Remote sensing image target trend detection method and device, equipment and medium

Abstract

The invention provides a remote sensing image target trend detection method which can be applied to the technical field of remote sensing image processing and analysis. The method comprises the steps that an input image is acquired and preprocessed; based on spectral features and geographic position prior information, a semantic segmentation network is adopted to separate a target area from the preprocessed image; constructing a cost measurement function by using an optimal transmission theory, solving an optimal transmission plan, and obtaining feature mapping after image smoothing; based on feature mapping, texture features and motion features are extracted, and spatial-temporal features of the target area are obtained through fusion; capturing a space-time dependency relationship in the multi-temporal remote sensing image feature sequence by adopting a time sequence analysis model; constructing a target trend prediction model based on the feature mapping, the spatio-temporal features and the spatio-temporal dependency relationship; and outputting a trend prediction result by using the target trend prediction model. The invention further provides a remote sensing image target trend detection device, equipment and a medium.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing and analysis, and more particularly to a remote sensing image target trend detection method, device, equipment and medium. Background Art

[0002] Currently, remote sensing image target trend detection mainly adopts methods based on pixel difference, feature extraction and classification, and time series analysis.

[0003] Pixel difference-based methods identify changes in target areas by comparing pixel values ​​in remote sensing images at different time points. While this approach is simple and straightforward, it is susceptible to environmental factors such as lighting conditions and seasonal variations, leading to a high false positive rate.

[0004] Feature extraction and classification methods first preprocess images to extract target features such as color, texture, and shape. Classification and trend prediction are then performed using machine learning or deep learning models. This method offers high accuracy, but suffers from high computational complexity and a strong reliance on feature selection. Common approaches include image preprocessing (denoising and enhancement), feature extraction (using algorithms such as SIFT and HOG), feature selection, model training (such as SVM and CNN), and trend prediction. These methods aim to extract target objects through segmentation of multi-temporal remote sensing images and analyze the results based on geometric attributes or spectral texture features. However, this method faces challenges arising from varying segmentation results, and the accuracy of trend detection depends heavily on segmentation quality.

[0005] Time series analysis methods treat remote sensing image data as time series data and apply statistical models (such as ARIMA and LSTM) to analyze target trends. While these methods can capture long-term trends, they lack consideration of the spatial characteristics of the image data. The key concept is to overlay multi-temporal data to extract target objects, identifying those with consistent shape, size, and location across multi-temporal imagery. Target trend detection is then performed using either a classification-first-then-detection or detection-first-then-classification approach. The architecture typically includes three main steps: input of semantically labeled data, classifier selection, and change extraction. Traditional classification-first-then-detection methods rely on comparing raw spectral data after classification. While they can provide detailed information on change category migration, they are limited by the accuracy of single-period remote sensing image classification. Single-temporal classification errors can propagate to the multi-temporal change detection process, leading to cumulative errors. Detection-first-then-classification methods, on the other hand, directly perform interpolation or overlay on multi-temporal data before classification to obtain comprehensive detection results.

[0006] However, existing methods have numerous shortcomings. Time series analysis often ignores spatial continuity in images, affecting the accuracy of target trend detection. Pixel difference methods are susceptible to environmental factors, leading to false positives. Traditional algorithms rely heavily on precise registration of multi-temporal data during target object extraction. Furthermore, due to the simultaneous use of image feature parameters from multiple time points, improper parameter settings can easily lead to over-segmentation or under-segmentation of object units, resulting in high heterogeneity in the extracted target objects, oversampling, and boundary discontinuities, leading to a "distribution gap." Furthermore, in remote sensing image trend analysis, a "spatiotemporal semantic gap" exists between low-level object changes and high-level semantic changes in targets. While traditional pixel-level or feature-based analysis methods can capture specific changes at the object level, they cannot determine overall classification transitions for high-level targets. Summary of the Invention

[0007] (1) Technical issues to be resolved

[0008] In order to solve the target heterogeneity caused by inconsistent feature parameters in the multi-temporal target extraction process when processing remote sensing images in existing methods and alleviate the "distributed gap" phenomenon, the present invention provides a remote sensing image target trend detection method, device, equipment and medium. Through the nonlinear mapping capability of optimal transmission, the hidden connection between low-level ground object changes and high-level scene semantic changes is established, and the "spatiotemporal semantic gap" phenomenon is alleviated through time series modeling, ultimately improving the accuracy and robustness of target trend detection and effectively utilizing the continuous spatial information implicit in multiple remote sensing images.

[0009] (2) Technical solution

[0010] In response to the above technical problems, embodiments of the present invention provide a method, apparatus, device, and medium for detecting target trends in remote sensing images.

[0011] According to a first aspect of the present invention, a method for detecting target trends in remote sensing images is provided, comprising: obtaining an input image, the input image comprising a plurality of continuous temporal remote sensing images; preprocessing the input image to obtain a preprocessed image; separating a target area from the preprocessed image using a semantic segmentation network based on spectral features and prior information of geographic location; utilizing optimal transmission theory and a Sinkhorn iterative algorithm to calculate the pixel-level Wasserstein distance between images, constructing a cost metric function, solving an optimal transmission plan, and obtaining a smoothed feature map of an image based on the optimal transmission plan; extracting texture features based on the feature map using a local binary pattern algorithm and an optical flow method to extract motion features, fusing the texture features and the motion features to obtain the spatiotemporal features of the target area; utilizing a time series analysis model to capture the spatiotemporal dependencies in the feature sequences of multi-temporal remote sensing images; constructing a target trend prediction model based on the feature map, the spatiotemporal features, and the spatiotemporal dependencies; utilizing the target trend prediction model to output trend prediction results, determining the trend type by setting a threshold, and drawing a trend graph.

[0012] In some exemplary embodiments, the input image is preprocessed, including: using Gaussian filtering to remove image noise and performing mask processing to reduce the influence of noise and interference areas; using scale-invariant feature transformation feature matching method to linearly normalize the image brightness to the same range to ensure brightness consistency between images; and using image registration and orthorectification methods to spatially align multi-temporal images to eliminate illumination and atmospheric differences.

[0013] In some exemplary embodiments, optimal transmission theory is utilized, the pixel-level Wasserstein distance between images is calculated based on the Sinkhorn iterative algorithm, a cost metric function is constructed, an optimal transmission plan is solved, and a smoothed feature map of an image is obtained based on the optimal transmission plan, including: utilizing optimal transmission theory, the pixel-level Wasserstein distance between images is calculated based on the Sinkhorn iterative algorithm, and a cost metric function for the optimal transmission map is constructed; a cost matrix of the two images is obtained based on the cost metric function, the cost matrix including correlation calculation results of corresponding pixel points of the two images; an optimal transmission plan between the two images is solved based on the cost matrix, the optimal transmission plan satisfies the optimal solution under specific constraints; and a smoothed feature map of an image is obtained based on the solved optimal transmission plan.

[0014] In some exemplary embodiments, the semantic segmentation network is a deep learning network trained on a large amount of labeled remote sensing image data to accurately identify and extract target areas.

[0015] In some exemplary embodiments, based on feature mapping, texture features are extracted in combination with a local binary pattern algorithm and motion features are extracted using an optical flow method, and the texture features and motion features are fused to obtain the spatiotemporal features of the target area, including: based on feature mapping, texture features of the target area are extracted using a local binary pattern algorithm, and a texture feature descriptor is generated by comparing the grayscale value relationship between a pixel point and its neighboring pixels to characterize the texture information of the target area; based on feature mapping, motion features of the target area are extracted using an optical flow method, and a motion vector of the pixel is calculated based on the temporal change of the pixel in the image sequence and the correlation between adjacent frames, thereby obtaining the motion information of the target area; and texture information and motion information are fused using feature splicing or weighted fusion to form the spatiotemporal features of the target area, and the spatiotemporal features can comprehensively reflect the characteristics of the target area at different times and spaces.

[0016] In some exemplary embodiments, a time series analysis model is used to capture the spatiotemporal dependencies in a feature sequence of multi-temporal remote sensing images, including: organizing the spatiotemporal features of the multi-temporal remote sensing images into a feature sequence in chronological order, and inputting the feature sequence into a time series analysis model, the time series analysis model including a long short-term memory network or a gated recurrent unit; using the time series analysis model, learning the dependencies between different time steps in the feature sequence to capture the changing patterns of the target area in the time dimension and the changing trends of spatial features over time.

[0017] In some exemplary embodiments, a target trend prediction model is used to output trend prediction results, the trend type is judged by setting a threshold, and a trend graph is drawn, including: the multi-phase remote sensing image data to be predicted is subjected to the same preprocessing, feature extraction and other steps as the training data, and then input into the trained target trend prediction model to obtain the prediction result of the target trend; according to actual application requirements and historical data statistics, thresholds of different trend types are set; the prediction results are compared with the set thresholds to judge the type of the target trend; and according to the judgment result of the type of the target trend, a trend graph is drawn in a visual manner to intuitively show the changing trend of the target area at different times.

[0018] According to a second aspect of the present invention, a remote sensing image target trend detection device is provided, comprising: a first acquisition module for acquiring an input image, the input image comprising a plurality of continuous temporal remote sensing images; a preprocessing module for preprocessing the input image to obtain a preprocessed image; a second acquisition module for separating a target region from the preprocessed image using a semantic segmentation network based on spectral features and prior information about geographic location; a third acquisition module for calculating the pixel-level Wasserstein distance between images using the Sinkhorn iterative algorithm using optimal transmission theory, constructing a cost metric function, solving an optimal transmission plan, and obtaining a smoothed feature map of an image based on the optimal transmission plan; a fourth acquisition module for extracting texture features based on the feature map in combination with a local binary pattern algorithm and an optical flow method to extract motion features, fusing the texture features and motion features to obtain spatiotemporal features of the target region; a fifth acquisition module for capturing the spatiotemporal dependencies in the feature sequences of multi-temporal remote sensing images using a time series analysis model; a model construction module for constructing a target trend prediction model based on the feature map, the spatiotemporal features, and the spatiotemporal dependencies; and a prediction module for outputting a trend prediction result using the target trend prediction model, determining the trend type by setting a threshold, and drawing a trend graph.

[0019] According to a third aspect of the present invention, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0020] The fourth aspect of the present invention further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the computer program or instructions are executed by a processor.

[0021] (3) Beneficial effects

[0022] It can be seen from the above technical solutions that the remote sensing image target trend detection method, device, equipment and medium provided by the embodiments of the present invention have at least the following beneficial effects:

[0023] (1) Through optimal transmission mapping, the heterogeneity of multi-temporal target area features is resolved; the hidden connection between low-level object changes and high-level scene semantic changes is established, which more accurately captures subtle changes between images, reduces noise interference, and improves the accuracy of trend detection.

[0024] (2) By extracting spatiotemporal features, the spatial continuity of remote sensing images is effectively utilized, and the spatial resolution and credibility of trend prediction are improved.

[0025] (3) Through the temporal modeling based on the GRU module, the spatiotemporal association is mined, important feature information is retained and irrelevant content is filtered out, thereby alleviating the spatiotemporal semantic gap phenomenon, ultimately improving the accuracy and robustness of target trend detection, and effectively utilizing the continuous spatial information implicit in multiple remote sensing images. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above contents and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0027] Figure 1 The following schematically shows a flow chart of a method for detecting target trends in remote sensing images according to an embodiment of the present invention;

[0028] Figure 2 The following schematically shows an architecture diagram of a method for detecting target trends in remote sensing images according to an embodiment of the present invention;

[0029] Figure 3 Schematically showing the target trend detection result of the optimal transmission remote sensing image according to an embodiment of the present invention;

[0030] Figure 4 A remote sensing image target trend detection device according to an embodiment of the present invention is schematically shown;

[0031] Figure 5 The block diagram schematically shows an electronic device for a method for detecting target trends in remote sensing images according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concept of the present invention.

[0033] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise," "include," etc. used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0034] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0035] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0036] Figure 1 A flowchart of a method for detecting target trends in remote sensing images according to an embodiment of the present invention is schematically shown; Figure 2 The following schematically shows an architecture diagram of a remote sensing image target trend detection method according to an embodiment of the present invention.

[0037] like Figure 1 and Figure 2 As shown, a method for detecting target trends in remote sensing images according to an embodiment of the present invention includes steps S110 to S180.

[0038] In step S110 , an input image is acquired, where the input image includes a plurality of continuous temporal remote sensing images.

[0039] In step S120 , the input image is preprocessed to obtain a preprocessed image.

[0040] In an embodiment of the present invention, the input image is preprocessed, including: using Gaussian filtering to remove image noise and performing mask processing to reduce the influence of noise and interference areas; using the Scale-Invariant Feature Transform (SIFT) feature matching method to linearly normalize the image brightness to the same range to ensure brightness consistency between images; and using image registration and orthorectification methods to spatially align multi-temporal images to eliminate lighting and atmospheric differences.

[0041] In step S130 , a semantic segmentation network is used to separate the target area from the preprocessed image based on the spectral features and the geographic location prior information.

[0042] In this embodiment of the present invention, the semantic segmentation network is a deep learning network trained on a large amount of annotated remote sensing image data to accurately identify and extract target areas. Step S130 specifically includes defining the region of interest using prior information such as geographic location and spectral characteristics. A threshold-based, morphological, or semantic segmentation network is then used to extract the relevant target area, reducing background interference and focusing on changes in the target area, thereby improving the accuracy of trend detection and the efficiency of subsequent analysis.

[0043] In step S140, the optimal transmission theory is used to calculate the pixel-level Wasserstein distance between images based on the Sinkhorn iterative algorithm, a cost metric function is constructed, an optimal transmission plan is solved, and a smoothed feature map of the image is obtained based on the plan.

[0044] In an embodiment of the present invention, step S140 may specifically include: utilizing optimal transmission theory and calculating the pixel-level Wasserstein distance between images based on the Sinkhorn iterative algorithm to construct a cost measurement function for the optimal transmission mapping; obtaining a cost matrix of the two images based on the cost measurement function, the cost matrix including the correlation calculation results of the corresponding pixel points of the two images; solving the optimal transmission plan between the two images based on the cost matrix, the optimal transmission plan meeting the optimal solution under specific constraints; and obtaining a smoothed feature map of the image based on the solved optimal transmission plan.

[0045] Introducing optimal transmission theory into trend detection in remote sensing images precisely quantifies inter-image variations and improves trend detection accuracy. This method constructs the forward transmission cost using a Gaussian kernel function and solves the optimal transmission scheme through Sinkhorn iteration, achieving information migration and fusion. This addresses target heterogeneity caused by inconsistent feature parameters during multi-temporal target extraction and alleviates the distribution gap phenomenon.

[0046] For example, by using the optimal transfer theory and the Sinkhorn iterative algorithm, the Wasserstein distance between pixels is calculated to alleviate the distribution gap in the extraction results and achieve smooth transition of image features and accurate quantification of changes.

[0047] Given a remote sensing image at time N and the remote sensing image at time N+1 , first construct the cost metric function of the optimal transmission mapping, set is a reproducing kernel Hilbert space (RKHS) and its associated kernel embedding Gaussian kernel function : .in Expressed as a d-dimensional linear space over the real number field R. Then we can get The cost matrix , contains the dot product results of the two images:

[0048] (1)

[0049] Based on the cost matrix, and The optimal transmission plan between The matrix is ​​defined as the unique solution of the following equation:

[0050] (2)

[0051] (3)

[0052] in, represents the transmission cost, is the entropy regularization function, Responsible for controlling the sparsity of the transmission cost matrix. U is the space of acceptable coupling, and the source domain vector of the optimal transmission is , the target domain vector is b, then Subject to the following conditions:

[0053] (4)

[0054] in, represents the space of positive real numbers, Represents an n-dimensional vector whose element value is 1.

[0055] Based on the solved transmission plan get Smoothed feature map :

[0056] (5)

[0057] in, is the transmission target vector ( ) is the number of elements in .

[0058] In step S150, based on feature mapping, the local binary pattern algorithm is combined to extract texture features and the optical flow method is combined to extract motion features, and the texture features and motion features are fused to obtain the spatiotemporal features of the target area. The spatiotemporal features include color, texture changes, and spatial position movement.

[0059] In an embodiment of the present invention, step S150 may specifically include: based on feature mapping, using a local binary pattern algorithm to extract texture features of the target area, and generating a texture feature descriptor by comparing the grayscale value relationship between the pixel point and its neighboring pixel points to characterize the texture information of the target area; based on feature mapping, using the optical flow method to extract motion features of the target area, and calculating the motion vector of the pixel according to the temporal change of the pixel in the image sequence and the correlation between adjacent frames, thereby obtaining the motion information of the target area; using feature splicing or weighted fusion to fuse the texture information and motion information to form the spatiotemporal features of the target area, and the spatiotemporal features can comprehensively reflect the characteristics of the target area at different times and spaces.

[0060] This structure is based on a multi-head self-attention module and takes multi-dimensional features such as color, texture, and position as input. After optimizing the relevant task objectives of trend prediction, it can adaptively fuse and filter key features, fully considering the spatial and temporal continuity of remote sensing images and enhancing the robustness of the model.

[0061] In step S160 , a time series analysis model is used to capture the spatiotemporal dependencies in the multi-temporal remote sensing image feature sequences.

[0062] In an embodiment of the present invention, step S160 may specifically include: organizing the spatiotemporal features of the multi-temporal remote sensing image into a feature sequence in chronological order, and inputting the feature sequence into a time series analysis model, the time series analysis model including a long short-term memory network or a gated recurrent unit; using the time series analysis model, learning the dependency between different time steps in the feature sequence to capture the change law of the target area in the time dimension and the change trend of the spatial features over time.

[0063] In step S170, a target trend prediction model is constructed based on the feature map, spatiotemporal features, and spatiotemporal dependencies.

[0064] This architecture, based on the GRU model, introduces update and reset gates to control the flow of information. The update gate determines the degree of integration of current information with historical information, while the reset gate regulates the influence of the previous state on the current moment. This effectively captures spatiotemporal correlations in image sequences and mitigates the spatiotemporal semantic gap. This architecture maintains modeling capabilities while reducing the number of parameters and effectively improving computational efficiency.

[0065] In step S180, the target trend prediction model is used to output the trend prediction result, the trend type is determined by setting a threshold, and a trend graph is drawn.

[0066] In an embodiment of the present invention, step S180 may specifically include: inputting the multi-temporal remote sensing image data to be predicted into the trained target trend prediction model after undergoing the same preprocessing, feature extraction and other steps as the training data to obtain the prediction results of the target trend; setting thresholds for different trend types based on actual application requirements and historical data statistics; comparing the prediction results with the set thresholds to determine the type of target trend; and drawing a trend graph in a visual manner based on the determination result of the target trend type to intuitively display the changing trend of the target area at different times.

[0067] Figure 3 The result of target trend detection of the optimally transmitted remote sensing image according to an embodiment of the present invention is schematically shown.

[0068] like Figure 3 As shown, Figure 3 From left to right in the center are earlier remote sensing images of towns in the desert region, later remote sensing images of towns after remediation, and a graph of the subsequent changes. The first two satellite remote sensing images provide a visual comparison of the same region over time, offering insights into spatial shifts such as urban relocation and adjustments in farmland patterns, and allowing for inferences about the speed and magnitude of these shifts. The third, abstract two-dimensional image provides a clearer quantification of these regional changes. Combined with the forecast results, regional development trends can be identified, such as the economic vitality and potential pressures reflected by sustained urban growth, and the agricultural processes and policy effects associated with farmland changes. Potential problems, such as insufficient regional development momentum or land use conflicts, can also be identified, providing important insights for regional planning, resource management, and policy development, ultimately contributing to sustainable regional development.

[0069] Figure 4 A remote sensing image target trend detection device according to an embodiment of the present invention is schematically shown.

[0070] like Figure 4 As shown, the remote sensing image target trend detection device 800 of this embodiment includes a first acquisition module 810, a preprocessing module 820, a second acquisition module 830, a third acquisition module 840, a fourth acquisition module 850, a fifth acquisition module 860, a model building module 870 and a prediction module 880.

[0071] The first acquisition module 810 is used to acquire input images, where the input images include a plurality of continuous temporal remote sensing images.

[0072] The preprocessing module 820 is used to preprocess the input image to obtain a preprocessed image.

[0073] The second acquisition module 830 is used to separate the target area from the pre-processed image using a semantic segmentation network based on the spectral characteristics and the geographic location prior information.

[0074] The third acquisition module 840 is used to use the optimal transmission theory and the Sinkhorn iterative algorithm to calculate the pixel-level Wasserstein distance between images, construct a cost metric function, solve the optimal transmission plan, and obtain a smoothed feature map of the image based on the plan.

[0075] The fourth acquisition module 850 is used to extract texture features based on feature mapping in combination with the local binary pattern algorithm and the optical flow method to extract motion features, and fuse the texture features and motion features to obtain the spatiotemporal features of the target area.

[0076] The fifth acquisition module 860 is used to capture the spatiotemporal dependency in the multi-temporal remote sensing image feature sequence using a time series analysis model.

[0077] The model building module 870 is used to build a target trend prediction model based on feature mapping, spatiotemporal features, and spatiotemporal dependencies.

[0078] The prediction module 880 is used to output trend prediction results using the target trend prediction model, determine the trend type by setting a threshold, and draw a trend graph.

[0079] According to an embodiment of the present invention, any multiple modules among the first acquisition module 810, the preprocessing module 820, the second acquisition module 830, the third acquisition module 840, the fourth acquisition module 850, the fifth acquisition module 860, the model building module 870, and the prediction module 880 may be combined into a single module for implementation, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in a single module. According to an embodiment of the present invention, at least one of the first acquisition module 810, the preprocessing module 820, the second acquisition module 830, the third acquisition module 840, the fourth acquisition module 850, the fifth acquisition module 860, the model building module 870, and the prediction module 880 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or may be implemented in hardware or firmware by any other reasonable means of circuit integration or packaging, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the first acquisition module 810, the preprocessing module 820, the second acquisition module 830, the third acquisition module 840, the fourth acquisition module 850, the fifth acquisition module 860, the model building module 870, and the prediction module 880 may be at least partially implemented as a computer program module, which, when executed, may perform the corresponding function.

[0080] Figure 5 The block diagram schematically shows an electronic device for a method for detecting target trends in remote sensing images according to an embodiment of the present invention.

[0081] like Figure 5 As shown, an electronic device 900 according to an embodiment of the present invention includes a processor 901, which can perform various appropriate actions and processes based on programs stored in a read-only memory (ROM) 902 or programs loaded from a storage unit 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or related chipsets and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0082] The RAM 903 stores various programs and data required for the operation of the electronic device 900. The processor 901, ROM 902, and RAM 903 are connected to each other via a bus 904. The processor 901 executes the programs in the ROM 902 and / or RAM 903 to perform various operations according to the method flow of the embodiment of the present invention. It should be noted that the programs may also be stored in one or more memories other than the ROM 902 and RAM 903. The processor 901 may also execute the programs stored in one or more memories to perform various operations according to the method flow of the embodiment of the present invention.

[0083] According to an embodiment of the present invention, electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to bus 904. Electronic device 900 may also include one or more of the following components connected to I / O interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 908 including a hard disk; and a communication section 909 including a network interface card such as a LAN card or modem. Communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to I / O interface 905 as needed. Removable media 911, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 910 as needed, so that computer programs read from the removable media can be installed into storage section 908 as needed.

[0084] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0085] According to an embodiment of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: 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), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present invention, a computer-readable storage medium may include the ROM 902 and / or RAM 903 described above, and / or one or more memories other than ROM 902 and RAM 903.

[0086] It will be understood by those skilled in the art that the features described in the various embodiments of the present invention may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention may be combined and / or coupled in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or couplings fall within the scope of the present invention.

Claims

1. A method for detecting target trends in remote sensing images, characterized in that: include: Acquiring an input image, wherein the input image includes a plurality of continuous time-phase remote sensing images; Preprocessing the input image to obtain a preprocessed image; Based on spectral features and geographic location prior information, a semantic segmentation network is used to separate the target area from the preprocessed image; Utilizing optimal transmission theory and the Sinkhorn iterative algorithm, the Wasserstein distance between images at the pixel level is calculated, a cost metric function is constructed, and an optimal transmission plan is solved. Based on the optimal transmission plan, a smoothed feature map of the image is obtained. Based on the feature map, the texture features are extracted by combining the local binary pattern algorithm and the motion features are extracted by the optical flow method, and the texture features and the motion features are fused to obtain the spatiotemporal features of the target area; A time series analysis model is used to capture the spatiotemporal dependencies in multi-temporal remote sensing image feature sequences; Building a target trend prediction model based on the feature map, the spatiotemporal features, and the spatiotemporal dependency; The target trend prediction model is used to output trend prediction results, the trend type is determined by setting a threshold, and a trend graph is drawn.

2. The method according to claim 1, characterized in that The preprocessing of the input image includes: Gaussian filtering is used to remove image noise and mask processing is performed to reduce the influence of noise and interference areas; Using scale-invariant feature transformation and feature matching to linearly normalize image brightness to the same range to ensure brightness consistency between images; and Multi-temporal images are spatially aligned through image registration and orthorectification methods to eliminate illumination and atmospheric differences.

3. The method according to claim 1, characterized in that The method utilizes the optimal transmission theory to calculate the pixel-level Wasserstein distance between images based on the Sinkhorn iterative algorithm, constructs a cost metric function, solves the optimal transmission plan, and obtains a smoothed feature map of the image based on the optimal transmission plan, including: Using the optimal transfer theory and the Sinkhorn iterative algorithm, the pixel-level Wasserstein distance between images is calculated to construct the cost metric function for the optimal transfer mapping. Obtaining a cost matrix of the two images based on the cost metric function, wherein the cost matrix includes correlation calculation results of corresponding pixel points of the two images; solving an optimal transmission plan between the two images based on the cost matrix, wherein the optimal transmission plan satisfies an optimal solution under specific constraints; and Based on the solved optimal transmission plan, a smoothed feature map of the image is obtained.

4. The method according to claim 1, wherein The semantic segmentation network is a deep learning network trained on a large amount of labeled remote sensing image data to accurately identify and extract target areas.

5. The method according to claim 1, wherein The method of extracting texture features based on the feature map and combining the local binary pattern algorithm with the optical flow method to extract motion features, fusing the texture features and the motion features to obtain the spatiotemporal features of the target area, includes: Based on the feature map, a local binary pattern algorithm is used to extract texture features of the target area, and a texture feature descriptor is generated by comparing the grayscale value relationship between a pixel point and its neighboring pixels to characterize the texture information of the target area; Based on the feature map, the optical flow method is used to extract motion features of the target area, and the motion vector of the pixel is calculated according to the temporal change of the pixel in the image sequence and the correlation between adjacent frames, thereby obtaining the motion information of the target area; The texture information and the motion information are fused by feature splicing or weighted fusion to form the spatiotemporal features of the target area. The spatiotemporal features can fully reflect the characteristics of the target area at different times and spaces.

6. The method according to claim 1, wherein The time series analysis model is used to capture the spatiotemporal dependency in the multi-temporal remote sensing image feature sequence, including: The spatiotemporal features of the multi-temporal remote sensing images are organized into a feature sequence in chronological order, and the feature sequence is input into a time series analysis model, wherein the time series analysis model includes a long short-term memory network or a gated recurrent unit; The time series analysis model is used to learn the dependencies between different time steps in the feature sequence, so as to capture the changing rules of the target area in the time dimension and the changing trends of spatial features over time.

7. The method according to claim 1, characterized in that The method of outputting a trend prediction result by using the target trend prediction model, determining a trend type by setting a threshold, and drawing a trend graph includes: The multi-temporal remote sensing image data to be predicted undergoes the same preprocessing and feature extraction steps as the training data, and is then input into the trained target trend prediction model to obtain the target trend prediction results. Set thresholds for different trend types based on actual application requirements and historical data statistics; Compare the prediction results with the set threshold to determine the type of target trend; According to the determination result of the target trend type, a trend graph is drawn in a visual manner to intuitively show the change trend of the target area at different times.

8. A remote sensing image target trend detection device, characterized in that: The device comprises: A first acquisition module is used to acquire an input image, wherein the input image includes a plurality of continuous time-phase remote sensing images; A preprocessing module, used for preprocessing the input image to obtain a preprocessed image; A second acquisition module is used to separate the target area from the pre-processed image using a semantic segmentation network based on spectral characteristics and geographic location prior information; a third acquisition module, configured to calculate the pixel-level Wasserstein distance between images using optimal transmission theory and the Sinkhorn iterative algorithm, construct a cost metric function, solve an optimal transmission plan, and obtain a smoothed feature map of an image based on the optimal transmission plan; A fourth acquisition module is configured to extract texture features based on the feature map in combination with a local binary pattern algorithm and a motion feature extracted using an optical flow method, and fuse the texture features and the motion features to obtain spatiotemporal features of the target area; The fifth acquisition module is used to capture the spatiotemporal dependency in the multi-temporal remote sensing image feature sequence using a time series analysis model; A model building module, configured to build a target trend prediction model based on the feature map, the spatiotemporal features, and the spatiotemporal dependency; The prediction module is used to output trend prediction results using the target trend prediction model, determine the trend type by setting a threshold, and draw a trend graph.

9. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more computer programs, The one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images