Method and device for monitoring aquatic vegetation types based on optical and sar satellite imagery

Abstract

The application relates to the technical field of water environment remote sensing monitoring, and provides a water vegetation type monitoring method and device based on optical and SAR satellite images. The method comprises the following steps: acquiring optical remote sensing image data and radar image data of the same water area after spatiotemporal registration; acquiring optical image band data, near-infrared reflectivity and red light band reflectivity of each pixel point from the optical remote sensing image data; selecting sub-channel features corresponding to the pixel points from cross-polarization channel feature data; obtaining normalized vegetation indexes of each pixel point based on the near-infrared reflectivity and the red light band reflectivity; obtaining classification results corresponding to each pixel point based on the normalized vegetation indexes, the optical image band data and the sub-channel features, so as to determine all water vegetation types in the water area. The method realizes large-range, high-frequency and high-precision monitoring of water vegetation types in large shallow lakes.

Description

Technical Field

[0001] This application relates to the field of water environment remote sensing monitoring technology, and in particular to a method and device for monitoring aquatic vegetation types based on optical and SAR (syntheticaperture radar) satellite imagery. Background Technology

[0002] Aquatic vegetation is a key indicator of the health of lake ecosystems. Its type and distribution dynamics directly reflect the nutrient status and ecological function of water bodies. Conducting large-scale, periodic, and precise monitoring of large shallow lakes such as Taihu Lake is of great significance for the treatment of lake eutrophication, water quality management, and ecological assessment.

[0003] Current mainstream monitoring methods have significant limitations: traditional field surveys, while capable of obtaining high-precision, fixed-point information, are time-consuming, labor-intensive, and have limited coverage, making simultaneous and rapid monitoring difficult; single optical remote sensing can perform large-scale periodic monitoring, but is susceptible to cloud and fog conditions leading to data loss, and suffers from issues such as "different objects with the same spectrum" and "same objects with different spectra," resulting in insufficient classification accuracy; synthetic aperture radar remote sensing can provide all-weather observation and information on ground structure, but lacks red, green, and blue optical information, making it impossible to accurately identify aquatic vegetation types independently. Existing methods cannot simultaneously meet the needs of large-scale, high-precision, and all-weather monitoring, necessitating the development of new monitoring technologies to overcome these bottlenecks. Summary of the Invention

[0004] In view of this, this application provides a method and device for monitoring aquatic vegetation types based on optical and SAR satellite imagery, which can realize large-scale, high-precision, all-weather monitoring of aquatic vegetation types.

[0005] This application provides a method and apparatus for monitoring aquatic vegetation types based on optical and SAR satellite imagery. The following describes this application from multiple aspects, and the embodiments and beneficial effects described below can be referenced interchangeably.

[0006] Firstly, this application provides a method for monitoring aquatic vegetation types based on optical and SAR satellite imagery, including:

[0007] Acquire optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered, wherein the radar image data includes cross-polarization channel feature data;

[0008] Optical data and optical image band data for each pixel are obtained from optical remote sensing image data. The optical data includes near-infrared reflectance and red band reflectance.

[0009] Select sub-channel features corresponding to pixels from the cross-polarization channel feature data;

[0010] Based on optical data, the normalized vegetation index for each pixel is obtained;

[0011] Based on the normalized vegetation index, optical image band data, and sub-channel features, classification results corresponding to each pixel are obtained to determine all aquatic vegetation types in the water area.

[0012] According to the embodiments of this application, the above-described technical solution of this application has at least the following beneficial effects:

[0013] By acquiring spatiotemporally registered optical remote sensing image data and radar image data containing cross-polarization channel feature data, optical data such as near-infrared reflectance and red band reflectance of pixels, as well as corresponding sub-channel features, are extracted. Combined with the calculated normalized vegetation index, multi-dimensional feature fusion and pixel classification are performed, integrating the spectral features of optical data with the land cover structure information of radar data, effectively solving the problem of different objects with the same spectrum and similar objects with different spectra. Furthermore, the all-weather, all-time observation capability of radar data compensates for the weather constraints of optical remote sensing. Simultaneously, it overcomes the shortcomings of traditional field surveys, such as being time-consuming, labor-intensive, and having insufficient coverage, enabling large-scale, high-frequency, and high-precision monitoring of aquatic vegetation types in large shallow lakes such as Taihu Lake, providing efficient and reliable data support for lake eutrophication control, water quality management, and ecological assessment.

[0014] In one possible implementation of the first aspect above, selecting sub-channel features corresponding to pixels from the cross-polarization channel feature data includes:

[0015] The cross-polarization channel feature data is preprocessed to obtain preprocessed channel feature data. The preprocessing includes at least one of the following: orbit correction, thermal noise removal, radiometric calibration, speckle filtering, and terrain correction.

[0016] From the preprocessed channel feature data, select the sub-channel features corresponding to the pixels.

[0017] According to the embodiments of this application, noise in the cross-polarization channel feature data can be effectively reduced by preprocessing the cross-polarization channel feature data.

[0018] In one possible implementation of the first aspect above, selecting the sub-channel feature corresponding to the pixel from the cross-polarization channel feature data further includes:

[0019] Nonlinear mathematical transformations are performed on the cross-polarization channel feature data to obtain enhanced cross-polarization channel feature data. The nonlinear mathematical transformations include at least one of logarithmic transformation and square root transformation.

[0020] Sub-channel features are selected from the enhanced cross-polarization channel feature data.

[0021] According to the embodiments of this application, by performing nonlinear mathematical transformation on the cross-polarization channel feature data, the data distribution of the features can be adjusted, the dynamic range can be compressed, and the subtle differences between features can be amplified, effectively alleviating the problem of insufficient discrimination of a single radar feature, thereby improving the precision and accuracy of aquatic vegetation type classification.

[0022] In one possible implementation of the first aspect described above, based on the normalized vegetation index, optical image band data, and sub-channel features, a classification result corresponding to each pixel is obtained, including:

[0023] Given that the normalized vegetation index is greater than a preset threshold, a high-dimensional feature vector is constructed based on optical image band data and sub-channel features.

[0024] High-dimensional feature vectors are used as input data for a feature-level combination classification model to obtain classification results.

[0025] According to the embodiments of this application, since the vegetation spectrum and spatial height are similar in high-density aquatic plant areas, constructing a high-dimensional feature vector can fuse multi-source difference information, amplify subtle feature differences, and improve classification accuracy.

[0026] In one possible implementation of the first aspect described above, the classification result corresponding to each pixel is obtained based on the normalized vegetation index, optical image band data, and sub-channel features, and further includes:

[0027] Under the condition that the normalized vegetation index is not greater than the preset threshold, the optical image band data is used as the input data of the optical feature classification model to obtain the first classification probability vector.

[0028] The sub-channel features are used as input data for the radar image feature classification model to obtain the second classification probability vector;

[0029] Based on preset weight coefficients, the first classification probability vector and the second classification probability vector are fused to obtain a fused vector;

[0030] Determine the maximum probability value of the fusion vector, and use the category corresponding to the maximum probability value as the classification result.

[0031] According to the implementation method of this application, for low vegetation cover areas, the probability vectors of optical and radar classification models are fused at the decision level. This not only leverages the spectral advantages of optics in distinguishing non-vegetation and low-coverage land features, but also makes up for the shortcomings of optics in identifying weak vegetation signals. This effectively avoids misclassification and omission in sparse vegetation areas and ensures the stable and accurate classification effect of the overall monitoring method in the whole area and under different vegetation densities.

[0032] In one possible implementation of the first aspect above, the high-dimensional feature vector is used as input data for a feature-level combination classification model to obtain classification results, including:

[0033] Standardize the high-dimensional feature vector to obtain a standardized high-dimensional feature vector. Standardization includes at least one of mean normalization and standard deviation.

[0034] The standardized high-dimensional feature vectors are used as input data for the feature-level combination classification model to obtain the classification results.

[0035] According to the embodiments of this application, by standardizing the high-dimensional feature vectors, model bias caused by differences in physical meaning and numerical range among the feature dimensions is eliminated.

[0036] In one possible implementation of the first aspect described above, optical image band data is used as input data for an optical feature classification model to obtain a first classification probability vector, including:

[0037] Standardize optical image band data to obtain standardized optical image band data. Standardization includes at least one of mean and standard deviation.

[0038] Standardized optical image band data are used as input data for the optical feature classification model to obtain the first classification probability vector;

[0039] Using the sub-channel features as input data for the radar image feature classification model, a second classification probability vector is obtained, including:

[0040] The sub-channel features are standardized to obtain standardized sub-channel features;

[0041] The standardized sub-channel features are used as input data for the radar image feature classification model to obtain the second classification probability vector.

[0042] According to the embodiments of this application, by standardizing the optical image band data and sub-channel features respectively, the differences in dimensions and numerical ranges between optical and radar data are eliminated, thereby improving the accuracy of model recognition.

[0043] Secondly, this application provides a device for monitoring aquatic vegetation types based on optical and SAR satellite imagery, comprising:

[0044] The data acquisition module is used to acquire optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered, wherein the radar image data includes cross-polarization channel feature data.

[0045] The remote sensing image processing module is used to obtain optical data and optical image band data of each pixel from the optical remote sensing image data, wherein the optical data includes near-infrared reflectance and red band reflectance;

[0046] The radar image processing module is used to select sub-channel features corresponding to the pixel from the cross-polarization channel feature data;

[0047] An index calculation module is used to obtain a normalized vegetation index for each pixel based on the optical data.

[0048] The classification module is used to obtain the classification result corresponding to each pixel based on the normalized vegetation index, optical image band data and sub-channel features, so as to determine all aquatic vegetation types in the water area.

[0049] Thirdly, this application provides an electronic device including a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by the processor to implement the aquatic vegetation type monitoring method based on optical and SAR satellite imagery disclosed in the first aspect and any possible implementation thereof.

[0050] Fourthly, this application provides a computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the aquatic vegetation type monitoring method based on optical and SAR satellite imagery disclosed in the first aspect and any possible implementation thereof.

[0051] Fifthly, this application provides a computer program product comprising: computer instructions that, when executed on an electronic device, cause the electronic device to perform the aquatic vegetation type monitoring method based on optical and SAR satellite imagery disclosed in the first aspect and any possible implementation thereof.

[0052] The beneficial effects of the second to fifth aspects can be found in the first aspect and the beneficial effects of any possible implementation of the first aspect, and will not be repeated here. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the aquatic vegetation type monitoring method in the embodiments of this application;

[0054] Figure 2 This is a flowchart illustrating the aquatic vegetation type monitoring method in the embodiments of this application.

[0055] Figure 3This is a flowchart of the high-dimensional feature vector classification process in the embodiments of this application;

[0056] Figure 4 This is a flowchart illustrating the calculation of independent classification results in the embodiments of this application;

[0057] Figure 5 This is a block diagram of the electronic device in the embodiments of this application;

[0058] Figure 6 This is a block diagram of a system-on-chip (SoC) in the embodiments of this application. Detailed Implementation

[0059] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] The technical problems to be solved by the embodiments of this application will be described below.

[0061] Aquatic vegetation is an important component of aquatic ecosystems such as lakes and rivers, playing a vital role in maintaining ecological balance, improving water quality, and providing habitats for organisms. With the impacts of global climate change and human activities, the distribution and types of aquatic vegetation have undergone significant changes, making accurate monitoring and classification of aquatic vegetation a key task for ecological environmental protection and water resource management.

[0062] However, existing aquatic vegetation monitoring technologies still suffer from the following problems: First, traditional field survey methods are time-consuming, labor-intensive, costly, and have limited sample representativeness, making it difficult to achieve synchronous, rapid, and comprehensive coverage of vast lake areas. Second, monitoring methods based on single optical remote sensing are easily constrained by weather conditions. Optical sensors cannot penetrate clouds and fog, resulting in large areas of missing data in images acquired under cloudy and rainy weather, severely affecting the timeliness and continuity of monitoring. Furthermore, the phenomena of "different objects with the same spectrum" and "same objects with different spectra" exist, making classification methods relying solely on spectral information prone to confusion and misclassification. Third, existing optical and SAR image combination technologies suffer from rigid architecture and blind strategy, i.e., a fixed "either / or" choice between feature-level and decision-level combinations, failing to leverage the advantages of each for different land cover characteristics. Finally, existing technologies suffer from feature redundancy and noise introduction problems, tending to indiscriminately use all SAR polarization channels and their derivative channels as input, which not only increases the computational burden but may even impair classification accuracy due to introduced noise.

[0063] Therefore, there is an urgent need for a monitoring method for aquatic vegetation types that can effectively combine the advantages of optical remote sensing and radar remote sensing, adopt flexible fusion strategies for different aquatic vegetation types, and optimize feature selection, so as to improve the accuracy and efficiency of aquatic vegetation classification.

[0064] Therefore, in order to solve the above problems, this application provides a method and device for monitoring aquatic vegetation types based on optical and SAR satellite imagery. By acquiring spatiotemporally registered optical remote sensing images and radar images containing cross-polarization channels of the same area, optical data and band data such as near-infrared and red light reflectance corresponding to the pixels are extracted. At the same time, cross-polarization sub-channel features are selected, and after calculating the normalized vegetation index, the normalized vegetation index, optical band data and radar sub-channel features are combined to perform pixel-level classification, so as to achieve accurate identification of aquatic vegetation types in the area.

[0065] To better understand the method for monitoring aquatic vegetation types based on optical and SAR satellite imagery according to embodiments of this application, the following description is provided in conjunction with the appendix to the specification. Figure 1 —4. The solutions corresponding to the methods and apparatus of the embodiments of this application will be described in detail.

[0066] Reference Figure 1 and Figure 2 , Figure 1 This is a schematic diagram of the aquatic vegetation type monitoring method in the embodiments of this application. Figure 2 This is a flowchart illustrating the aquatic vegetation type monitoring method in this application.

[0067] like Figure 1 As shown in the embodiments of this application, a method for monitoring aquatic vegetation types based on optical and SAR satellite imagery includes steps S10-S50.

[0068] S10 involves acquiring optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered.

[0069] like Figure 2 As shown, step S10 corresponds to step S1, acquiring optical remote sensing image data and radar image data.

[0070] It should be noted that radar image data includes cross-polarization channel (vertical transmit, horizontal receive, VH) feature data and single-polarization channel (vertical transmit, vertical receive, VV) feature data. VH feature data is more sensitive to volume scattering effects caused by vegetation canopy, while VV feature data is more sensitive to surface scattering caused by smooth water surfaces. To highlight vegetation information, this application selects only VH feature data as the core information source and actively discards VV feature data and derived features such as polarization ratios calculated from both, avoiding the introduction of noise information irrelevant or redundant to the vegetation classification and monitoring task.

[0071] S20: Obtain optical data and optical image band data for each pixel from optical remote sensing image data.

[0072] The optical data includes near-infrared reflectance and red light band reflectance.

[0073] like Figure 2 As shown, step S20 corresponds to step S2, which involves optical remote sensing image data processing.

[0074] S30: Select sub-channel features corresponding to the pixel from the cross-polarization channel feature data.

[0075] like Figure 2 As shown, step S30 corresponds to step S3, which involves radar image data processing.

[0076] It is understandable that the cross-polarization channel feature data and the optical remote sensing image data are in one-to-one correspondence. Based on the spatiotemporal registration relationship between the optical image and the radar image, the coordinate position of each pixel in the radar image is determined, and then the data corresponding to that position is extracted from the cross-polarization channel feature data of the radar image data as the sub-channel feature.

[0077] In some embodiments, selecting sub-channel features corresponding to pixels from cross-polarization channel feature data includes:

[0078] The cross-polarization channel feature data is preprocessed to obtain preprocessed channel feature data. The preprocessing includes at least one of the following: orbit correction, thermal noise removal, radiometric calibration, speckle filtering, and terrain correction.

[0079] From the preprocessed channel feature data, select the sub-channel features corresponding to the pixels.

[0080] Understandably, by performing preprocessing on the cross-polarization channel feature data, such as orbit correction, thermal noise removal, radiometric calibration, speckle filtering, and terrain correction, radar data noise can be effectively suppressed, imaging errors eliminated, and terrain and geometric distortions corrected. This significantly improves the quality and reliability of the cross-polarization channel feature data, enabling the selected sub-channel features to more accurately reflect the true structural information of aquatic vegetation. This provides stable and high-quality radar feature input for subsequent multi-feature fusion classification, further improving the accuracy and stability of aquatic vegetation type identification.

[0081] In some embodiments, selecting sub-channel features corresponding to pixels from cross-polarization channel feature data further includes:

[0082] Nonlinear mathematical transformations are performed on the cross-polarization channel feature data to obtain enhanced cross-polarization channel feature data.

[0083] Sub-channel features are selected from the enhanced cross-polarization channel feature data.

[0084] It should be noted that the above-mentioned nonlinear mathematical transformations can include at least one of the logarithmic transformation and the square root transformation.

[0085] In this embodiment, the cross-polarization channel feature data is first subjected to a logarithmic transformation to obtain the transformed channel feature data, and then the transformed channel feature data is subjected to a square root transformation to obtain the enhanced cross-polarization channel feature data. The specific processing procedures for the logarithmic transformation and the square root transformation can be referred to the prior art, and will not be described in detail here.

[0086] The expression for the transform channel feature data is:

[0087]

[0088] The expression for the enhanced cross-polarization channel characteristic data is:

[0089]

[0090] Where A represents the transform channel feature data, v represents the cross-polarization channel feature data, σ represents the minimum positive constant used to avoid taking the logarithm of zero, and B represents the enhanced cross-polarization channel feature data.

[0091] It is understandable that performing a logarithmic transformation on the cross-polarization channel feature data first can effectively enhance the contrast in low-value areas, and then performing a square root transformation can effectively unify the image of extremely high values.

[0092] By performing nonlinear mathematical transformations on the cross-polarization channel feature data, the feature differences between different aquatic vegetation structures can be effectively enhanced, highlighting weakly different features and suppressing data redundancy and interference information. This results in enhanced cross-polarization channel feature data with higher feature recognition. Based on this, sub-channel features can be selected to further improve the ability of subsequent classification models to distinguish similar vegetation types, effectively alleviating the problem of insufficient discrimination of single radar features, thereby improving the precision and accuracy of aquatic vegetation type classification.

[0093] S40, based on optical data, yields the normalized difference vegetation index (NDVI) for each pixel.

[0094] like Figure 2 As shown, step S40 corresponds to step S4, which calculates the normalized vegetation index.

[0095] In some embodiments, the above-mentioned normalized vegetation index can be expressed as:

[0096]

[0097] Where NDVI(p) represents the normalized vegetation index of pixel p, x NIR Let x be the near-infrared reflectance of pixel p. RED Let p be the red band reflectance of pixel p.

[0098] S50, based on the normalized vegetation index, optical image band data and sub-channel features, obtains the classification result corresponding to each pixel to determine all aquatic vegetation types in the water area.

[0099] like Figure 2 As shown, step S50 corresponds to step S5, which determines whether the normalized vegetation index is greater than the threshold. If yes, feature-level combination classification is performed; otherwise, decision-level combination classification is performed, and finally, a thematic map of the classification results is drawn.

[0100] It should be noted that when neither optical remote sensing image data nor radar image data contains aquatic vegetation, the classification result for each pixel is water body; when both contain aquatic vegetation, the classification result for each pixel may be either aquatic vegetation type or water body. Furthermore, the aquatic vegetation type monitoring method in this embodiment only identifies aquatic vegetation and water bodies, and does not identify other objects, including but not limited to structures placed in the water.

[0101] Therefore, compared with the prior art, the aquatic vegetation type monitoring method based on optical and SAR satellite imagery provided in this application acquires optical remote sensing images and radar image data containing cross-polarization channels that are spatiotemporally registered in the same water area. It extracts optical band data, near-infrared reflectance, and red band reflectance of each pixel from the optical remote sensing images, calculates the normalized vegetation index based on the near-infrared reflectance and red band reflectance, and finally classifies the optical band data in combination with sub-channel features. This effectively solves the technical problems of low efficiency in traditional field surveys, susceptibility of single optical remote sensing to weather and spectral confusion, and insufficient classification accuracy due to lack of spectral information in single radar data. It achieves all-weather, large-scale, and high-precision aquatic vegetation type identification, and has the beneficial effects of good monitoring continuity, high classification accuracy, and applicability to complex lake environments.

[0102] The following is a detailed explanation of step S50.

[0103] refer to Figure 3 , Figure 3 A flowchart of high-dimensional feature vector classification in an embodiment of this application is shown.

[0104] like Figure 3 As shown, based on the normalized vegetation index, optical image band data, and sub-channel features, the classification result corresponding to each pixel is obtained, including steps S51-S52.

[0105] In S51, when the normalized vegetation index is determined to be greater than the preset threshold, a high-dimensional feature vector is constructed based on optical image band data and sub-channel features.

[0106] For example, the threshold mentioned above can be set to 0.2. It is understood that, based on historical research literature and remote sensing application experience of Taihu Lake, aquatic vegetation with an NDVI value above 0.2 usually indicates a relatively lush state. Therefore, in the absence of detailed sample data or to simplify calculations, 0.2 can be used as an initial, universally applicable reference threshold.

[0107] In some embodiments, the high-dimensional feature vector is obtained by feature stitching optical image band data and sub-channel features.

[0108] It should be noted that the above high-dimensional feature vector can be represented as:

[0109]

[0110] Where, x feature x is a high-dimensional feature vector. opt For optical image band data, x sar This is a sub-channel feature.

[0111] It is understood that, in the embodiments of this application, the Normalized Difference Vegetation Index (NDVI) is preferred as the basis for judging the switching combination strategy. It can intuitively reflect vegetation cover and growth vitality, clearly distinguish between high-density and low-density aquatic plant areas, and provide an objective and stable threshold basis for switching classification strategies. Its calculation only requires two bands, red and near-infrared, making the algorithm simple, efficient, and computationally inefficient. It also exhibits strong robustness to illumination and atmospheric interference, making it suitable for large-scale automated processing of remote sensing images. Furthermore, NDVI, as a universal vegetation index, is compatible with most optical satellite sensors, facilitating the migration and reuse of the method across different regions and data sources.

[0112] It should be noted that other spectral indices that can effectively distinguish between high-density aquatic plants and water / sparse aquatic plants can be used to replace or supplement NDVI. For example, the enhanced vegetation index (EVI) and soil adjusted vegetation index (SAVI) can be used to assess vegetation cover, or the normalized difference water index (NDWI) can be used to specifically identify water bodies, so as to achieve the same or similar functions of land cover prediction and strategy scheduling.

[0113] S52 uses high-dimensional feature vectors as input data for a feature-level combination classification model to obtain classification results.

[0114] It should be noted that the above classification results can be expressed as:

[0115]

[0116] in, x represents the classification result of pixel p. p Let f be the high-dimensional feature vector of pixel p, and let f represent the feature-level combination classification model.

[0117] In some embodiments, high-dimensional feature vectors are used as input data for a feature-level combination classification model to obtain classification results, including steps S521-S522.

[0118] S521 standardizes the high-dimensional feature vector to obtain a standardized high-dimensional feature vector.

[0119] It should be noted that the above standardization includes at least one of mean normalization and standard deviation.

[0120] S522 uses standardized high-dimensional feature vectors as input data for a feature-level combination classification model to obtain classification results.

[0121] It is understandable that by standardizing high-dimensional feature vectors, differences in numerical magnitude and units between features from different sources can be eliminated, thereby improving model learning efficiency and classification stability, and ultimately enhancing the recognition accuracy of high-density aquatic vegetation areas.

[0122] In some embodiments, the above feature-level combined classification model is a support vector machine (SVC).

[0123] It should be noted that when training a feature-level combined classification model, a high-dimensional feature vector is used as input, and the true class label of this vector is used as the supervision signal for model training. The training process of the feature-level combined classification model aims to solve an optimization problem: finding an optimal hyperplane that can separate samples of different classes with the maximum margin. Its objective function is usually expressed as:

[0124]

[0125] Where w and b define the hyperplane, is the slack variable, c is the regularization coefficient used to balance the margin maximization and the misclassified sample penalty, and i is the index of the training sample, representing the traversal from the first sample to the last sample. Its function is to ensure that the slack variables of all samples are accumulated into the objective function.

[0126] Besides Support Vector Machines (SVMs) which can be used as the feature-level combination classification model mentioned above, other machine learning or deep learning algorithms can also be used as alternatives, such as Random Forest, Gradient Boosting Decision Trees (e.g., XGBoost, LightGBM), or Multi-layer Perceptron (MLP) neural network models, which will not be elaborated on in this article.

[0127] refer to Figure 4 , Figure 4 A flowchart illustrating the calculation of independent classification results in an embodiment of this application is shown.

[0128] like Figure 4 As shown in the embodiment of this application, the classification result corresponding to each pixel is obtained based on the normalized vegetation index, optical image band data and sub-channel features, and the method also includes steps S53-S56.

[0129] In step S53, under the condition that the normalized vegetation index is not greater than the preset threshold, the optical image band data is used as the input data of the optical feature classification model to obtain the first classification probability vector.

[0130] It should be noted that the selection and training of optical feature classification models can refer to the section on feature-level combination classification models above, and will not be repeated here.

[0131] In some embodiments, optical image band data is used as input data for an optical feature classification model to obtain a first classification probability vector, including steps S531-S532.

[0132] S531 standardizes the optical image band data to obtain standardized optical image band data. Standardization includes at least one of mean and standard deviation.

[0133] S532 uses standardized optical image band data as input data for the optical feature classification model to obtain the first classification probability vector.

[0134] It should be noted that the beneficial effects of standardizing optical image band data can be found above, and will not be repeated here.

[0135] S54. The sub-channel features are used as input data for the radar image feature classification model to obtain the second classification probability vector.

[0136] It should be noted that the selection and training of radar image feature classification models can refer to the section on feature-level combination classification models mentioned above, and will not be repeated here.

[0137] In some embodiments, the sub-channel features are used as input data for the radar image feature classification model to obtain a second classification probability vector, including steps S541-S542.

[0138] In S541, the sub-channel features are standardized to obtain standardized sub-channel features. The standardization includes at least one of mean and standard deviation.

[0139] S542 uses the standardized sub-channel features as input data for the radar image feature classification model to obtain the second classification probability vector.

[0140] It should be noted that the beneficial effects of standardizing sub-channel features can be found in the above text, and will not be repeated here.

[0141] S55, based on preset weight coefficients, merge the first classification probability vector and the second classification probability vector to obtain a fused vector.

[0142] It should be noted that the fusion vector can be represented as:

[0143]

[0144] in, This represents the fusion vector of pixel p. Let p be the first classification probability vector. Let α be the second classification probability vector of pixel p, and α be the weight coefficient.

[0145] For example, in the embodiments of this application, the above-mentioned weighting coefficients Setting the weight to 0.5 is understandable because, in low-vegetation-coverage areas targeted by decision-level fusion, optical and radar data are sensitive to the spectral characteristics and physical structure of ground features, respectively, providing equally important and highly complementary classification criteria. Therefore, in the absence of clear prior information indicating that one is superior, assigning equal weight to both is a robust choice that balances their respective advantages and disadvantages and reduces the risk of misjudgment from a single data source.

[0146] It is understandable that the weighting coefficients As an enhancement, this weighting coefficient Adaptive weighting can be applied based on the performance of optical and SAR expert models on independent validation sets, allocating higher weights to models that perform better. Furthermore, more complex fusion rules, such as Bayesian fusion or Dempster-Shafer evidence theory, can be used to replace the weighted average method to handle uncertainties in more complex situations.

[0147] S56. Determine the maximum probability value of the fusion vector and take the category corresponding to the maximum probability value as the classification result.

[0148] In some embodiments, through mathematical functions (such as...) The function is used to determine the maximum probability value, where the classification result can be expressed as:

[0149]

[0150] in, This is the classification result for pixel p.

[0151] Understandably, for areas with low vegetation cover, optical and radar classification models are used to output probability vectors and then fused together for decision-making according to preset weights. This approach fully leverages the spectral advantages of optical data in identifying non-vegetation and low-coverage land features, while using radar data to compensate for the insufficient identification of optical data under weak vegetation signals. The decision-level fusion improves classification reliability and avoids misclassification and omissions that are prone to occur in sparse vegetation areas due to a single data source. This ensures that the overall monitoring method can maintain stable and accurate classification results across the entire region and under different vegetation densities.

[0152] In practical applications, after calculating the classification result of each pixel, it is output in a universal, lossless, and high-precision raster image format file. Staff can import this file into a professional-grade geographic information system software platform, and reorganize and render the classification results of each pixel based on the row and column spatial position of each pixel, finally generating a complete thematic map of aquatic vegetation type classification. According to this thematic map of aquatic vegetation type classification, the aquatic vegetation type of each region can be clearly seen.

[0153] Secondly, this application provides a device for monitoring aquatic vegetation types based on optical and SAR satellite imagery, comprising:

[0154] The data acquisition module is used to acquire optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered, wherein the radar image data includes cross-polarization channel feature data.

[0155] The remote sensing image processing module is used to obtain optical data and optical image band data of each pixel from the optical remote sensing image data, wherein the optical data includes near-infrared reflectance and red band reflectance;

[0156] The radar image processing module is used to select sub-channel features corresponding to the pixel from the cross-polarization channel feature data;

[0157] An index calculation module is used to obtain a normalized vegetation index for each pixel based on the optical data.

[0158] The classification module is used to obtain the classification result corresponding to each pixel based on the normalized vegetation index, optical image band data and sub-channel features, so as to determine all aquatic vegetation types in the water area.

[0159] In the apparatus of this application embodiment, each module executes the method of the above embodiment, and its specific functions and corresponding technical effects can be referred to the above embodiment. Figures 1-4 The methods explained will not be repeated here.

[0160] This application provides an electronic device, which includes a processor and a memory. The memory stores at least one instruction or at least one program. When the processor loads and executes the instruction or program, the electronic device performs the aquatic vegetation type monitoring method described in the above embodiments. Its specific functions and corresponding technical effects can be found in the above embodiments. Figure 1 - Figure 4 The methods for monitoring aquatic vegetation types based on optical and SAR satellite imagery, as explained above, will not be elaborated upon here. The following section will combine... Figure 5 The electronic devices described in the embodiments of this application will be described in detail.

[0161] refer to Figure 5 The diagram shows a block diagram of an electronic device 1200 according to one embodiment of this application. The electronic device 1200 may include one or more processors 1201 coupled to a controller hub 1203. In at least one embodiment, the controller hub 1203 communicates with the processor 1201 via a multi-branch bus such as a front side bus (FSB) 1210, a point-to-point interface such as a quick path interconnect (QPI), or a similar connection. The processor 1201 executes instructions controlling general types of data processing operations. In one embodiment, the controller hub 1203 includes, but is not limited to, a graphics memory controller hub (GMCH) (not shown) and an input / output hub (IOH) (which may be on a separate chip) (not shown), wherein the GMCH includes memory and a graphics controller and is coupled to the IOH.

[0162] Electronic device 1200 may also include a coprocessor 1202 and a memory 1204 coupled to a controller hub 1203. Alternatively, one or both of the memory and the GMCH may be integrated within the processor (as described in this application), with memory 1204 and coprocessor 1202 directly coupled to processor 1201 and controller hub 1203, which resides on a single chip with the IOH. Memory 1204 may be, for example, dynamic random access memory (DRAM), phase change memory (PCM), or a combination of both. In one embodiment, coprocessor 1202 is a dedicated processor, such as, for example, a high-throughput MIC (many integerized core) processor, a network or communication processor, a compression engine, a graphics processor, a general-purpose computing on GPU (GPGPU), or an embedded processor, etc. Optional properties of coprocessor 1202 are indicated by dashed lines. Figure 5 middle.

[0163] As a computer-readable storage medium, memory 1204 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. For example, memory 1204 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device such as one or more hard-disk drives (HDDs), one or more compact disc (CD) drives, and / or one or more digital versatile disc (DVD) drives.

[0164] In one embodiment, electronic device 1200 may further include a network interface controller (NIC) 1206. Network interface 1206 may include a transceiver for providing a radio interface for electronic device 1200 to communicate with any other suitable device, such as a front-end module, antenna, etc. In various embodiments, network interface 1206 may be integrated with other components of electronic device 1200. Network interface 1206 can implement the functions of the communication unit in the above embodiments.

[0165] Electronic device 1200 may further include input / output (I / O) device 1205. I / O device 1205 may include: a user interface designed to enable a user to interact with electronic device 1200; a peripheral component interface designed to enable peripheral components to also interact with electronic device 1200; and / or sensors designed to determine environmental conditions and / or location information related to electronic device 1200.

[0166] It is worth noting that, Figure 5 This is merely an example. That is, although... Figure 5 The electronic device 1200 shown includes multiple devices such as a processor 1201, a coprocessor 1202, a controller hub 1203, and a memory 1204. However, in practical applications, devices using the methods of this application may include only a portion of the devices in the electronic device 1200. For example, it may include only the processor 1201 and the network interface 1206. Figure 5 The properties of the optional devices are shown in dashed lines. According to some embodiments of this application, the memory 1204, which is a computer-readable storage medium, stores instructions or programs that, when executed on a computer, perform the aquatic vegetation type monitoring method described in the above embodiments. Specific details can be found in the methods described in the above embodiments, and will not be repeated here.

[0167] Now for reference Figure 6The diagram shown is a block diagram of a system-on-chip (SoC) 1300 according to an embodiment of this application. Figure 6 In the diagram, similar components share the same reference numerals. Additionally, dashed boxes are an optional feature for more advanced SoCs. Figure 6 In this SoC 1300, the following are included: an interconnect unit 1350 coupled to an application processor 1310; a system proxy unit 1380; a bus controller unit 1390; an integrated memory controller unit 1340; a group or one or more coprocessors 1320, which may include integrated graphics logic, an image processor, an audio processor, and a video processor; a static random access memory (SRAM) unit 1330; and a direct memory access (DMA) unit 1360. In one embodiment, the coprocessor 1320 includes a dedicated processor, such as, for example, a network or communication processor, a compression engine, a GPGPU, a high-throughput MIC processor, or an embedded processor.

[0168] The static random access memory (SRAM) unit 1330 may include one or more computer-readable media for storing data and / or instructions. The computer-readable storage medium may store instructions, specifically, temporary and permanent copies of those instructions. These instructions may include, when executed by at least one unit in the processor, causing the SoC 1300 to perform the aquatic vegetation type monitoring method according to the above embodiments, as detailed in the methods described above, which will not be repeated here.

[0169] This application provides a computer-readable storage medium storing at least one instruction or at least one program. The instruction or program is loaded and executed by a processor to implement the aquatic vegetation type monitoring method described in the above embodiments. Its specific functions and corresponding technical effects can be found in the above embodiments. Figures 1-4 The methods for monitoring aquatic vegetation types based on optical and SAR satellite imagery, as explained above, will not be elaborated upon here.

[0170] This application provides a computer program product, including computer instructions. When the computer instructions are executed on an electronic device, the electronic device causes the electronic device to implement the aquatic vegetation type monitoring method described in the above embodiments. Its specific functions and corresponding technical effects can be found in the above embodiments. Figures 1-4 The methods for monitoring aquatic vegetation types based on optical and SAR satellite imagery, as explained above, will not be elaborated upon here.

[0171] Various embodiments of the mechanisms disclosed in this application can be implemented in hardware, software, firmware, or combinations of these implementation methods. Embodiments of this application can be implemented as computer programs or program code executable on a programmable system, the programmable system including at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device.

[0172] Program code can be applied to input instructions to execute the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, the processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application-specific integrated circuit (ASIC), or a microprocessor.

[0173] The program code can be implemented using a high-level procedural language or an object-oriented programming language to communicate with the processing system. Assembly language or machine language can also be used when needed. In fact, the mechanisms described in this application are not limited to any particular programming language. In either case, the language can be a compiled language or an interpreted language.

[0174] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored thereon on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or through other computer-readable media. Therefore, machine-readable media may include any mechanism for storing or transmitting information in a machine-readable (e.g., computer-readable) form, including but not limited to floppy disks, optical disks, CD-ROMs, compact disc read-only memory (CD-ROMs), magneto-optical disks, read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic cards or optical cards, flash memory, or tangible machine-readable storage for transmitting information (e.g., carrier waves, infrared signals, digital signals, etc.) using the Internet in the form of electrical, optical, acoustic, or other forms of propagated signals. Therefore, machine-readable media include any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a machine-readable (e.g., computer-readable) form.

[0175] In the accompanying drawings, some structural or methodological features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be necessary. Rather, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the accompanying drawings. Furthermore, including structural or methodological features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may be omitted or may be combined with other features.

[0176] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0177] It should be noted that all units / modules mentioned in the device embodiments of this application are logical units / modules. Physically, a logical unit / module can be a physical unit / module, a part of a physical unit / module, or a combination of multiple physical units / modules. The physical implementation of these logical units / modules themselves is not the most important factor; the combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed in this application. Furthermore, to highlight the innovative aspects of this application, the above-described device embodiments of this application have not introduced units / modules that are not closely related to solving the technical problems proposed in this application. This does not mean that the above-described device embodiments do not contain other units / modules.

[0178] It should be noted that in the examples and description of this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0179] Although this application has been illustrated and described with reference to certain preferred embodiments thereof, those skilled in the art should understand that various changes in form and detail may be made thereto without departing from the spirit and scope of this application.

Claims

1. A method for monitoring aquatic vegetation types based on optical and SAR satellite imagery, characterized in that, include: Acquire optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered, wherein the radar image data includes cross-polarization channel feature data; Optical data and optical image band data for each pixel are obtained from the optical remote sensing image data, wherein the optical data includes near-infrared reflectance and red band reflectance; Select the sub-channel feature corresponding to the pixel from the cross-polarization channel feature data; Based on the optical data, a normalized vegetation index is obtained for each pixel. Based on the normalized vegetation index, the optical image band data, and the sub-channel features, a classification result corresponding to each pixel is obtained to determine all aquatic vegetation types in the water area.

2. The method according to claim 1, characterized in that, The step of selecting the sub-channel feature corresponding to the pixel from the cross-polarization channel feature data includes: The cross-polarization channel feature data is preprocessed to obtain preprocessed channel feature data, wherein the preprocessing includes at least one of orbit correction, thermal noise removal, radiometric calibration, speckle filtering, and terrain correction; From the preprocessed channel feature data, select the sub-channel feature corresponding to the pixel.

3. The method according to claim 1, characterized in that, The step of selecting the sub-channel feature corresponding to the pixel from the cross-polarization channel feature data further includes: The cross-polarization channel feature data is subjected to a nonlinear mathematical transformation to obtain enhanced cross-polarization channel feature data, wherein the nonlinear mathematical transformation includes at least one of logarithmic transformation and square root transformation; The sub-channel features are selected from the enhanced cross-polarization channel feature data.

4. The method according to claim 1, characterized in that, The classification result corresponding to each pixel is obtained based on the normalized vegetation index, the optical image band data, and the sub-channel features, including: If the normalized vegetation index is determined to be greater than a preset threshold, a high-dimensional feature vector is constructed based on the optical image band data and the sub-channel features. The high-dimensional feature vector is used as input data for the feature-level combination classification model to obtain the classification result.

5. The method according to claim 1, characterized in that, The step of obtaining a classification result corresponding to each pixel based on the normalized vegetation index, the optical image band data, and the sub-channel features also includes: If the normalized vegetation index is determined to be no greater than a preset threshold, the optical image band data is used as the input data of the optical feature classification model to obtain the first classification probability vector. The sub-channel features are used as input data for the radar image feature classification model to obtain the second classification probability vector; Based on preset weight coefficients, the first classification probability vector and the second classification probability vector are fused to obtain a fused vector; The maximum probability value of the fusion vector is determined, and the category corresponding to the maximum probability value is taken as the classification result.

6. The method according to claim 4, characterized in that, The step of using the high-dimensional feature vector as input data for a feature-level combination classification model to obtain the classification result includes: The high-dimensional feature vector is standardized to obtain a standardized high-dimensional feature vector, wherein the standardization includes at least one of mean normalization and standard deviation. The standardized high-dimensional feature vector is used as input data for the feature-level combination classification model to obtain the classification result.

7. The method according to claim 5, characterized in that, The step of using the optical image band data as input data for the optical feature classification model to obtain the first classification probability vector includes: The optical image band data is standardized to obtain standardized optical image band data, wherein the standardization includes at least one of mean normalization and standard deviation. The standardized optical image band data is used as input data for the optical feature classification model to obtain the first classification probability vector; The step of using the sub-channel features as input data for the radar image feature classification model to obtain the second classification probability vector includes: The sub-channel features are standardized to obtain standardized sub-channel features; The standardized sub-channel features are used as input data for the radar image feature classification model to obtain the second classification probability vector.

8. A device for monitoring aquatic vegetation types based on optical and SAR satellite imagery, characterized in that, include: The data acquisition module is used to acquire optical remote sensing image data and radar image data of the same water area that have been spatiotemporally registered, wherein the radar image data includes cross-polarization channel feature data. The remote sensing image processing module is used to obtain optical data and optical image band data of each pixel from the optical remote sensing image data, wherein the optical data includes near-infrared reflectance and red band reflectance; The radar image processing module is used to select sub-channel features corresponding to the pixel from the cross-polarization channel feature data; An index calculation module is used to obtain a normalized vegetation index for each pixel based on the optical data. The classification module is used to obtain a classification result corresponding to each pixel based on the normalized vegetation index, the optical image band data, and the sub-channel features, so as to determine all aquatic vegetation types in the water area.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the method for monitoring aquatic vegetation types based on optical and SAR satellite imagery as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction or at least one program, which is loaded and executed by a processor to implement the method for monitoring aquatic vegetation types based on optical and SAR satellite imagery as described in any one of claims 1 to 7.

Images