302 is further configured to: The objective function of the cGAN model includes a gradient penalty term to further stabilize the training process and prevent mode collapse.

[0065] In some embodiments, module 302 is further configured to: Using training samples to alternately train the generator and discriminator includes: First fix the generator and train the discriminator, then fix the discriminator and train the generator.

[0066] In some embodiments, module 302 is further configured to: Interpolation analysis algorithm is used to monitor and warn of changes in urban features.

[0067] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0068] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device and a readable storage medium.

[0069] Figure 4A schematic block diagram of an electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0070] The electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in the ROM 402 or a computer program loaded from the storage unit 408 into the RAM 403. In the RAM 403, various programs and data required for the operation of the electronic device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An I / O interface 405 is also connected to the bus 404.

[0071] Multiple components in the electronic device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0072] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above, such as method 100. For example, in some embodiments, the method 100 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform the method 100 in any other appropriate manner (e.g., by means of firmware).

[0073] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0074] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0075] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0076] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0077] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0078] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0079] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of this disclosure can be achieved, and this document is not limited here.

[0080] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for monitoring, analyzing and warning urban changes based on remote sensing data, comprising: Obtaining before and after time series images of urban features; The before and after time series images of the urban features are input into the pre-trained urban feature change monitoring model to obtain the urban feature change results; wherein, The pre-trained urban feature change monitoring model is built based on the cGAN model, including: Generator and Discriminator; The pre-trained urban land feature change monitoring model is trained by the following steps: Obtaining before and after time series images of urban features; Annotate the change area of ​​the before and after time series images of the urban features to obtain a corresponding change label map, and use the before and after time series images of the urban features and the corresponding change label map as training samples; The generator and discriminator are trained alternately using training samples to obtain a model for monitoring urban land feature changes.

2. The method according to claim 1, characterized in that The generator includes a U-Net network and a BiSeNet network; wherein, The U-Net network consists of an encoder and a decoder and introduces skip connections; The BiSeNet network introduces a feature fusion module and an attention refinement module.

3. The method according to claim 1, characterized in that The discriminator includes an LcGAN discriminator and a PatchGAN discriminator; wherein, The LcGAN discriminator evaluates the image globally in combination with the object category labels; The PatchGAN discriminator evaluates local patches of pixels in an image.

4. The method according to claim 1, characterized in that: The overall objective function of the urban land feature change monitoring model is: ， in, is the objective function of the cGAN model, is the objective function for a pixel pair, is the weight coefficient; where Said Includes a space penalty term.

5. The method according to claim 4, characterized in that The objective function of the cGAN model includes a gradient penalty term to further stabilize the training process and prevent mode collapse.

6. The method according to claim 1, characterized in that The alternate training of the generator and the discriminator using the training samples comprises: First fix the generator and train the discriminator, then fix the discriminator and train the generator.

7. The method according to claim 1, characterized in that The method further comprises: Interpolation analysis algorithm is used to monitor and warn of changes in urban features.

8. An urban change monitoring, analysis and early warning system based on remote sensing ground data, comprising: Acquisition module, used to obtain before and after time series images of urban features; The monitoring and early warning module is used to input the before and after time series images of the urban features into the pre-trained urban feature change monitoring model to obtain the urban feature change results; wherein, The pre-trained urban feature change monitoring model is built based on the cGAN model, including: Generator and Discriminator; The pre-trained urban land feature change monitoring model is trained by the following steps: Obtaining before and after time series images of urban features; Annotate the change area of ​​the before and after time series images of the urban features to obtain a corresponding change label map, and use the before and after time series images of the urban features and the corresponding change label map as training samples; The generator and discriminator are trained alternately using training samples to obtain a model for monitoring urban land feature changes.

9. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.

Images