Remote sensing satellite data transmission analysis method and system storage medium

Abstract

The invention discloses a remote sensing satellite data transmission analysis method and system and a storage medium. Collecting first remote sensing data in real time to perform remote sensing imaging, performing region clustering analysis based on image features through a preset clustering algorithm, and selecting a low feature region; importing the first remote sensing data into a conversion model for data dimension reduction to form intermediate data; transmitting the intermediate data to a preset terminal, and raising the dimension of the intermediate data based on the conversion model to obtain second remote sensing data; acquiring comparison remote sensing imaging data based on an observation area of real-time acquisition remote sensing data, performing image feature extraction from the comparison remote sensing imaging data based on a low feature area to obtain corrected feature data, performing imaging analysis on second remote sensing data, and obtaining real-time transmission remote sensing image data based on a DCA feature fusion algorithm; and sending the real-time transmission remote sensing image data to a preset terminal device. According to the invention, the stability of real-time transmission can be effectively realized, and the completeness of remote sensing data features can be ensured.

Description

Technical Field

[0001] This invention relates to the field of remote sensing data analysis, and more specifically, to a remote sensing satellite data transmission and analysis method and system storage medium. Background Technology

[0002] With the rapid development of remote sensing technology, hyperspectral Earth observation remote sensing satellites are increasingly widely used in resource surveys, environmental monitoring, disaster assessment, and other fields. However, the amount of data collected by hyperspectral satellites is enormous, and traditional data transmission methods often face problems such as limited downlink bandwidth and low transmission efficiency, which seriously restricts the real-time application and analysis of remote sensing data. Therefore, how to improve the downlink transmission efficiency of hyperspectral satellite data and how to effectively transmit remote sensing imaging data and restore data to ground terminals have become key issues that urgently need to be addressed in the field of remote sensing technology. Summary of the Invention

[0003] This invention overcomes the shortcomings of existing technologies and proposes a remote sensing satellite data transmission and analysis method and a system storage medium.

[0004] The first aspect of this invention provides a remote sensing satellite data transmission method, comprising:

[0005] Real-time acquisition of hyperspectral remote sensing satellite data; preprocessing of the remote sensing satellite data to form first remote sensing data.

[0006] Based on the first remote sensing data, remote sensing imaging is performed, and a region clustering analysis based on image features is conducted using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them.

[0007] A transformation model based on a neural network model is constructed. The first remote sensing data is imported into the transformation model to perform data dimensionality reduction and form intermediate data.

[0008] The intermediate data is transmitted to a preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data;

[0009] Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Based on the low feature area, image features are extracted from the comparative remote sensing imaging data to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data.

[0010] The real-time remote sensing image data is sent to a preset terminal device.

[0011] In this scheme, the real-time acquisition of hyperspectral remote sensing satellite data and the preprocessing of the remote sensing satellite data to form the first remote sensing data specifically involve:

[0012] Real-time acquisition of hyperspectral remote sensing satellite data and labeling of the hyperspectral remote sensing satellite data as real-time remote sensing data;

[0013] The real-time remote sensing data is preprocessed by data cleaning, noise reduction, and radiometric calibration to obtain the first remote sensing data.

[0014] In this scheme, the step of performing remote sensing imaging based on the first remote sensing data and conducting region clustering analysis based on image features using a preset clustering algorithm to obtain multiple sets of image regions and then filtering out low-feature regions specifically involves:

[0015] Resampling and image synthesis are performed on the first remote sensing data to obtain initial image data;

[0016] The initial image data is divided into grid regions to form multiple sub-regions and corresponding image data for multiple regions.

[0017] The image data of the multiple regions are subjected to SIFT-based feature extraction to obtain multiple feature data;

[0018] The multiple feature data are used as clustering sample data. Based on the DBSCAN clustering algorithm, the sample data are clustered and grouped to form multiple groups of image regions.

[0019] In this scheme, the step of performing remote sensing imaging based on the first remote sensing data and conducting region clustering analysis based on image features using a preset clustering algorithm to obtain multiple sets of image regions and then filtering out low-feature regions from them further includes:

[0020] Based on the multiple sets of image regions, the image feature data of each set of image regions are integrated to form multiple sets of integrated feature data;

[0021] By using the threshold method, pixel-level noise data statistics and proportion calculations are performed on the integrated feature data, and the noise proportion value is obtained.

[0022] Multiple noise percentage values ​​were obtained by analyzing multiple sets of integrated feature data;

[0023] The image regions with the highest noise ratio are marked and these regions are designated as low-feature regions.

[0024] In this scheme, the construction of a transformation model based on a neural network model, and the import of the first remote sensing data into the transformation model for dimensionality reduction to form intermediate data, specifically involves:

[0025] A conversion model based on a neural network model is constructed, wherein the conversion model is based on an autoencoder and includes an encoder and a decoder;

[0026] Lossless historical remote sensing data is obtained from the system database, and the lossless historical remote sensing data is imported into the conversion model. The data is reduced in dimensionality and compressed by the encoder, and restored in dimensionality by the decoder. The parameters are optimized by joint training. In the joint training, the loss function is set to optimize and adjust the parameters of the encoder and decoder based on the mean square error.

[0027] The data is trained cyclically using lossless historical remote sensing data until the data features represented in the low-dimensional representation of the transformation model meet expectations.

[0028] In this scheme, the construction of a transformation model based on a neural network model, and the import of the first remote sensing data into the transformation model for dimensionality reduction to form intermediate data, specifically involves:

[0029] The first remote sensing data transformation model is used to reduce the dimensionality of the data to obtain intermediate data;

[0030] Calculate the structural similarity index S between the first remote sensing data and the intermediate data. If S is lower than the preset threshold, then perform data dimensionality reduction again.

[0031] In this scheme, the intermediate data is transmitted to a preset terminal, and the intermediate data is then upscaled based on a transformation model to obtain the second remote sensing data. Specifically, this involves:

[0032] Intermediate data is transmitted to the user's preset terminal via satellite terminal;

[0033] On the user's preset terminal, intermediate data is imported into the conversion model, and the data is upgraded and restored based on the decoder in the conversion model to generate second remote sensing data.

[0034] In this scheme, the observation area based on real-time acquired remote sensing data is used to obtain comparative remote sensing imaging data. Image features are extracted from the comparative remote sensing imaging data based on low-feature areas to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. Specifically:

[0035] Based on the observation area of ​​real-time acquired remote sensing data, comparative remote sensing imaging data is obtained;

[0036] Based on low-feature regions, the comparative remote sensing imaging data are labeled accordingly to obtain the region to be analyzed.

[0037] Based on the region to be analyzed, image data of the corresponding region is obtained from the comparative remote sensing imaging data. The obtained image data is then subjected to SIFT-based feature extraction to obtain corrected feature data.

[0038] The second remote sensing data is used for imaging analysis to obtain the target image;

[0039] The DCA feature fusion algorithm is used to fuse features of the target image based on the corrected feature data, and to form real-time transmitted remote sensing image data.

[0040] A second aspect of the present invention also provides a remote sensing satellite data transmission system, the system comprising: a memory and a processor, wherein the memory includes a remote sensing satellite data transmission program, and the remote sensing satellite data transmission program, when executed by the processor, performs the following steps:

[0041] Real-time acquisition of hyperspectral remote sensing satellite data; preprocessing of the remote sensing satellite data to form first remote sensing data.

[0042] Based on the first remote sensing data, remote sensing imaging is performed, and a region clustering analysis based on image features is conducted using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them.

[0043] A transformation model based on a neural network model is constructed. The first remote sensing data is imported into the transformation model to perform data dimensionality reduction and form intermediate data.

[0044] The intermediate data is transmitted to a preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data;

[0045] Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Based on the low feature area, image features are extracted from the comparative remote sensing imaging data to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data.

[0046] The real-time remote sensing image data is sent to a preset terminal device.

[0047] A third aspect of the present invention also provides a computer-readable storage medium including a remote sensing satellite data transmission program, which, when executed by a processor, implements the steps of the remote sensing satellite data transmission method as described in any of the preceding claims.

[0048] This invention discloses a remote sensing satellite data transmission and analysis method and a system storage medium. The method involves real-time acquisition of first remote sensing data for remote sensing imaging, followed by region clustering analysis based on image features using a preset clustering algorithm to select low-feature regions. The first remote sensing data is then imported into a transformation model for dimensionality reduction, forming intermediate data. This intermediate data is transmitted to a preset terminal, and its dimensionality is increased based on the transformation model to obtain second remote sensing data. Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Image features are extracted from the comparative remote sensing imaging data based on the low-feature regions to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and real-time transmitted remote sensing image data is obtained based on a DCA feature fusion algorithm. This real-time transmitted remote sensing image data is then sent to the preset terminal device. This invention effectively achieves stable real-time transmission and ensures the integrity of remote sensing data features. Attached Figure Description

[0049] Figure 1 A flowchart of a remote sensing satellite data transmission method according to the present invention is shown;

[0050] Figure 2 The flowchart of the first remote sensing data acquisition process of the present invention is shown;

[0051] Figure 3 A block diagram of a remote sensing satellite data transmission system according to the present invention is shown. Detailed Implementation

[0052] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0053] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0054] Figure 1 A flowchart of a remote sensing satellite data transmission method according to the present invention is shown.

[0055] like Figure 1 As shown, the first aspect of the present invention provides a remote sensing satellite data transmission method, comprising:

[0056] S102, real-time acquisition of hyperspectral remote sensing satellite data, and preprocessing of the remote sensing satellite data to form first remote sensing data;

[0057] S104, Based on the first remote sensing data, perform remote sensing imaging and perform region clustering analysis based on image features using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them;

[0058] S106, Construct a transformation model based on a neural network model, import the first remote sensing data into the transformation model to perform data dimensionality reduction, and form intermediate data;

[0059] S108, the intermediate data is transmitted to the preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data;

[0060] S110: Based on the observation area of ​​the real-time acquisition of remote sensing data, acquire comparative remote sensing imaging data, extract image features from the comparative remote sensing imaging data based on low feature areas, obtain corrected feature data, perform imaging analysis on the second remote sensing data, and fuse the imaging data and corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data.

[0061] The real-time remote sensing image data is sent to a preset terminal device.

[0062] Figure 2 A flowchart of the first remote sensing data acquisition process of the present invention is shown.

[0063] According to an embodiment of the present invention, the real-time acquisition of hyperspectral remote sensing satellite data and the preprocessing of the remote sensing satellite data to form first remote sensing data specifically include:

[0064] S202, Real-time acquisition of hyperspectral remote sensing satellite data and marking the hyperspectral remote sensing satellite data as real-time remote sensing data;

[0065] S204, perform data cleaning, noise reduction, and radiometric calibration preprocessing on the real-time remote sensing data to obtain the first remote sensing data.

[0066] It should be noted that the real-time acquisition of hyperspectral remote sensing satellite data refers to data acquisition at the satellite terminal, and the corresponding observation target area refers to the observation area.

[0067] According to an embodiment of the present invention, the step of performing remote sensing imaging based on the first remote sensing data and performing region clustering analysis based on image features using a preset clustering algorithm to obtain multiple groups of image regions and then filtering out low-feature regions specifically involves:

[0068] Resampling and image synthesis are performed on the first remote sensing data to obtain initial image data;

[0069] The initial image data is divided into grid regions to form multiple sub-regions and corresponding image data for multiple regions.

[0070] The image data of the multiple regions are subjected to SIFT-based feature extraction to obtain multiple feature data;

[0071] The multiple feature data are used as clustering sample data. Based on the DBSCAN clustering algorithm, the sample data are clustered and grouped to form multiple groups of image regions.

[0072] It should be noted that each of the multiple image regions includes at least one sub-region, and generally more than two. By clustering the regions, the regional image data in all the corresponding sub-regions of each region group have similar characteristics, including similarity in noise ratio. That is, after grouping, some regions with more noise and lower features can be filtered out, so that remote sensing image enhancement can be performed based on the low-feature regions in the subsequent process.

[0073] It is worth mentioning that in the real-time transmission of remote sensing data, due to the limitations of the transmitted data, it is necessary to reduce or compress the data volume as much as possible and transmit the data between the satellite and the ground terminal. Because the original hyperspectral data has a high dimension, it is difficult to transmit all the data. However, after compression and dimensionality reduction, there may be image distortion or excessive noise in the remote sensing image of a certain area, resulting in unsatisfactory results in some areas. Therefore, this invention performs imaging analysis on the original remote sensing data and, based on a clustering algorithm, performs feature analysis and clustering grouping on different image regions. This effectively classifies potential noise points, significantly improving the effectiveness and targeting of subsequent feature fusion and image enhancement. Existing technologies lack preliminary analysis and clustering of the original imaging images, making it difficult to identify transmission defect areas and perform targeted feature fusion enhancement. Furthermore, after dimensionality reduction and transmission to ground users is interrupted, there is some data loss in the dimensionality-reduced data, and the restored data may still contain distortion and loss of detail in certain areas. Therefore, after obtaining the second remote sensing data, this invention pre-acquires corresponding comparison images for low-feature areas (the comparison images are historical remote sensing data images from Earth observations, which have a certain reference value), and extracts corresponding feature data through aggressive DCA-based feature fusion to obtain corrected imaging data. This significantly improves the degree of image detail restoration. During dimensionality-reduction-based transmission, it effectively ensures the stability of real-time transmission and the integrity of important remote sensing data features, greatly enhancing the further application of remote sensing technology.

[0074] According to an embodiment of the present invention, the step of performing remote sensing imaging based on the first remote sensing data and performing region clustering analysis based on image features using a preset clustering algorithm to obtain multiple groups of image regions and filter out low-feature regions from them further includes:

[0075] Based on the multiple sets of image regions, the image feature data of each set of image regions are integrated to form multiple sets of integrated feature data;

[0076] By using the threshold method, pixel-level noise data statistics and proportion calculations are performed on the integrated feature data, and the noise proportion value is obtained.

[0077] Multiple noise percentage values ​​were obtained by analyzing multiple sets of integrated feature data;

[0078] The image regions with the highest noise ratio are marked and these regions are designated as low-feature regions.

[0079] It should be noted that, among the multiple noise proportion values, each group of image regions has one noise proportion value. When marking the group of image regions with the largest noise proportion value, a corresponding group of image regions is obtained. Since a group of image regions includes multiple sub-regions, the corresponding multiple sub-regions are summed to form a large region as a low-feature region. Assuming there are a total of N groups of image regions, N noise proportion values ​​are optimally obtained.

[0080] The noise percentage reflects the proportion of distorted and noisy points in an image region. The higher the noise percentage, the more noisy points there are in the image region, and the lower the amount of image representation feature data. This makes it easier for data loss to occur during dimensionality reduction. Therefore, special labeling is required.

[0081] According to an embodiment of the present invention, the construction of a transformation model based on a neural network model, wherein the first remote sensing data is imported into the transformation model for dimensionality reduction to form intermediate data, specifically includes:

[0082] A conversion model based on a neural network model is constructed, wherein the conversion model is based on an autoencoder and includes an encoder and a decoder;

[0083] Lossless historical remote sensing data is obtained from the system database, and the lossless historical remote sensing data is imported into the conversion model. The data is reduced in dimensionality and compressed by the encoder, and restored in dimensionality by the decoder. The parameters are optimized by joint training. In the joint training, the loss function is set to optimize and adjust the parameters of the encoder and decoder based on the mean square error.

[0084] The data is trained cyclically using lossless historical remote sensing data until the data features represented in the low-dimensional representation of the transformation model meet expectations.

[0085] It should be noted that the system database is used to store historical data records, and the lossless historical remote sensing data specifically refers to data records that have been pre-selected by the user, which are used to train the conversion model and adjust the corresponding parameters.

[0086] According to an embodiment of the present invention, the construction of a transformation model based on a neural network model, wherein the first remote sensing data is imported into the transformation model for dimensionality reduction to form intermediate data, specifically includes:

[0087] The first remote sensing data transformation model is used to reduce the dimensionality of the data to obtain intermediate data;

[0088] Calculate the structural similarity index S between the first remote sensing data and the intermediate data. If S is lower than the preset threshold, then perform data dimensionality reduction again.

[0089] It should be noted that the structural similarity index S is used to calculate the reconstruction error and determine whether important features are retained after the data is reduced in dimensionality.

[0090] According to an embodiment of the present invention, the step of transmitting intermediate data to a preset terminal and then performing dimensionality upscaling on the intermediate data based on a transformation model to obtain second remote sensing data specifically involves:

[0091] Intermediate data is transmitted to the user's preset terminal via satellite terminal;

[0092] On the user's preset terminal, intermediate data is imported into the conversion model, and the data is upgraded and restored based on the decoder in the conversion model to generate second remote sensing data.

[0093] It should be noted that the conversion model is a shared model between the satellite terminal and the user's preset terminal.

[0094] According to an embodiment of the present invention, the observation area based on real-time acquired remote sensing data is used to obtain comparative remote sensing imaging data. Image features are extracted from the comparative remote sensing imaging data based on low-feature regions to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. Specifically:

[0095] Based on the observation area of ​​real-time acquired remote sensing data, comparative remote sensing imaging data is obtained;

[0096] Based on low-feature regions, the comparative remote sensing imaging data are labeled accordingly to obtain the region to be analyzed.

[0097] Based on the region to be analyzed, image data of the corresponding region is obtained from the comparative remote sensing imaging data. The obtained image data is then subjected to SIFT-based feature extraction to obtain corrected feature data.

[0098] The second remote sensing data is used for imaging analysis to obtain the target image;

[0099] The DCA feature fusion algorithm is used to fuse features of the target image based on the corrected feature data, and to form real-time transmitted remote sensing image data.

[0100] Figure 3 A block diagram of a remote sensing satellite data transmission system according to the present invention is shown.

[0101] A second aspect of the present invention also provides a remote sensing satellite data transmission system 3, the system comprising: a memory 31 and a processor 32, wherein the memory includes a remote sensing satellite data transmission program, and the remote sensing satellite data transmission program, when executed by the processor, performs the following steps:

[0102] Real-time acquisition of hyperspectral remote sensing satellite data; preprocessing of the remote sensing satellite data to form first remote sensing data.

[0103] Based on the first remote sensing data, remote sensing imaging is performed, and a region clustering analysis based on image features is conducted using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them.

[0104] A transformation model based on a neural network model is constructed. The first remote sensing data is imported into the transformation model to perform data dimensionality reduction and form intermediate data.

[0105] The intermediate data is transmitted to a preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data;

[0106] Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Based on the low feature area, image features are extracted from the comparative remote sensing imaging data to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data.

[0107] The real-time remote sensing image data is sent to a preset terminal device.

[0108] According to an embodiment of the present invention, the real-time acquisition of hyperspectral remote sensing satellite data and the preprocessing of the remote sensing satellite data to form first remote sensing data specifically include:

[0109] Real-time acquisition of hyperspectral remote sensing satellite data and labeling of the hyperspectral remote sensing satellite data as real-time remote sensing data;

[0110] The real-time remote sensing data is preprocessed by data cleaning, noise reduction, and radiometric calibration to obtain the first remote sensing data.

[0111] It should be noted that the real-time acquisition of hyperspectral remote sensing satellite data refers to data acquisition at the satellite terminal, and the corresponding observation target area refers to the observation area.

[0112] According to an embodiment of the present invention, the step of performing remote sensing imaging based on the first remote sensing data and performing region clustering analysis based on image features using a preset clustering algorithm to obtain multiple groups of image regions and then filtering out low-feature regions specifically involves:

[0113] Resampling and image synthesis are performed on the first remote sensing data to obtain initial image data;

[0114] The initial image data is divided into grid regions to form multiple sub-regions and corresponding image data for multiple regions.

[0115] The image data of the multiple regions are subjected to SIFT-based feature extraction to obtain multiple feature data;

[0116] The multiple feature data are used as clustering sample data. Based on the DBSCAN clustering algorithm, the sample data are clustered and grouped to form multiple groups of image regions.

[0117] It should be noted that each of the multiple image regions includes at least one sub-region, and generally more than two. By clustering the regions, the regional image data in all the corresponding sub-regions of each region group have similar characteristics, including similarity in noise ratio. That is, after grouping, some regions with more noise and lower features can be filtered out, so that remote sensing image enhancement can be performed based on the low-feature regions in the subsequent process.

[0118] It is worth mentioning that in the real-time transmission of remote sensing data, due to the limitations of the transmitted data, it is necessary to reduce or compress the data volume as much as possible and transmit the data between the satellite and the ground terminal. Because the original hyperspectral data has a high dimension, it is difficult to transmit all the data. However, after compression and dimensionality reduction, there may be image distortion or excessive noise in the remote sensing image of a certain area, resulting in unsatisfactory results in some areas. Therefore, this invention performs imaging analysis on the original remote sensing data and, based on a clustering algorithm, performs feature analysis and clustering grouping on different image regions. This effectively classifies potential noise points, significantly improving the effectiveness and targeting of subsequent feature fusion and image enhancement. Existing technologies lack preliminary analysis and clustering of the original imaging images, making it difficult to identify transmission defect areas and perform targeted feature fusion enhancement. Furthermore, after dimensionality reduction and transmission to ground users is interrupted, there is some data loss in the dimensionality-reduced data, and the restored data may still contain distortion and loss of detail in certain areas. Therefore, after obtaining the second remote sensing data, this invention pre-acquires corresponding comparison images for low-feature areas (the comparison images are historical remote sensing data images from Earth observations, which have a certain reference value), and extracts corresponding feature data through aggressive DCA-based feature fusion to obtain corrected imaging data. This significantly improves the degree of image detail restoration. During dimensionality-reduction-based transmission, it effectively ensures the stability of real-time transmission and the integrity of important remote sensing data features, greatly enhancing the further application of remote sensing technology.

[0119] According to an embodiment of the present invention, the step of performing remote sensing imaging based on the first remote sensing data and performing region clustering analysis based on image features using a preset clustering algorithm to obtain multiple groups of image regions and filter out low-feature regions from them further includes:

[0120] Based on the multiple sets of image regions, the image feature data of each set of image regions are integrated to form multiple sets of integrated feature data;

[0121] By using the threshold method, pixel-level noise data statistics and proportion calculations are performed on the integrated feature data, and the noise proportion value is obtained.

[0122] Multiple noise percentage values ​​were obtained by analyzing multiple sets of integrated feature data;

[0123] The image regions with the highest noise ratio are marked and these regions are designated as low-feature regions.

[0124] It should be noted that, among the multiple noise proportion values, each group of image regions has one noise proportion value. When marking the group of image regions with the largest noise proportion value, a corresponding group of image regions is obtained. Since a group of image regions includes multiple sub-regions, the corresponding multiple sub-regions are summed to form a large region as a low-feature region. Assuming there are a total of N groups of image regions, N noise proportion values ​​are optimally obtained.

[0125] The noise percentage reflects the proportion of distorted and noisy points in an image region. The higher the noise percentage, the more noisy points there are in the image region, and the lower the amount of image representation feature data. This makes it easier for data loss to occur during dimensionality reduction. Therefore, special labeling is required.

[0126] According to an embodiment of the present invention, the construction of a transformation model based on a neural network model, wherein the first remote sensing data is imported into the transformation model for dimensionality reduction to form intermediate data, specifically includes:

[0127] A conversion model based on a neural network model is constructed, wherein the conversion model is based on an autoencoder and includes an encoder and a decoder;

[0128] Lossless historical remote sensing data is obtained from the system database, and the lossless historical remote sensing data is imported into the conversion model. The data is reduced in dimensionality and compressed by the encoder, and restored in dimensionality by the decoder. The parameters are optimized by joint training. In the joint training, the loss function is set to optimize and adjust the parameters of the encoder and decoder based on the mean square error.

[0129] The data is trained cyclically using lossless historical remote sensing data until the data features represented in the low-dimensional representation of the transformation model meet expectations.

[0130] It should be noted that the system database is used to store historical data records, and the lossless historical remote sensing data specifically refers to data records that have been pre-selected by the user, which are used to train the conversion model and adjust the corresponding parameters.

[0131] According to an embodiment of the present invention, the construction of a transformation model based on a neural network model, wherein the first remote sensing data is imported into the transformation model for dimensionality reduction to form intermediate data, specifically includes:

[0132] The first remote sensing data transformation model is used to reduce the dimensionality of the data to obtain intermediate data;

[0133] Calculate the structural similarity index S between the first remote sensing data and the intermediate data. If S is lower than the preset threshold, then perform data dimensionality reduction again.

[0134] It should be noted that the structural similarity index S is used to calculate the reconstruction error and determine whether important features are retained after the data is reduced in dimensionality.

[0135] According to an embodiment of the present invention, the step of transmitting intermediate data to a preset terminal and then performing dimensionality upscaling on the intermediate data based on a transformation model to obtain second remote sensing data specifically involves:

[0136] Intermediate data is transmitted to the user's preset terminal via satellite terminal;

[0137] On the user's preset terminal, intermediate data is imported into the conversion model, and the data is upgraded and restored based on the decoder in the conversion model to generate second remote sensing data.

[0138] It should be noted that the conversion model is a shared model between the satellite terminal and the user's preset terminal.

[0139] According to an embodiment of the present invention, the observation area based on real-time acquired remote sensing data is used to obtain comparative remote sensing imaging data. Image features are extracted from the comparative remote sensing imaging data based on low-feature regions to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. Specifically:

[0140] Based on the observation area of ​​real-time acquired remote sensing data, comparative remote sensing imaging data is obtained;

[0141] Based on low-feature regions, the comparative remote sensing imaging data are labeled accordingly to obtain the region to be analyzed.

[0142] Based on the region to be analyzed, image data of the corresponding region is obtained from the comparative remote sensing imaging data. The obtained image data is then subjected to SIFT-based feature extraction to obtain corrected feature data.

[0143] The second remote sensing data is used for imaging analysis to obtain the target image;

[0144] The DCA feature fusion algorithm is used to fuse features of the target image based on the corrected feature data, and to form real-time transmitted remote sensing image data.

[0145] A third aspect of the present invention also provides a computer-readable storage medium including a remote sensing satellite data transmission program, which, when executed by a processor, implements the steps of the remote sensing satellite data transmission method as described in any of the preceding claims.

[0146] This invention discloses a remote sensing satellite data transmission and analysis method and a system storage medium. The method involves real-time acquisition of first remote sensing data for remote sensing imaging, followed by region clustering analysis based on image features using a preset clustering algorithm to select low-feature regions. The first remote sensing data is then imported into a transformation model for dimensionality reduction, forming intermediate data. This intermediate data is transmitted to a preset terminal, and its dimensionality is increased based on the transformation model to obtain second remote sensing data. Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Image features are extracted from the comparative remote sensing imaging data based on the low-feature regions to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and real-time transmitted remote sensing image data is obtained based on a DCA feature fusion algorithm. This real-time transmitted remote sensing image data is then sent to the preset terminal device. This invention effectively achieves stable real-time transmission and ensures the integrity of remote sensing data features.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and 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. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0148] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0149] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0150] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0152] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A remote sensing satellite data transmission method, characterized in that, include: Real-time acquisition of hyperspectral remote sensing satellite data; preprocessing of the remote sensing satellite data to form first remote sensing data. Based on the first remote sensing data, remote sensing imaging is performed, and a region clustering analysis based on image features is conducted using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them. A transformation model based on a neural network model is constructed. The first remote sensing data is imported into the transformation model to perform data dimensionality reduction and form intermediate data. The intermediate data is transmitted to a preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data; Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Based on the low feature area, image features are extracted from the comparative remote sensing imaging data to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. The real-time remote sensing image data is sent to a preset terminal device.

2. The remote sensing satellite data transmission method according to claim 1, characterized in that, The real-time acquisition of hyperspectral remote sensing satellite data, followed by preprocessing of the remote sensing satellite data to form the first remote sensing data, specifically involves: Real-time acquisition of hyperspectral remote sensing satellite data and labeling of the hyperspectral remote sensing satellite data as real-time remote sensing data; The real-time remote sensing data is preprocessed by data cleaning, noise reduction, and radiometric calibration to obtain the first remote sensing data.

3. The remote sensing satellite data transmission method according to claim 2, characterized in that, The process of performing remote sensing imaging based on the first remote sensing data and conducting region clustering analysis based on image features using a preset clustering algorithm to obtain multiple groups of image regions and then filtering out low-feature regions is as follows: Resampling and image synthesis are performed on the first remote sensing data to obtain initial image data; The initial image data is divided into grid regions to form multiple sub-regions and corresponding image data for multiple regions. The image data of the multiple regions are subjected to SIFT-based feature extraction to obtain multiple feature data; The multiple feature data are used as clustering sample data. Based on the DBSCAN clustering algorithm, the sample data are clustered and grouped to form multiple groups of image regions.

4. The remote sensing satellite data transmission method according to claim 3, characterized in that, The step of performing remote sensing imaging based on the first remote sensing data and conducting region clustering analysis based on image features using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them also includes: Based on the multiple sets of image regions, the image feature data of each set of image regions are integrated to form multiple sets of integrated feature data; By using the threshold method, pixel-level noise data statistics and proportion calculations are performed on the integrated feature data, and the noise proportion value is obtained. Multiple noise percentage values ​​were obtained by analyzing multiple sets of integrated feature data; The image regions with the highest noise ratio are marked and these regions are designated as low-feature regions.

5. A remote sensing satellite data transmission method according to claim 4, characterized in that, The construction of a transformation model based on a neural network model involves importing the first remote sensing data into the transformation model for dimensionality reduction to form intermediate data. Specifically: A conversion model based on a neural network model is constructed, wherein the conversion model is based on an autoencoder and includes an encoder and a decoder; Lossless historical remote sensing data is obtained from the system database, and the lossless historical remote sensing data is imported into the conversion model. The data is reduced in dimensionality and compressed by the encoder, and restored in dimensionality by the decoder. The parameters are optimized by joint training. In the joint training, the loss function is set to optimize and adjust the parameters of the encoder and decoder based on the mean square error. The data is trained cyclically using lossless historical remote sensing data until the data features represented in the low-dimensional representation of the transformation model meet expectations.

6. A remote sensing satellite data transmission method according to claim 5, characterized in that, The construction of a transformation model based on a neural network model involves importing the first remote sensing data into the transformation model for dimensionality reduction to form intermediate data. Specifically: The first remote sensing data transformation model is used to reduce the dimensionality of the data to obtain intermediate data; Calculate the structural similarity index S between the first remote sensing data and the intermediate data. If S is lower than the preset threshold, then perform data dimensionality reduction again.

7. A remote sensing satellite data transmission method according to claim 6, characterized in that, The process of transmitting intermediate data to a preset terminal and then upscaling the intermediate data based on a transformation model to obtain second remote sensing data specifically involves: Intermediate data is transmitted to the user's preset terminal via satellite terminal; On the user's preset terminal, intermediate data is imported into the conversion model, and the data is upgraded and restored based on the decoder in the conversion model to generate second remote sensing data.

8. A remote sensing satellite data transmission method according to claim 7, characterized in that, The observation area based on real-time acquired remote sensing data is used to obtain comparative remote sensing imaging data. Image features are extracted from the comparative remote sensing imaging data based on low-feature areas to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. Specifically: Based on the observation area of ​​real-time acquired remote sensing data, comparative remote sensing imaging data is obtained; Based on low-feature regions, the comparative remote sensing imaging data are labeled accordingly to obtain the region to be analyzed. Based on the region to be analyzed, image data of the corresponding region is obtained from the comparative remote sensing imaging data. The obtained image data is then subjected to SIFT-based feature extraction to obtain corrected feature data. The second remote sensing data is used for imaging analysis to obtain the target image; The DCA feature fusion algorithm is used to fuse features of the target image based on the corrected feature data, and to form real-time transmitted remote sensing image data.

9. A remote sensing satellite data transmission system, characterized in that, The system includes: a memory and a processor. The memory includes a remote sensing satellite data transmission program. When the remote sensing satellite data transmission program is executed by the processor, it performs the following steps: Real-time acquisition of hyperspectral remote sensing satellite data; preprocessing of the remote sensing satellite data to form first remote sensing data. Based on the first remote sensing data, remote sensing imaging is performed, and a region clustering analysis based on image features is conducted using a preset clustering algorithm to obtain multiple sets of image regions and filter out low-feature regions from them. A transformation model based on a neural network model is constructed. The first remote sensing data is imported into the transformation model to perform data dimensionality reduction and form intermediate data. The intermediate data is transmitted to a preset terminal, and the intermediate data is upgraded based on the transformation model to obtain the second remote sensing data; Based on the observation area of ​​the real-time acquired remote sensing data, comparative remote sensing imaging data is obtained. Based on the low feature area, image features are extracted from the comparative remote sensing imaging data to obtain corrected feature data. Imaging analysis is performed on the second remote sensing data, and the imaging data is fused with the corrected feature data based on the DCA feature fusion algorithm to obtain real-time transmitted remote sensing image data. The real-time remote sensing image data is sent to a preset terminal device.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a remote sensing satellite data transmission program, which, when executed by a processor, implements the steps of the remote sensing satellite data transmission method as described in any one of claims 1 to 8.

Images