Water and soil loss monitoring method and device based on multi-source remote sensing data fusion

Abstract

The invention provides a water and soil loss monitoring method and device based on multi-source remote sensing data fusion, and relates to the technical field of deep learning, and the method comprises the steps: building a cross-modal physical response consistency mapping relation based on the collected optical remote sensing data and synthetic aperture radar data for a monitored landform region, and carrying out the consistency correction processing; extracting shared semantic features from different remote sensing data after processing, and performing structure compensation and sensitive weight adjustment; in the process of extracting the shared semantic features, correcting confusion region features in the optical remote sensing data; and processing confidence regions of different remote sensing features in the fusion feature space, and driving water and soil loss space partition extraction and water and soil loss quantitative evaluation by taking the fusion features as input, thereby realizing water and soil loss monitoring. According to the method, a multi-source remote sensing data-oriented cross-modal physical response consistency modeling and semantic sharing feature optimization mechanism is established, and the physical rationality and semantic expression capability of a fusion result are improved.

Description

Technical Field

[0001] The present application relates to the technical field of deep learning, and specifically to a soil and water loss monitoring method and device using multi-source remote sensing data fusion. Background Art

[0002] Soil and water loss monitoring refers to the technical process of systematically observing, quantifying, and spatially evaluating the erosion, transportation, and deposition of surface soils due to rainfall, runoff, wind, or human disturbance. The goal is to identify the spatial extent, evolutionary trends, and intensity changes of soil and water loss, and to assess the ecological and environmental risks and degree of land degradation in a watershed. Soil and water loss monitoring typically combines ground-based surveys with remote sensing data acquisition. Through terrain analysis, surface cover change detection, soil erosion model-driven development, and meteorological factor correlation reasoning, a monitoring system is established that integrates multi-temporal, multi-scale, and multi-source data. This allows for continuous tracking and precise early warning of soil and water loss dynamics in complex geomorphic areas, small watersheds, and even regional scales. It also provides a scientific basis for the layout of soil and water conservation projects, the formulation of ecological restoration strategies, and comprehensive watershed management.

[0003] Multi-source remote sensing monitoring of soil erosion usually includes optical remote sensing measurements and radar remote sensing measurements. Optical remote sensing measurements of soil erosion mainly rely on the high-resolution capture of visible light, near-infrared and short-wave infrared radiation signals reflected from the surface. By analyzing the surface vegetation cover index, bare land index, soil color changes and surface moisture characteristics, the spatial distribution and changing dynamics of soil erosion patches, erosion gullies and debris flow areas are extracted. It has the advantages of rich bands, fine classification, and high temporal and spatial resolution, but is easily affected by observation interference under cloudy, rainy or shielded conditions. Radar remote sensing measurement uses a synthetic aperture radar system to actively transmit microwave signals and receive their surface scattered echoes. Based on the backscattering intensity, polarization characteristics and interference change analysis, it inverts the surface roughness, soil moisture and micro-topography disturbance process. It can penetrate clouds and partial vegetation cover to achieve all-weather and all-day soil and water loss monitoring. It is particularly suitable for extracting the degree of erosion of exposed surfaces, changes in landslide volume and fine-scale landform deformation. The coordinated integration of optical remote sensing measurement and radar remote sensing measurement has become an important trend to improve the accuracy and reliability of dynamic soil and water loss monitoring.

[0004] Current multi-source remote sensing data fusion algorithms generally adopt surface feature combination methods such as pixel-level overlay, feature splicing or decision-level voting. They usually rely only on empirical normalization to perform simple processing of different source data. They lack systematic modeling and physical consistency alignment of the deep essential differences between optical remote sensing data and synthetic aperture radar remote sensing data in observation mechanism, response scale, physical magnitude and error characteristics. As a result, information redundancy, modal conflict amplification and ground feature category identification errors occur between different modal features during the fusion process, seriously affecting the reliability of the fused features and the accuracy and stability of downstream soil and water loss monitoring tasks. Therefore, it is urgent to establish cross-modal physical response consistency modeling and semantic sharing feature optimization mechanism for multi-source remote sensing data to improve the physical rationality and semantic expression ability of the fusion results. Summary of the Invention

[0005] This application provides a soil and water loss monitoring method and device for multi-source remote sensing data fusion, establishes a cross-modal physical response consistency modeling and semantic sharing feature optimization mechanism for multi-source remote sensing data, and improves the physical rationality and semantic expression ability of the fusion results.

[0006] In a first aspect of the present application, a soil and water loss monitoring method using multi-source remote sensing data fusion is provided, the method comprising:

[0007] Based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, a cross-modal physical response consistency mapping relationship is established, and the optical remote sensing data and the synthetic aperture radar remote sensing data are corrected for consistency in physical magnitude, response scale, and radiation dynamic range through a unified physical scale standardization mechanism;

[0008] After completing the physical scale normalization process, shared semantic features are extracted from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, and structural compensation and sensitivity weight adjustment are performed on the optical remote sensing features and the synthetic aperture radar remote sensing features using a multi-scale residual structure and a dynamic spatial attention mechanism, wherein the optical remote sensing features are features corresponding to the corrected optical remote sensing data, and the synthetic aperture radar remote sensing features are features corresponding to the corrected synthetic aperture radar remote sensing data;

[0009] In the process of extracting shared semantic features, the structure and texture features of the synthetic aperture radar remote sensing data are used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network, thereby suppressing modal conflict features and improving boundary recognition capabilities;

[0010] After generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, the confidence areas in the fusion feature space are eliminated and confidence-weighted, and the fusion features are used as input to drive the spatial zoning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the soil erosion distribution pattern and spatiotemporal evolution trend of the monitored landform area.

[0011] On the basis of the above technical solution, preferably, the establishment of a cross-modal physical response consistency mapping relationship based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area specifically includes:

[0012] Based on the ground object classification results of the monitored landform area, a ground object spectral reflectance characteristic model of the optical remote sensing data and a ground object electromagnetic scattering response model of the synthetic aperture radar remote sensing data are respectively constructed to clarify the physical response mechanism of different ground object categories under different remote sensing data;

[0013] Based on the ground object spectral reflectance characteristic model and the ground object electromagnetic scattering response model, a mathematical mapping relationship between the optical remote sensing data and the synthetic aperture radar remote sensing data in terms of physical magnitude, response scale and radiation dynamic range is established, so that the observation responses of the optical remote sensing data and the synthetic aperture radar remote sensing data to the same ground object category are aligned.

[0014] Based on the above technical solution, preferably, the consistency correction processing of the physical magnitude, response scale and radiation dynamic range of the optical remote sensing data and the synthetic aperture radar remote sensing data is performed through a unified physical scale standardization mechanism, specifically including:

[0015] Based on the mathematical mapping relationship, a unified physical scale normalization mechanism is designed to normalize the optical remote sensing data and the synthetic aperture radar remote sensing data to a unified physical dimension and dynamic range, thereby eliminating observation scale deviations caused by differences in sensor characteristics;

[0016] Based on the standardized optical remote sensing data and the standardized synthetic aperture radar remote sensing data, a physical consistency check is performed on each corresponding surface unit of the monitored landform area to ensure that the consistency of the physical response meets the preset threshold requirements, and the data areas that do not meet the consistency requirements are masked. After processing, the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data are obtained respectively.

[0017] On the basis of the above technical solution, preferably, after completing the physical scale normalization process, extracting shared semantic features from the corrected optical remote sensing data and synthetic aperture radar remote sensing data, and using a multi-scale residual structure and a dynamic spatial attention mechanism to perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and synthetic aperture radar remote sensing features, specifically including:

[0018] Performing multi-scale feature extraction on the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, respectively, and extracting the optical remote sensing features and the synthetic aperture radar remote sensing features using convolution kernels of different scales;

[0019] In the process of extracting the optical remote sensing features and the synthetic aperture radar remote sensing features, a multi-scale residual structure is superimposed, and a cross-scale residual connection method is used to retain key boundary information and structural detail information in the features of each scale;

[0020] Based on the optical remote sensing features and the synthetic aperture radar remote sensing features output by the multi-scale residual structure, a dynamic spatial attention mechanism is introduced to generate position-sensitive weight maps for the optical remote sensing feature channel and the synthetic aperture radar remote sensing feature channel respectively, and dynamically weighted modulate the importance of features at different spatial positions;

[0021] Based on the weighted results of the dynamic spatial attention mechanism, the optical remote sensing features and the synthetic aperture radar remote sensing features are enhanced in sensitive areas and suppressed in redundant areas;

[0022] The optical remote sensing features and synthetic aperture radar remote sensing features that have been weighted and adjusted by multi-scale residual structure compensation and dynamic spatial attention mechanism are used as the shared semantic features.

[0023] Based on the above technical solution, preferably, in the process of extracting the shared semantic features, the structure and texture features of the synthetic aperture radar remote sensing data are used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network, thereby suppressing the modal conflict features and improving the boundary recognition capability, which specifically includes:

[0024] Extracting a surface fine-grained texture pattern based on the corrected synthetic aperture radar remote sensing data using structural texture feature extraction to generate a synthetic aperture radar structural texture feature map;

[0025] Extracting an optical remote sensing confusion region feature map based on the corrected optical remote sensing data, wherein the confusion region is a spatial location where boundaries are blurred or object categories overlap;

[0026] Constructing a learning distillation network to guide the learning, using the synthetic aperture radar structure texture feature map as a guidance signal and the optical remote sensing confused region feature map as a learning signal, and adopting a structure preservation loss function and a modal consistency loss function for joint optimization to drive the optical remote sensing confused region feature map to gradually approach the spatial distribution pattern of the synthetic aperture radar structure texture feature map;

[0027] During the guided learning distillation process, the boundary position of the confused region is corrected through local alignment within the distillation network, while modal conflict features and artifact features in the feature map of the optical remote sensing confused region are suppressed, thereby improving spatial boundary clarity and semantic differentiation.

[0028] The optical remote sensing features are corrected by a guided learning distillation network to obtain the processed optical remote sensing features, and the synthetic aperture radar remote sensing features are corrected by a guided learning distillation network to obtain the processed synthetic aperture radar remote sensing features.

[0029] On the basis of the above technical solution, preferably, after generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, the confidence areas in the fusion feature space are eliminated and confidence weighted, and the fusion features are used as input to drive the spatial partitioning extraction of soil and water loss and the quantitative assessment of soil and water loss, thereby realizing the monitoring and assessment of the soil and water loss distribution pattern and spatiotemporal evolution trend of the monitored landform area, specifically including:

[0030] Taking the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features as input, generating the fused features through feature cascade and cross-modal attention fusion mechanism, wherein the fused features include both optical remote sensing feature information and synthetic aperture radar remote sensing feature information;

[0031] Perform confidence prediction on the fusion feature based on a Bayesian deep neural network, and generate a confidence map corresponding to each spatial position in the fusion feature space, wherein the confidence map represents the confidence level of the fusion feature at each position;

[0032] According to the spatial distribution result of the confidence map, a confidence rejection threshold is set, and low confidence areas in the confidence map that are lower than the rejection threshold are masked or zeroed in the fusion feature;

[0033] For a high confidence region in the confidence map that is higher than a confidence weighted threshold, weighted amplification processing is performed on the feature response according to the confidence value thereof to obtain a feature fusion region;

[0034] The fusion features of the feature fusion area are input into the regional connectivity analysis and change detection algorithm to extract the distribution pattern of soil erosion patches, and the soil erosion intensity index is inverted to realize the identification of soil erosion distribution pattern and dynamic monitoring of spatiotemporal evolution trend in the monitored landform area.

[0035] Based on the above technical solution, preferably, after completing the physical scale normalization process, extracting shared semantic features from the corrected optical remote sensing data and synthetic aperture radar remote sensing data further includes:

[0036] The optical remote sensing features and synthetic aperture radar remote sensing features processed by the multi-scale residual structure and dynamic spatial attention mechanism are mapped to the same shared semantic feature space to obtain shared optical remote sensing features and shared radar remote sensing features;

[0037] In the shared semantic feature space, a standardized category center vector is defined for each feature category. By constructing a semantic consistency loss function, the shared semantic features of samples of the same category are constrained to be close to the same category center vector, and the shared semantic features of samples of different categories are constrained to be far away from other category centers.

[0038] Calculating the mutual information between the shared optical remote sensing feature and the shared radar remote sensing feature in a shared semantic feature space, and maximizing the mutual information between the shared optical remote sensing feature and the shared radar remote sensing feature using a cross-modal mutual information maximization mechanism;

[0039] In the process of calculating the mutual information, a semantic consistency loss function and a cross-modal mutual information maximization loss function are jointly optimized, while maintaining a balance between category discrimination and modality preservation of shared semantic features, and optimizing the shared optical remote sensing features and the shared radar remote sensing features respectively;

[0040] The optimized shared optical remote sensing feature is used as the guidance signal, and the optimized shared radar remote sensing feature is used as the learning signal.

[0041] In a second aspect of the present application, a soil and water loss monitoring device using multi-source remote sensing data fusion is provided. The device is used to perform any of the above-described soil and water loss monitoring methods using multi-source remote sensing data fusion. The device includes an acquisition module, a processing module, and an output module, wherein:

[0042] The acquisition module is configured to establish a cross-modal physical response consistency mapping relationship based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, and perform consistency correction processing on the optical remote sensing data and the synthetic aperture radar remote sensing data for physical magnitude, response scale, and radiation dynamic range through a unified physical scale standardization mechanism;

[0043] The processing module is configured to extract shared semantic features from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data after completing the physical scale normalization process, and perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and the synthetic aperture radar remote sensing features using a multi-scale residual structure and a dynamic spatial attention mechanism, wherein the optical remote sensing features are features corresponding to the corrected optical remote sensing data, and the synthetic aperture radar remote sensing features are features corresponding to the corrected synthetic aperture radar remote sensing data;

[0044] The processing module is configured to correct the confusing region features in the optical remote sensing data using the structural texture features of the synthetic aperture radar remote sensing data by guiding the learning distillation network during the process of extracting the shared semantic features, thereby suppressing modal conflict features and improving boundary recognition capability;

[0045] The output module is used to generate fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, and then perform elimination processing and confidence weighting processing on the confidence areas in the fusion feature space, and use the fusion features as input to drive the spatial zoning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the soil erosion distribution pattern and spatiotemporal evolution trend of the monitored landform area.

[0046] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0047] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any one of the methods described above is executed.

[0048] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0049] 1. This application systematically introduces a cross-modal physical response consistency modeling mechanism into the multi-source remote sensing data fusion process. First, based on the feature type classification of the monitored geomorphological area, a feature spectral reflectance characteristic model for optical remote sensing data and a feature electromagnetic scattering response model for synthetic aperture radar remote sensing data are constructed, respectively. A unified mathematical mapping relationship and physical scale standardization mechanism are established to ensure the consistency of the physical magnitude, response scale, and radiation dynamic range of the optical and synthetic aperture radar remote sensing data. On this basis, a multi-scale residual structure and a dynamic spatial attention mechanism are used to extract shared semantic features after structural compensation. A guided learning distillation network is used to correct the optical remote sensing confusing region features using synthetic aperture radar structural texture features. A semantic consistency loss function and a cross-modal mutual information maximization mechanism are further introduced to jointly optimize the spatial distribution of shared semantic features. This ensures the consistency of the fused data at the physical level and strengthens the collaborative expression of features from different modalities at the semantic level. Ultimately, the physical rationality and semantic expression ability of the fused features are significantly improved, enabling high-precision, high-robustness, and high-spatiotemporal resolution dynamic evolution analysis for soil and water loss monitoring tasks.

[0050] 2. Based on the classification of land object types, the land object spectral reflectance characteristic model of optical remote sensing data and the land object electromagnetic scattering response model of synthetic aperture radar remote sensing data were constructed respectively, and a mathematical mapping relationship was established to achieve unified alignment of optical remote sensing data and synthetic aperture radar remote sensing data in physical magnitude, response scale and radiation dynamic range, thereby ensuring the consistency of observations of the same land object category by different remote sensing data sources, and improving the physical rationality of subsequent fusion modeling and the interoperability of cross-modal data.

[0051] 3. Through a unified physical scale standardization mechanism, optical remote sensing data and synthetic aperture radar remote sensing data are normalized to a unified physical scale. After standardization, physical consistency verification and mask processing are performed on each surface unit, eliminating scale deviation and noise interference caused by differences in observation characteristics of different sensors, ensuring high consistency and reliability of input feature data, and providing stable and accurate basic data support for subsequent shared semantic feature extraction and fusion feature generation.

[0052] 4. In the feature extraction process, a multi-scale residual structure and dynamic spatial attention mechanism are introduced to perform multi-scale structure compensation and spatially sensitive weight adjustment on optical remote sensing features and synthetic aperture radar remote sensing features, effectively enhancing the expression ability of ground object boundaries, changing areas and fine-grained texture features, while suppressing redundant and irrelevant regional features, significantly improving the discrimination ability and spatial expression accuracy of shared semantic features.

[0053] 5. By guided learning distillation networks, the features of optical remote sensing confusion areas are corrected using synthetic aperture radar structural texture features as guidance signals, and modal conflict features and artifact features are suppressed during the local alignment process. This effectively improves the discrimination accuracy and spatial consistency of blurred boundary areas in optical remote sensing data, thereby enhancing the expressiveness and stability of shared semantic features in change detection and small-scale surface disturbance recognition.

[0054] 6. After the fusion features are generated, the confidence of the fusion features is predicted through a Bayesian deep neural network. Based on the confidence map, low-confidence areas are eliminated and high-confidence areas are weighted, further improving the reliability and decision weight of the fusion features. At the same time, the fusion features are used to drive the extraction of soil and water loss patch distribution and intensity inversion, achieving the refinement of spatial pattern identification and high-precision analysis of evolution trends in dynamic soil and water loss monitoring.

[0055] 7. By mapping the optical remote sensing features and synthetic aperture radar remote sensing features processed by the multi-scale residual structure and dynamic spatial attention mechanism into a unified shared semantic feature space, and introducing the semantic consistency loss function and the cross-modal mutual information maximization mechanism for joint optimization, the category discrimination and modality retention of the shared semantic features of different modalities are effectively improved, and the coordinated and consistent expression of optical remote sensing features and synthetic aperture radar remote sensing features in the shared space is achieved, laying a solid semantic foundation for guiding learning distillation and fusion feature generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of a method for monitoring soil and water loss by fusing multi-source remote sensing data disclosed in an embodiment of the present application;

[0057] Figure 2 This is a module diagram of a soil and water loss monitoring device that integrates multi-source remote sensing data disclosed in an embodiment of the present application;

[0058] Figure 3 This is a structural diagram of an electronic device disclosed in an embodiment of the present application.

[0059] Explanation of the reference numerals: 201, acquisition module; 202, processing module; 203, output module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0060] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0061] In the description of the embodiments of this application, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a concrete manner.

[0062] In the description of the embodiments of the present application, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0063] Soil and water loss monitoring involves systematically observing the erosion, transport, and deposition of surface soils under the influence of rainfall, runoff, wind, or human disturbance. Combining ground surveys with remote sensing data analysis, it identifies the spatial extent, evolutionary trends, and intensity of soil and water loss, supporting ecological risk assessment and governance decisions. Remote sensing monitoring primarily relies on optical and radar remote sensing measurements. Optical remote sensing captures visible and infrared reflectance to extract vegetation, bare ground, and moisture characteristics, but is subject to climatic interference. Radar remote sensing inverts surface roughness and microtopography through microwave scattering, enabling all-weather observation. The synergistic integration of these two methods is key to improving monitoring accuracy. However, existing multi-source fusion algorithms generally rely on pixel-level stacking and empirical normalization, lacking in-depth modeling of the physical response mechanisms and error characteristics of optical and synthetic aperture radar remote sensing data. This leads to information redundancy, modal conflict, and object recognition errors. There is an urgent need to develop cross-modal physical response consistency modeling and semantic shared feature optimization mechanisms to enhance the physical rationality and semantic accuracy of fused monitoring systems.

[0064] This embodiment discloses a soil and water loss monitoring method based on multi-source remote sensing data fusion. Figure 1 , including the following steps S110-S140:

[0065] S110, based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, establish a cross-modal physical response consistency mapping relationship, and perform consistency correction processing on the physical magnitude, response scale and radiation dynamic range of the optical remote sensing data and synthetic aperture radar remote sensing data through a unified physical scale standardization mechanism.

[0066] The embodiment of the present application discloses a method for monitoring soil and water loss using multi-source remote sensing data fusion, which is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablet computers, wearable devices, and personal computers (PCs), and can also be a background server that runs the method for monitoring soil and water loss using multi-source remote sensing data fusion. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0067] In one possible implementation, a cross-modal physical response consistency mapping relationship is established based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, specifically including: based on the result of the land feature type classification in the monitored landform area, constructing a land feature spectral reflectance characteristic model for optical remote sensing data and a land feature electromagnetic scattering response model for synthetic aperture radar remote sensing data, respectively, to clarify the physical response mechanism of different land feature categories under different remote sensing data; based on the land feature spectral reflectance characteristic model and the land feature electromagnetic scattering response model, establishing a mathematical mapping relationship between optical remote sensing data and synthetic aperture radar remote sensing data in terms of physical magnitude, response scale and radiation dynamic range, so that the observation responses of optical remote sensing data and synthetic aperture radar remote sensing data to the same land feature category are aligned.

[0068] Specifically, first, based on the results of the land feature classification in the monitored landform area, it is necessary to use existing ground survey data, classification samples or historical remote sensing image data, and adopt object-oriented classification methods or supervised learning classification methods to finely divide the surface in the monitored landform area, identify land feature categories including vegetation cover, bare soil, rock outcrops, water bodies, farmland and construction land, and provide a standardized land feature category basis for subsequent modeling. At the same time, it is ensured that the land feature type classification has spatial connectivity, consistency and temporal stability, so as to ensure the accuracy and representativeness of land feature response modeling.

[0069] Subsequently, after completing the classification of land object types, a land object spectral reflectance characteristic model is constructed for each land object category based on the multispectral band reflectance information of optical remote sensing data. This model forms a parameterized description for characterizing the optical remote sensing reflection behavior of different land object categories by statistically analyzing the reflectance mean, standard deviation and spectral shape parameters of samples of each category in different bands. At the same time, auxiliary features such as vegetation index, bare land index and humidity index are introduced to enrich the discrimination of the land object spectral reflectance characteristic model; and based on the backscattering intensity, polarization ratio and coherence parameters of synthetic aperture radar remote sensing data, a land object electromagnetic scattering response model is constructed. This model takes the land object category as a unit, records its backscattering characteristics and change trends under VV polarization, VH polarization or dual polarization combination, and takes into account the incident angle sensitivity and surface roughness modulation effect to form a complete land object electromagnetic scattering response parameter set.

[0070] Next, based on the ground object spectral reflectance characteristic model and the ground object electromagnetic scattering response model, the response curve morphology, physical magnitude range and scale differences of different ground object categories in optical remote sensing data and synthetic aperture radar remote sensing data are analyzed. By constructing a multivariable mapping function, the reflectivity feature space of optical remote sensing data and the scattering intensity feature space of synthetic aperture radar remote sensing data are mathematically mapped, and a conversion relationship between the two in physical magnitude, response scale and radiation dynamic range is established. The mapping function can use polynomial regression, local weighted regression or deep regression network to take into account both nonlinear relationship modeling and high-dimensional feature matching accuracy.

[0071] Finally, based on the mathematical mapping relationship, the optical remote sensing data and synthetic aperture radar remote sensing data in the monitored landform area are respectively subjected to physical magnitude normalization, response scale alignment and radiation dynamic range standardization, so that the physical response characteristics of the same land feature category in the optical remote sensing data and synthetic aperture radar remote sensing data are aligned, ensuring the physical consistency and semantic compatibility in the subsequent shared semantic feature extraction, fusion feature generation and soil and water loss monitoring model input, thereby significantly improving the overall accuracy and stability of the fusion system.

[0072] In one possible implementation, a uniform physical scale standardization mechanism is used to perform consistency correction processing on the physical magnitude, response scale, and radiation dynamic range of optical remote sensing data and synthetic aperture radar remote sensing data. Specifically, the following steps are performed: based on a mathematical mapping relationship, a uniform physical scale standardization mechanism is designed to normalize the optical remote sensing data and synthetic aperture radar remote sensing data to a uniform physical dimension and dynamic range, respectively, to eliminate the observation scale deviation caused by differences in sensor characteristics; based on the standardized optical remote sensing data and the standardized synthetic aperture radar remote sensing data, a physical consistency check is performed on each corresponding surface unit in the monitored landform area to ensure that the consistency of the physical response meets the preset threshold requirements, and mask processing is performed on the data areas that do not meet the consistency requirements. After processing, the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data are obtained respectively.

[0073] Specifically, based on the mathematical mapping relationship between the optical remote sensing data and the synthetic aperture radar remote sensing data established above, a unified physical scale standardization mechanism is designed. By determining a unified physical dimension benchmark and radiation dynamic range benchmark, the reflectivity characteristics of the optical remote sensing data and the backscattering characteristics of the synthetic aperture radar remote sensing data are subjected to physical magnitude standardization. That is, their respective characteristic values are linearly transformed into a set unified physical interval, the reflectivity value of the optical remote sensing data is standardized to the interval of 0 to 1, and the backscattering coefficient of the synthetic aperture radar remote sensing data is standardized to the interval of -1 to 1. At the same time, the scale response curve is normalized according to the mapping relationship, so that the response dynamic range of different remote sensing data to the same ground object category remains consistent, and the dimension deviation and scale distortion caused by different sensor observation mechanisms and calibration standards are completely eliminated.

[0074] After completing the standardization process, for each corresponding surface unit in the monitored geomorphological area, the standardized optical remote sensing data feature vector and the standardized synthetic aperture radar remote sensing data feature vector are extracted, and the similarity index of the two sets of feature vectors in the unified feature space, such as cosine similarity, Euclidean distance or Mahalanobis distance, is calculated and compared with the preset consistency threshold. If the similarity index meets the set threshold requirement, it is considered that the optical remote sensing data and synthetic aperture radar remote sensing data of the surface unit meet the consistency standard in physical response.

[0075] For surface units that do not meet the physical consistency threshold requirements, mask processing operations are performed to mark the corresponding areas as invalid or high-uncertainty areas in subsequent processing flows, to avoid interference of abnormal response data in shared semantic feature extraction and fusion feature generation. At the same time, small missing areas can be repaired based on neighborhood interpolation or local reconstruction strategies to improve overall data continuity and spatial integrity.

[0076] Through the above-mentioned unified physical scale standardization mechanism and physical consistency verification processing, the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data are obtained respectively. These two types of corrected remote sensing data will serve as standard inputs for subsequent multimodal feature extraction, guided learning distillation and fusion feature generation, ensuring that the subsequent soil and water loss monitoring system can carry out high-precision dynamic analysis and evolution trend assessment based on physically consistent basic data.

[0077] S120, after completing the physical scale normalization processing, extracts shared semantic features from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, and uses a multi-scale residual structure and a dynamic spatial attention mechanism to perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and the synthetic aperture radar remote sensing features.

[0078] In one possible implementation, after completing the physical scale normalization processing, the shared semantic features are extracted from the corrected optical remote sensing data and the synthetic aperture radar remote sensing data, and the multi-scale residual structure and the dynamic spatial attention mechanism are used to perform structural compensation and sensitive weight adjustment on the optical remote sensing features and the synthetic aperture radar remote sensing features, specifically including: performing multi-scale feature extraction on the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, extracting the optical remote sensing features and the synthetic aperture radar remote sensing features through convolution kernels of different scales; in the process of extracting the optical remote sensing features and the synthetic aperture radar remote sensing features, superimposing the multi-scale residual structure, and using the cross-scale residual connection mechanism to extract the optical remote sensing features and the synthetic aperture radar remote sensing features. The proposed method retains the key boundary information and structural detail information in the features of each scale; based on the optical remote sensing features and synthetic aperture radar remote sensing features output by the multi-scale residual structure, a dynamic spatial attention mechanism is introduced to generate position-sensitive weight maps for the optical remote sensing feature channels and synthetic aperture radar remote sensing feature channels respectively, and dynamically weighted modulate the importance of features at different spatial positions; based on the weighted results of the dynamic spatial attention mechanism, the sensitive areas of the optical remote sensing features and synthetic aperture radar remote sensing features are enhanced and the redundant areas are suppressed; the optical remote sensing features and synthetic aperture radar remote sensing features after multi-scale residual structure compensation and dynamic spatial attention mechanism weighted adjustment are used as shared semantic features.

[0079] Specifically, the optical remote sensing data and synthetic aperture radar remote sensing data that have undergone physical scale normalization are first input into the multi-scale feature extraction module respectively. The multi-scale feature extraction module adopts a convolution kernel structure with different receptive fields, and independently extracts local fine-grained features and global coarse-grained features at each scale, thereby obtaining optical remote sensing features and synthetic aperture radar remote sensing features at multiple scales to ensure that the change patterns at different spatial scales are fully captured, thereby improving the spatial diversity and structural integrity of feature representation.

[0080] In the process of multi-scale feature extraction, a multi-scale residual structure is superimposed. The multi-scale residual structure introduces a cross-scale residual connection mechanism to jump-connect and superimpose residuals on low-level scale features and high-level scale features, avoiding gradient vanishing and feature information loss caused by the increase in the number of convolution layers during the feature extraction process. At the same time, it effectively retains key boundary detail information and surface microstructure features, and enhances the continuity expression capability of optical remote sensing features and synthetic aperture radar remote sensing features in spatial boundary areas.

[0081] Based on the optical remote sensing features and synthetic aperture radar remote sensing features output by the multi-scale residual structure, a dynamic spatial attention mechanism is introduced. The dynamic spatial attention mechanism generates position-sensitive weight maps for the optical remote sensing feature channels and synthetic aperture radar remote sensing feature channels according to the local feature response intensity and global feature dependency of each spatial position, and dynamically adjusts the importance of different spatial regions in the feature fusion process to adaptively highlight key change areas and suppress redundant irrelevant areas.

[0082] Based on the position-sensitive weight map generated by the dynamic spatial attention mechanism, the optical remote sensing features and synthetic aperture radar remote sensing features are enhanced in sensitive areas and suppressed in redundant areas respectively. The feature response is weighted and amplified in the sensitive areas, and the feature amplitude is appropriately attenuated in the redundant areas, so that the final feature representation is more focused on soil erosion patches, gully erosion channels and change-sensitive areas, significantly improving the feature's perception of small disturbances and structural changes.

[0083] The optical remote sensing features and synthetic aperture radar remote sensing features that have undergone multi-scale residual structure compensation and weighted adjustment using the dynamic spatial attention mechanism are fused at the feature level to form shared semantic features with complete structure, clear boundaries, consistent semantics, and modal collaboration. This lays a high-quality feature foundation for subsequent guided learning of distillation network optimization and fusion feature generation, further improving the spatial resolution and classification accuracy of soil and water loss monitoring tasks.

[0084] S130, in the process of extracting shared semantic features, the structural texture features of synthetic aperture radar remote sensing data are used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network, thereby suppressing modal conflict features and improving boundary recognition capabilities.

[0085] In a possible implementation, in the process of extracting shared semantic features, the structural texture features of synthetic aperture radar remote sensing data are used to correct the features of the confused areas in the optical remote sensing data through a guided learning distillation network, so as to suppress the modal conflict features and improve the boundary recognition ability, specifically including: extracting the fine-grained texture pattern of the surface based on the corrected synthetic aperture radar remote sensing data using the structural texture feature extraction to generate a synthetic aperture radar structural texture feature map; extracting the optical remote sensing confused area feature map based on the corrected optical remote sensing data, wherein the confused area is a spatial location with blurred boundaries or overlapping ground object categories; constructing a guided learning distillation network with the synthetic aperture radar structural texture feature map as the guidance signal, and optical remote sensing is used as the training set. The remote sensing confused area feature map is used as the learning signal, and the structure preservation loss function and the modal consistency loss function are jointly optimized to drive the optical remote sensing confused area feature map to gradually approach the spatial distribution pattern of the synthetic aperture radar structure texture feature map; in the guided learning distillation process, the boundary position of the confused area is corrected through local alignment within the distillation network, and the modal conflict features and artifact features in the optical remote sensing confused area feature map are suppressed, thereby improving the spatial boundary clarity and semantic distinction; the optical remote sensing features are corrected by the guided learning distillation network to obtain the processed optical remote sensing features, and the synthetic aperture radar remote sensing features are corrected by the guided learning distillation network to obtain the processed synthetic aperture radar remote sensing features.

[0086] Specifically, first, based on the synthetic aperture radar remote sensing data that has been normalized for physical scale and verified for physical consistency, a structural texture feature extraction module is adopted. Texture operators such as local variance, structural tensor features, Gabor filter response or small-scale directional gradient histogram are used to extract fine-grained surface texture patterns at different spatial scales, thereby generating a synthetic aperture radar structural texture feature map that can reflect the changes in surface roughness, the morphology of erosion grooves and the continuity of the boundaries of the objects, which serves as a high-confidence spatial structure reference signal to guide the learning stage.

[0087] Subsequently, based on the optical remote sensing data that has undergone physical scale standardization and physical consistency verification, the boundary ambiguity detection algorithm and category overlap analysis method are used to extract the feature map of the optical remote sensing confusion area. The confusion area is defined as the location where the spectral characteristics are unclear and the category discrimination is uncertain at the junction of vegetation coverage, bare land exposure, water body boundaries and human activity interference areas, providing a focus area for subsequent distillation guidance.

[0088] Next, a guided learning distillation network is constructed, using the synthetic aperture radar structure texture feature map as the guiding signal and the optical remote sensing confusion area feature map as the learning signal. During the network training process, a structure preservation loss function is introduced to maintain the consistency of the surface spatial structure. At the same time, a modal consistency loss function is introduced to minimize the difference in spatial distribution patterns between optical remote sensing features and synthetic aperture radar remote sensing features. Through a joint optimization mechanism, the optical remote sensing confusion area features are driven to gradually approach the synthetic aperture radar structure texture feature map, thereby enhancing boundary clarity and the separability of ground object categories.

[0089] Afterwards, during the guided learning distillation process, a local alignment module is set up inside the distillation network. This module performs local similarity calculation and alignment optimization on the local neighborhood of each spatial unit in the optical remote sensing confusion area feature map. The local alignment operation further corrects the boundary position offset and eliminates modal conflict features and artifact features caused by sensor differences or observation errors, thereby effectively improving the spatial boundary clarity and category semantic distinction.

[0090] Finally, the processed optical remote sensing features are obtained by guiding the learning of the distillation network to correct the optical remote sensing features, and the processed synthetic aperture radar remote sensing features are obtained by guiding the learning of the distillation network to correct the synthetic aperture radar remote sensing features. The two serve as the basic input for subsequent fusion feature generation and soil and water loss monitoring tasks, ensuring that the fusion features of multi-source remote sensing data have a unified spatial structure expression and stable physical consistency support.

[0091] In a possible implementation, after completing the physical scale standardization processing, shared semantic features are extracted from the corrected optical remote sensing data and synthetic aperture radar remote sensing data, specifically including: mapping the optical remote sensing features and synthetic aperture radar remote sensing features processed by the multi-scale residual structure and dynamic spatial attention mechanism to the same shared semantic feature space to obtain shared optical remote sensing features and shared radar remote sensing features; in the shared semantic feature space, a standardized category center vector is defined for each ground feature category, and by constructing a semantic consistency loss function, the shared semantic features of samples of the same category are constrained to be close to the same category center vector, and the shared semantic features of samples of different categories are constrained to be close to the same category center vector. The semantic features are far away from the centers of other categories; the mutual information of shared optical remote sensing features and shared radar remote sensing features in the shared semantic feature space is calculated, and the mutual information between shared optical remote sensing features and shared radar remote sensing features is maximized by adopting the cross-modal mutual information maximization mechanism; in the process of calculating the mutual information, the semantic consistency loss function and the cross-modal mutual information maximization loss function are jointly optimized, while maintaining the balance between the category discrimination and modality retention of the shared semantic features, and optimizing the shared optical remote sensing features and the shared radar remote sensing features respectively; the optimized shared optical remote sensing features are used as the guidance signal, and the optimized shared radar remote sensing features are used as the learning signal.

[0092] Specifically, the optical remote sensing data and synthetic aperture radar remote sensing data that have undergone physical scale standardization and physical consistency verification are respectively processed by multi-scale residual structure compensation and dynamic spatial attention mechanism weighted processing, and then input into the shared semantic encoder module. The shared semantic encoder maps the optical remote sensing features and synthetic aperture radar remote sensing features to the same shared semantic feature space through a set of weight-sharing deep feature mapping networks, ensuring that different modal features in the shared semantic feature space have a unified semantic interpretation and similar spatial distribution structure, thereby forming shared optical remote sensing features and shared radar remote sensing features, respectively.

[0093] In the shared semantic feature space, based on the classification results of land feature types in the monitored landform area, a standardized category center vector is defined for each land feature category. Each category center vector represents the ideal position of the category in the shared semantic feature space. By constructing a semantic consistency loss function, an attraction constraint is imposed on the shared optical remote sensing features and shared radar remote sensing features of the same category, so that their feature vectors are close to the corresponding category center vector. At the same time, an exclusion constraint is imposed on features of different categories to ensure that the shared semantic features of different categories maintain a sufficient distinguishing distance in space, thereby improving category discrimination.

[0094] The mutual information of shared optical remote sensing features and shared radar remote sensing features in the shared semantic feature space is calculated separately. The mutual information reflects the size of shared information and the degree of mutual dependence between features of different modalities. The cross-modal mutual information maximization mechanism is adopted to maximize the mutual information between shared optical remote sensing features and shared radar remote sensing features, thereby strengthening the alignment degree and complementary information utilization efficiency of the two modalities in the shared semantic feature space, and enhancing the consistency and collaborative expression ability of modal fusion.

[0095] During the training process, the semantic consistency loss function and the cross-modal mutual information maximization loss function are jointly optimized. While maintaining the aggregation of samples of the same category and the distinguishability of samples of different categories, the modal consistency and complementary information retention capabilities between shared optical remote sensing features and shared radar remote sensing features are improved. Through joint loss optimization, the shared semantic features have both high category discrimination and high modality retention, ensuring the stability and reliability of feature expression in the subsequent fusion feature generation and guided learning distillation process.

[0096] The shared optical remote sensing features after the above joint optimization are used as guidance signals, and the shared radar remote sensing features after the above joint optimization are used as learning signals, and are respectively input into the guided learning distillation network to further correct the confusing regional features in the optical remote sensing data, suppress the modal conflict features, and improve the spatial boundary clarity of the final generated features and the discriminative performance of the soil and water loss monitoring model.

[0097] S140, after generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, performs elimination processing and confidence weighting processing on the confidence areas in the fusion feature space, and uses the fusion features as input to drive the spatial zoning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the soil erosion distribution pattern and spatiotemporal evolution trend in the monitored geomorphic area.

[0098] In a possible implementation, after generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, the confidence areas in the fusion feature space are eliminated and confidence weighted, and the fusion features are used as input to drive the spatial partitioning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the distribution pattern and spatiotemporal evolution trend of soil erosion in the monitored landform area. Specifically, the method includes: taking the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features as input, generating fusion features through feature cascade and cross-modal attention fusion mechanism, wherein the fusion features include both optical remote sensing feature information and synthetic aperture radar remote sensing feature information; confidence evaluation of the fusion features based on Bayesian deep neural network. The confidence map corresponding to each spatial position in the fusion feature space is generated. The confidence map represents the confidence level of the fusion feature at each position. According to the spatial distribution results of the confidence map, the confidence rejection threshold is set, and the low-confidence areas in the confidence map below the rejection threshold are masked or set to zero in the fusion feature. For the high-confidence areas in the confidence map above the confidence weighted threshold, the feature response is weighted amplified according to its confidence value to obtain the feature fusion area. The fusion features of the feature fusion area are input into the regional connectivity analysis and change detection algorithm to extract the distribution pattern of soil and water loss patches, and the soil and water loss intensity index is inverted to realize the identification of soil and water loss distribution pattern and dynamic monitoring of spatiotemporal evolution trend in the monitored landform area.

[0099] Specifically, the optical remote sensing features and synthetic aperture radar remote sensing features after correction processing by the guided learning distillation network are respectively input into the feature fusion module. The feature fusion module adopts feature cascade and cross-modal attention fusion mechanism. By splicing on the feature dimension and introducing the cross-modal attention matrix, the fusion weights of the optical remote sensing features and the synthetic aperture radar remote sensing features are dynamically adjusted to generate fusion features containing optical remote sensing feature information and synthetic aperture radar remote sensing feature information, thereby realizing the coordinated enhancement of modal information and the extraction of complementary features, and forming a complete and unified fusion feature expression.

[0100] The confidence prediction of the fusion features is performed based on the Bayesian deep neural network. The Bayesian deep neural network generates a confidence map corresponding to each spatial position in the fusion feature space by performing multiple forward inferences on the fusion features and analyzing the mean and variance of the predicted output. The confidence map quantifies the reliability and uncertainty of the fusion features at different spatial positions, providing a basis for subsequent confidence-guided feature optimization.

[0101] According to the spatial distribution results of the confidence map, a confidence rejection threshold is set, and masking or zeroing is performed on the low-confidence areas in the confidence map that are lower than the rejection threshold in the fusion feature. By eliminating the low-confidence areas, abnormal feature responses caused by observation errors, occlusion effects or modal conflicts can be effectively eliminated, thereby improving the overall robustness and noise suppression capabilities of the fusion feature.

[0102] For high-confidence areas in the confidence map that are higher than the confidence weighted threshold, the fused feature response is weightedly amplified according to the confidence value of each spatial position to strengthen the feature expression of the high-confidence area, so that it has higher weight and decision-making influence in the subsequent soil and water loss spatial zoning extraction and soil and water loss quantitative assessment process, forming a feature fusion area to ensure the physical reliability and spatial accuracy of the extraction results.

[0103] Taking the fusion features of the feature fusion area as input, we first apply the regional connectivity analysis and change detection algorithm to extract the distribution pattern of soil erosion patches, and determine the soil erosion units by analyzing the morphological characteristics, change trends and texture structures of the connected areas in the fusion features. Then, combined with the soil erosion intensity inversion model, we invert the soil erosion intensity index based on the fusion features and external auxiliary factors (such as slope, rainfall intensity, etc.), so as to achieve accurate identification of the soil erosion distribution pattern in the monitored geomorphological area and dynamic monitoring and quantitative evaluation of the spatiotemporal evolution trend, thus supporting watershed management and ecological restoration decision-making.

[0104] In one possible implementation, first, the fusion features of the feature fusion area are taken as input, and the regional connectivity analysis module is applied to extract the connected areas in the fusion features based on the spatial adjacency relationship and pixel feature similarity. The regional connectivity analysis module adopts the eight-neighborhood connectivity judgment rule and combines the connected component labeling algorithm to identify continuous and consistent soil erosion candidate patches in the fusion feature space, providing a basic spatial unit for subsequent change detection and soil erosion unit determination.

[0105] Subsequently, based on the connected areas obtained, the change detection module was applied to compare the multi-temporal fusion features of the monitored geomorphic area and extract the changed areas. By calculating the change amplitude, change rate and change direction of the fusion features in the time series, the connected areas with significant dynamic evolution characteristics were screened out to ensure that the soil and water loss extraction process not only relies on static spatial distribution, but also captures dynamic evolution trends, thereby improving the ability to identify loss-sensitive areas.

[0106] Next, for the candidate connected areas of soil and water loss determined after change detection, the morphological characteristics of each connected area are extracted, including indicators such as area, perimeter, shape index, compactness, fractal dimension, etc. At the same time, texture structure characteristics are extracted, such as local variance, energy, entropy and homogeneity indicators. By comprehensively analyzing the morphological characteristics and texture structure characteristics, it is determined whether each connected area meets the spatial morphological characteristic standards of the soil and water loss unit, and noise patches or non-erosive change areas are eliminated.

[0107] Afterwards, the connected areas determined to be soil erosion units are used as the results of soil erosion spatial zoning, and the soil erosion intensity is further quantified based on the inversion model. The soil erosion intensity inversion model uses fusion features as the main input, combined with auxiliary factors such as slope, aspect, rainfall intensity, vegetation cover index, soil type, etc., through weighted regression model, machine learning regression model or process model based on modified RUSLE equation to calculate the soil loss index per unit area and form a spatial distribution map of soil erosion intensity.

[0108] Finally, the distribution pattern of soil erosion patches and soil erosion intensity indicators are superimposed to construct a comprehensive evolution map of soil erosion in the monitored geomorphological area. Through time series analysis, the expansion trend, evolution rate and potential risk areas of the erosion units are evaluated to support the optimization decision-making of watershed management planning, ecological restoration design and soil and water conservation measures, and ensure that the monitoring system has spatial accuracy, dynamic responsiveness and application guidance.

[0109] This embodiment also discloses a soil and water loss monitoring device that integrates multi-source remote sensing data. Figure 2 , comprising an acquisition module 201, a processing module 202 and an output module 203, the device is used to execute any of the above-mentioned soil and water loss monitoring methods using multi-source remote sensing data fusion, wherein:

[0110] The acquisition module 201 is used to establish a cross-modal physical response consistency mapping relationship based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, and to perform consistency correction processing on the physical magnitude, response scale and radiation dynamic range of the optical remote sensing data and the synthetic aperture radar remote sensing data through a unified physical scale standardization mechanism.

[0111] The processing module 202 is used to extract shared semantic features from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data after completing the physical scale normalization processing, and use a multi-scale residual structure and a dynamic spatial attention mechanism to perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and the synthetic aperture radar remote sensing features, wherein the optical remote sensing features are features corresponding to the corrected optical remote sensing data, and the synthetic aperture radar remote sensing features are features corresponding to the corrected synthetic aperture radar remote sensing data.

[0112] The processing module 202 is used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network to use the structural texture features of the synthetic aperture radar remote sensing data in the process of extracting shared semantic features, thereby suppressing the modal conflict features and improving the boundary recognition ability.

[0113] The output module 203 is used to generate fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, and then perform elimination processing and confidence weighting processing on the confidence areas in the fusion feature space, and use the fusion features as input to drive the spatial zoning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the soil erosion distribution pattern and spatiotemporal evolution trend of the monitored geomorphic area.

[0114] In one possible embodiment, the processing module 202 is used to construct a ground object spectral reflectance characteristic model for optical remote sensing data and a ground object electromagnetic scattering response model for synthetic aperture radar remote sensing data based on the ground object type classification results of the monitored landform area, so as to clarify the physical response mechanism of different ground object categories under different remote sensing data.

[0115] Processing module 202 is used to establish a mathematical mapping relationship between optical remote sensing data and synthetic aperture radar remote sensing data in terms of physical magnitude, response scale and radiation dynamic range based on the ground object spectral reflectance characteristic model and the ground object electromagnetic scattering response model, so that the observation responses of optical remote sensing data and synthetic aperture radar remote sensing data to the same ground object category are aligned.

[0116] In one possible implementation, the processing module 202 is used to design a unified physical scale normalization mechanism based on a mathematical mapping relationship, normalize the optical remote sensing data and the synthetic aperture radar remote sensing data to a unified physical dimension and dynamic range, and eliminate the observation scale deviation caused by differences in sensor characteristics.

[0117] The output module 203 is used to perform physical consistency verification on each corresponding surface unit in the monitored landform area based on the standardized optical remote sensing data and the standardized synthetic aperture radar remote sensing data to ensure that the consistency of the physical response meets the preset threshold requirements, and to mask the data areas that do not meet the consistency requirements. After processing, the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data are obtained respectively.

[0118] In a possible implementation, the acquisition module 201 is used to perform multi-scale feature extraction on the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, respectively, and extract the optical remote sensing features and the synthetic aperture radar remote sensing features through convolution kernels of different scales.

[0119] The processing module 202 is used to superimpose multi-scale residual structures in the process of extracting optical remote sensing features and synthetic aperture radar remote sensing features, and use a cross-scale residual connection method to retain key boundary information and structural detail information in features of each scale.

[0120] Processing module 202 is used to introduce a dynamic spatial attention mechanism based on the optical remote sensing features and synthetic aperture radar remote sensing features output by the multi-scale residual structure, generate position-sensitive weight maps for the optical remote sensing feature channel and the synthetic aperture radar remote sensing feature channel respectively, and dynamically weight the importance of features at different spatial positions.

[0121] The processing module 202 is used to enhance the sensitive areas and suppress the redundant areas of the optical remote sensing features and the synthetic aperture radar remote sensing features based on the weighted results of the dynamic spatial attention mechanism.

[0122] The processing module 202 is configured to use the optical remote sensing features and the synthetic aperture radar remote sensing features that have undergone multi-scale residual structure compensation and weighted adjustment using a dynamic spatial attention mechanism as shared semantic features.

[0123] In a possible implementation, the acquisition module 201 is configured to extract a surface fine-grained texture pattern using structural texture feature extraction based on the corrected synthetic aperture radar remote sensing data, and generate a synthetic aperture radar structural texture feature map.

[0124] The acquisition module 201 is used to extract an optical remote sensing confusion region feature map based on the corrected optical remote sensing data, wherein the confusion region is a spatial location with blurred boundaries or overlapping ground object categories.

[0125] Processing module 202 is used to construct a guided learning distillation network, using the synthetic aperture radar structure texture feature map as a guidance signal and the optical remote sensing confusion area feature map as a learning signal. It adopts a joint optimization of the structure preservation loss function and the modal consistency loss function to drive the optical remote sensing confusion area feature map to gradually approach the spatial distribution pattern of the synthetic aperture radar structure texture feature map.

[0126] Processing module 202 is used to correct the boundary positions of the confused regions through local alignment within the distillation network during the guided learning distillation process, while suppressing modal conflict and artifact features in the optical remote sensing feature map of the confused regions, thereby improving spatial boundary clarity and semantic distinction. The optical remote sensing features are corrected by the guided learning distillation network to obtain processed optical remote sensing features, and the synthetic aperture radar remote sensing features are corrected by the guided learning distillation network to obtain processed synthetic aperture radar features.

[0127] In one possible implementation, the processing module 202 is used to take the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features as input, and generate fused features through feature cascade and cross-modal attention fusion mechanism, where the fused features include both optical remote sensing feature information and synthetic aperture radar remote sensing feature information.

[0128] The processing module 202 is used to perform confidence prediction on the fused features based on the Bayesian deep neural network, and generate a confidence map corresponding to each spatial position in the fused feature space, where the confidence map represents the confidence level of the fused features at each position.

[0129] The processing module 202 is used to set a confidence rejection threshold according to the spatial distribution result of the confidence map, and perform mask processing or zero processing on low-confidence areas in the confidence map that are lower than the rejection threshold in the fusion feature.

[0130] The processing module 202 is configured to perform weighted amplification processing on the feature response of the high confidence region in the confidence map that is higher than the confidence weighted threshold according to its confidence value to obtain a feature fusion region.

[0131] The processing module 202 is used to input the fusion features of the fusion area to extract the distribution pattern of soil erosion patches based on regional connectivity analysis and change detection algorithm, and to invert the soil erosion intensity index to realize the identification of soil erosion distribution pattern and dynamic monitoring of spatiotemporal evolution trend in the monitored landform area.

[0132] In one possible implementation, the processing module 202 is used to map the optical remote sensing features and synthetic aperture radar remote sensing features processed by the multi-scale residual structure and the dynamic spatial attention mechanism to the same shared semantic feature space to obtain shared optical remote sensing features and shared radar remote sensing features.

[0133] Processing module 202 is used to define a standardized category center vector for each feature category in the shared semantic feature space, and to constrain the shared semantic features of samples of the same category to be close to the same category center vector, and the shared semantic features of samples of different categories to be far away from other category centers by constructing a semantic consistency loss function.

[0134] The processing module 202 is used to calculate the mutual information between the shared optical remote sensing features and the shared radar remote sensing features in the shared semantic feature space, and adopt a cross-modal mutual information maximization mechanism to maximize the mutual information between the shared optical remote sensing features and the shared radar remote sensing features.

[0135] Processing module 202 is used to jointly optimize the semantic consistency loss function and the cross-modal mutual information maximization loss function in the process of calculating the mutual information, while maintaining the balance between the shared semantic features in terms of category discrimination and modality retention, and optimizing the shared optical remote sensing features and the shared radar remote sensing features respectively.

[0136] The processing module 202 is configured to use the optimized shared optical remote sensing features as guidance signals and the optimized shared radar remote sensing features as learning signals.

[0137] It should be noted that the above embodiments provide devices that implement their functions using only the division of the above functional modules as examples. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0138] This embodiment also discloses an electronic device, referring to Figure 3 The electronic device may include: at least one processor 301 , at least one communication bus 302 , a user interface 303 , a network interface 304 , and at least one memory 305 .

[0139] The communication bus 302 is used to implement the connection and communication between these components.

[0140] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0141] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0142] The processor 301 may include one or more processing cores. The processor 301 utilizes various interfaces and lines to connect various parts of the entire server. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and calling data stored in the memory 305, the processor 301 performs various server functions and processes data. Optionally, the processor 301 may be implemented in at least one hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 301 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display; and the modem is used to handle wireless communications. It is understood that the modem may not be integrated into the processor 301 and may be implemented separately on a single chip.

[0143] The memory 305 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 305 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store data involved in the above-mentioned method embodiments, etc. The memory 305 may also optionally be at least one storage device located away from the aforementioned processor 301. The memory 305, as a computer storage medium, may include an operating system, a network communication module, a user interface 303 module, and an application program for a soil and water loss monitoring method that integrates multi-source remote sensing data.

[0144] exist Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 301 can be used to call an application program stored in the memory 305 for a soil and water loss monitoring method that integrates multi-source remote sensing data. When executed by one or more processors 301, the electronic device executes one or more methods in the above-mentioned embodiments.

[0145] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, 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 preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0146] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0148] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0149] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0150] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory 305 and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory 305 includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disk.

[0151] The present application also discloses a computer-readable storage medium storing instructions, which, when executed by one or more processors 301 , enable an electronic device to execute one or more methods in the above embodiments.

[0152] The above are merely exemplary embodiments of the present disclosure and are not intended to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A soil and water loss monitoring method based on multi-source remote sensing data fusion, characterized in that: The method comprises: Based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, a cross-modal physical response consistency mapping relationship is established, and the optical remote sensing data and the synthetic aperture radar remote sensing data are corrected for consistency in physical magnitude, response scale, and radiation dynamic range through a unified physical scale standardization mechanism; After completing the physical scale normalization process, shared semantic features are extracted from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, and structural compensation and sensitivity weight adjustment are performed on the optical remote sensing features and the synthetic aperture radar remote sensing features using a multi-scale residual structure and a dynamic spatial attention mechanism, wherein the optical remote sensing features are features corresponding to the corrected optical remote sensing data, and the synthetic aperture radar remote sensing features are features corresponding to the corrected synthetic aperture radar remote sensing data; In the process of extracting shared semantic features, the structure and texture features of the synthetic aperture radar remote sensing data are used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network, thereby suppressing modal conflict features and improving boundary recognition capabilities; After generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, the confidence areas in the fusion feature space are eliminated and confidence-weighted, and the fusion features are used as input to drive the spatial zoning extraction of soil erosion and the quantitative assessment of soil erosion, thereby realizing the monitoring and assessment of the soil erosion distribution pattern and spatiotemporal evolution trend of the monitored landform area.

2. The method for monitoring soil and water loss by fusion of multi-source remote sensing data according to claim 1, characterized in that: The establishment of a cross-modal physical response consistency mapping relationship based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area specifically includes: Based on the ground object classification results of the monitored landform area, a ground object spectral reflectance characteristic model of the optical remote sensing data and a ground object electromagnetic scattering response model of the synthetic aperture radar remote sensing data are respectively constructed to clarify the physical response mechanism of different ground object categories under different remote sensing data; Based on the ground object spectral reflectance characteristic model and the ground object electromagnetic scattering response model, a mathematical mapping relationship between the optical remote sensing data and the synthetic aperture radar remote sensing data in terms of physical magnitude, response scale and radiation dynamic range is established, so that the observation responses of the optical remote sensing data and the synthetic aperture radar remote sensing data to the same ground object category are aligned.

3. The method for soil and water loss monitoring based on multi-source remote sensing data fusion according to claim 2, characterized in that: The consistency correction processing of the physical magnitude, response scale and radiation dynamic range of the optical remote sensing data and the synthetic aperture radar remote sensing data is performed through a unified physical scale standardization mechanism, specifically including: Based on the mathematical mapping relationship, a unified physical scale normalization mechanism is designed to normalize the optical remote sensing data and the synthetic aperture radar remote sensing data to a unified physical dimension and dynamic range, thereby eliminating observation scale deviations caused by differences in sensor characteristics; Based on the standardized optical remote sensing data and the standardized synthetic aperture radar remote sensing data, a physical consistency check is performed on each corresponding surface unit of the monitored landform area to ensure that the consistency of the physical response meets the preset threshold requirements, and the data areas that do not meet the consistency requirements are masked. After processing, the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data are obtained respectively.

4. The method for soil and water loss monitoring based on multi-source remote sensing data fusion according to claim 1, characterized in that: After completing the physical scale normalization process, the shared semantic features are extracted from the corrected optical remote sensing data and synthetic aperture radar remote sensing data, and the multi-scale residual structure and dynamic spatial attention mechanism are used to perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and synthetic aperture radar remote sensing features, specifically including: Performing multi-scale feature extraction on the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data, respectively, and extracting the optical remote sensing features and the synthetic aperture radar remote sensing features using convolution kernels of different scales; In the process of extracting the optical remote sensing features and the synthetic aperture radar remote sensing features, a multi-scale residual structure is superimposed, and a cross-scale residual connection method is used to retain key boundary information and structural detail information in the features of each scale; Based on the optical remote sensing features and the synthetic aperture radar remote sensing features output by the multi-scale residual structure, a dynamic spatial attention mechanism is introduced to generate position-sensitive weight maps for the optical remote sensing feature channel and the synthetic aperture radar remote sensing feature channel respectively, and dynamically weighted modulate the importance of features at different spatial positions; Based on the weighted results of the dynamic spatial attention mechanism, the optical remote sensing features and the synthetic aperture radar remote sensing features are enhanced in sensitive areas and suppressed in redundant areas; The optical remote sensing features and synthetic aperture radar remote sensing features that have undergone multi-scale residual structure compensation and dynamic spatial attention mechanism weighted adjustment are used as the shared semantic features.

5. The method for soil and water loss monitoring based on multi-source remote sensing data fusion according to claim 1, characterized in that: In the process of extracting the shared semantic features, the method further comprises: guiding the learning distillation network to correct the confusing region features in the optical remote sensing data using the structural texture features of the synthetic aperture radar remote sensing data, thereby suppressing the modal conflict features and improving the boundary recognition capability. Extracting a surface fine-grained texture pattern based on the corrected synthetic aperture radar remote sensing data using structural texture feature extraction to generate a synthetic aperture radar structural texture feature map; Extracting an optical remote sensing confusion region feature map based on the corrected optical remote sensing data, wherein the confusion region is a spatial location where boundaries are blurred or object categories overlap; Constructing a learning distillation network to guide the learning, using the synthetic aperture radar structure texture feature map as a guidance signal and the optical remote sensing confused region feature map as a learning signal, and adopting a structure preservation loss function and a modal consistency loss function for joint optimization to drive the optical remote sensing confused region feature map to gradually approach the spatial distribution pattern of the synthetic aperture radar structure texture feature map; During the guided learning distillation process, the boundary position of the confused region is corrected through local alignment within the distillation network, while modal conflict features and artifact features in the feature map of the optical remote sensing confused region are suppressed, thereby improving spatial boundary clarity and semantic differentiation. The optical remote sensing features are corrected by a guided learning distillation network to obtain the processed optical remote sensing features, and the synthetic aperture radar remote sensing features are corrected by a guided learning distillation network to obtain the processed synthetic aperture radar remote sensing features.

6. The method for soil and water loss monitoring based on multi-source remote sensing data fusion according to claim 1, characterized in that: After generating fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, the confidence regions in the fusion feature space are eliminated and confidence weighted, and the fusion features are used as input to drive the spatial partitioning extraction of soil and water loss and the quantitative assessment of soil and water loss, thereby achieving the monitoring and assessment of the soil and water loss distribution pattern and spatiotemporal evolution trend of the monitored landform area, specifically including: Taking the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features as input, generating the fused features through feature cascade and cross-modal attention fusion mechanism, wherein the fused features include both optical remote sensing feature information and synthetic aperture radar remote sensing feature information; Perform confidence prediction on the fusion feature based on a Bayesian deep neural network, and generate a confidence map corresponding to each spatial position in the fusion feature space, wherein the confidence map represents the confidence level of the fusion feature at each position; According to the spatial distribution result of the confidence map, a confidence rejection threshold is set, and low confidence areas in the confidence map that are lower than the rejection threshold are masked or zeroed in the fusion feature; For a high confidence region in the confidence map that is higher than a confidence weighted threshold, weighted amplification processing is performed on the feature response according to the confidence value thereof to obtain a feature fusion region; The fusion features of the feature fusion area are input into the regional connectivity analysis and change detection algorithm to extract the distribution pattern of soil erosion patches, and the soil erosion intensity index is inverted to realize the identification of soil erosion distribution pattern and dynamic monitoring of spatiotemporal evolution trend in the monitored landform area.

7. The method for soil and water loss monitoring based on multi-source remote sensing data fusion according to claim 5, characterized in that: After completing the physical scale normalization process, extracting shared semantic features from the corrected optical remote sensing data and synthetic aperture radar remote sensing data, specifically including: The optical remote sensing features and synthetic aperture radar remote sensing features processed by the multi-scale residual structure and dynamic spatial attention mechanism are mapped to the same shared semantic feature space to obtain shared optical remote sensing features and shared radar remote sensing features; In the shared semantic feature space, a standardized category center vector is defined for each feature category. By constructing a semantic consistency loss function, the shared semantic features of samples of the same category are constrained to be close to the same category center vector, and the shared semantic features of samples of different categories are constrained to be far away from other category centers. Calculating the mutual information between the shared optical remote sensing feature and the shared radar remote sensing feature in a shared semantic feature space, and maximizing the mutual information between the shared optical remote sensing feature and the shared radar remote sensing feature using a cross-modal mutual information maximization mechanism; In the process of calculating the mutual information, a semantic consistency loss function and a cross-modal mutual information maximization loss function are jointly optimized, while maintaining a balance between category discrimination and modality preservation of shared semantic features, and optimizing the shared optical remote sensing features and the shared radar remote sensing features respectively; The optimized shared optical remote sensing feature is used as the guidance signal, and the optimized shared radar remote sensing feature is used as the learning signal.

8. A soil and water loss monitoring device based on multi-source remote sensing data fusion, characterized in that: The device is used to execute a soil and water loss monitoring method based on multi-source remote sensing data fusion according to any one of claims 1 to 7, and the device comprises an acquisition module (201), a processing module (202) and an output module (203), wherein: The acquisition module (201) is used to establish a cross-modal physical response consistency mapping relationship based on the collected optical remote sensing data and synthetic aperture radar data for the monitored landform area, and perform consistency correction processing on the physical magnitude, response scale and radiation dynamic range of the optical remote sensing data and the synthetic aperture radar remote sensing data through a unified physical scale standardization mechanism; The processing module (202) is used to extract shared semantic features from the corrected optical remote sensing data and the corrected synthetic aperture radar remote sensing data after completing the physical scale normalization process, and use a multi-scale residual structure and a dynamic spatial attention mechanism to perform structural compensation and sensitivity weight adjustment on the optical remote sensing features and the synthetic aperture radar remote sensing features, wherein the optical remote sensing features are features corresponding to the corrected optical remote sensing data, and the synthetic aperture radar remote sensing features are features corresponding to the corrected synthetic aperture radar remote sensing data; The processing module (202) is used to correct the confusing area features in the optical remote sensing data by guiding the learning distillation network in the process of extracting the shared semantic features, using the structural texture features of the synthetic aperture radar remote sensing data, thereby suppressing modal conflict features and improving boundary recognition capabilities; The output module (203) is used to generate fusion features based on the processed optical remote sensing features and the processed synthetic aperture radar remote sensing features, and then perform a removal process and a confidence weighting process on the confidence areas in the fusion feature space, and use the fusion features as input to drive the spatial partitioning extraction of soil and water loss and the quantitative assessment of soil and water loss, thereby realizing the monitoring and assessment of the soil and water loss distribution pattern and spatiotemporal evolution trend of the monitored landform area.

9. An electronic device, characterized in that: The electronic device comprises a processor (301), a communication bus (302), a user interface (303), a network interface (304) and a memory (305), wherein the memory (305) is used to store instructions, the user interface (303) and the network interface (304) are both used to communicate with other devices, the communication bus (302) is used to realize connection and communication between components in the electronic device, and the processor (301) is used to execute the instructions stored in the memory (305) so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is executed.

Images