Remote sensing image water Rayleigh scattering proximity effect correction method and device

Abstract

The invention provides a remote sensing image water Rayleigh scattering proximity effect correction method. The method comprises the following steps: constructing a data set of a water Rayleigh scattering proximity effect correction model for training; generating a structured training sample based on the data set and the observation conditions, atmosphere and environmental parameters corresponding to the data set to train at least one preset model; collecting satellite remote sensing image data, judging a water body region type of the satellite remote sensing image data, and extracting corresponding water body morphological characteristic parameters from the satellite remote sensing image data; and inputting the water body morphological characteristic parameters, observation conditions and atmosphere and environment parameters as characteristic values into the at least one model to obtain a Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data. The invention discloses a remote sensing image water Rayleigh scattering proximity effect correction device, electronic equipment applying the method and a computer readable storage medium in an exemplarily manner.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water color remote sensing monitoring, in particular to a remote sensing image water body Rayleigh scattering adjacent effect correction method, device and computer readable storage medium. BACKGROUND

[0002] In the field of water color remote sensing monitoring, accurate extraction of water-leaving radiance through atmospheric correction is the core cornerstone of water quality evaluation, ecological environment monitoring and other research and application. At present, a series of mature atmospheric correction algorithms have been developed for open ocean water bodies, but when applied to remote sensing images, the interference of land adjacent effect can seriously affect the accuracy of water body signal extraction, making it difficult for traditional atmospheric correction algorithms to achieve the expected effect.

[0003] In actual monitoring scenarios, the coastal area has complex terrain and various types of land and water interface; and the inland water body, especially the narrow river, is often surrounded by high reflectivity land. These complex environments make it difficult to accurately extract water-leaving radiance from satellite-observed water body pixels affected by the reflection of surrounding adjacent land. In addition, sea ice or water bodies around islands are also affected by adjacent effects. For example, islands and sea ice areas exhibit "water-land" or "water-ice" structural characteristics, i.e., water body pixels are surrounded by high reflectivity targets (such as ice cover, sand islands, etc.), and Rayleigh scattering adjacent effect presents radial and multi-directional characteristics; in addition, the sea ice surface reflectivity is high, often exceeding 0.6 in the visible light band, making water body pixels more susceptible to strong adjacent effects. In addition, Rayleigh scattering adjacent effect is significantly affected by the sun and satellite observation angle. Under different combinations of solar zenith angle and satellite observation zenith angle, the amount of adjacent scattering radiation received by water body pixels differs significantly, and shows high sensitivity to observation direction, which undoubtedly further increases the complexity and difficulty of data processing. At present, although some adjacent effect correction algorithms have been proposed, most of them focus on removing aerosol scattering and lack systematic consideration of Rayleigh scattering adjacent effect. Therefore, it is of great significance to develop a Rayleigh scattering adjacent effect correction method and system specifically for remote sensing image water body pixels to improve the quality of water body optical remote sensing data. SUMMARY

[0004] The present application provides a remote sensing image water body Rayleigh scattering adjacent effect correction method, device and computer readable storage medium, which can improve the quality of water body optical remote sensing data and is of great significance for quickly and accurately obtaining water body optical remote sensing information.

[0005] In a first aspect, a method for correcting a water body Rayleigh scattering proximity effect in remote sensing images is provided, including: constructing a data set for training a water body Rayleigh scattering proximity effect correction model, wherein the data set includes parameters: Rayleigh scattering optical thickness, solar zenith angle, observation zenith angle, relative azimuth angle, wind speed, surface reflectivity, and water body morphological parameters; generating a structured training sample based on the data set and corresponding observation conditions, atmospheric and environmental parameters, to train a preset at least one model, and setting a preset training end condition of the at least one model as: taking the radiance in a pure water body case as a reference, calculating a ratio of radiance with proximity effect to the reference, constructing a target correction factor, and repeating the process until the ratio reaches a predetermined numerical range; collecting satellite remote sensing image data, obtaining Rayleigh scattering optical thickness, observation zenith angle, and relative azimuth angle of each waveband of the image according to information of the image, obtaining wind speed data from observation data, and judging a water body region type of the satellite remote sensing image data, and extracting corresponding water body morphological characteristic parameters therefrom, wherein the water body morphological characteristic parameters include: water body type, morphological parameter, and spatial structure parameter; inputting the water body morphological characteristic parameters and observation conditions, atmospheric and environmental parameters as characteristic values into the at least one model, to obtain a Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data.

[0006] In some embodiments, constructing the data set for training the water body Rayleigh scattering proximity effect correction model includes: based on the data set, setting a plurality of parameter combinations, inputting them into an NUS-MC model, and obtaining atmospheric top Rayleigh scattering radiance in corresponding cases; and for a composite water body scene with a land proximity effect, further introducing a surface reflectivity parameter and a water body morphological parameter for simulation.

[0007] In some embodiments, for a composite water body scene with a land proximity effect, the surface reflectivity parameter and the water body morphological parameter are further introduced for simulation, including: for a river type water body, expressing spatial morphological characteristics thereof by a river width, a pixel distance from a shore, and a river direction angle; or for a lake type water body, dividing it into two categories: a long and narrow lake and a circular lake, wherein the long and narrow lake is expressed by the same morphological parameters as the river; and the circular lake is expressed by a lake radius and a pixel distance from a shore as main characteristics; or for a coastal near-shore water body, including: expressing water body morphology by a distance from a shore to simulate changes in a degree of land radiation interference; or for an island or sea ice region, including: expressing a distance from a water body pixel to a geometric center of an island or sea ice, and a shortest distance to a boundary, as characteristics for modeling Rayleigh scattering proximity interference.

[0008] In some embodiments, the at least one model includes: XGBoost, CatBoost, lightGBM, random forest, support vector machine, and neural network.

[0009] In some embodiments, collecting satellite remote sensing image data includes: preprocessing the satellite remote sensing image data to be processed, including radiometric calibration.

[0010] In some embodiments, the method further includes: performing pixel-by-pixel correction on the radiance value in the original satellite image according to the Rayleigh scattering proximity effect correction factor.

[0011] In some embodiments, the method further includes: outputting Rayleigh scattering corrected images for remote sensing analysis.

[0012] In the second aspect, the method for correcting the Rayleigh scattering proximity effect of water bodies in the aforementioned remote sensing images includes: a radiation transmission data simulation module for constructing a data set for training a water body Rayleigh scattering proximity effect correction model, wherein the data set includes parameters: Rayleigh scattering optical thickness, solar zenith angle, observation zenith angle, relative azimuth, wind speed, surface reflectivity, and water body morphological parameters; a machine learning model training module for generating structured training samples based on the data set and the observation conditions, atmospheric and environmental parameters corresponding to the data set, for training at least one preset model, and setting the preset training end condition of the at least one model to: taking the radiance in the pure water body case as a benchmark, calculating the ratio of the radiance containing the proximity effect to it, and constructing a target correction model. a positive factor until the ratio reaches a predetermined numerical range; a satellite image water body and morphological feature extraction module, used to collect satellite remote sensing image data, obtain the Rayleigh scattering optical thickness, observation zenith angle, and relative azimuth of each band of the image according to the image information, obtain wind speed data from the observation data, and determine the water body area type of the satellite remote sensing image data, and extract corresponding water body morphological characteristic parameters therefrom, wherein the water body morphological characteristic parameters include: water body type, morphological parameters, and spatial structure parameters; a satellite image correction module, used to input the water body morphological characteristic parameters and observation conditions, atmospheric and environmental parameters as characteristic values ​​into the at least one model to obtain the Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data.

[0013] In a third aspect, the present invention provides an electronic device comprising: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the above-mentioned method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images is implemented.

[0014] In a fourth aspect, the present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores program code, and the program code can be called by a processor to execute the above-mentioned remote sensing image water body Rayleigh scattering proximity effect correction method.

[0015] Compared with the prior art, the present application can at least achieve one of the following beneficial effects: Based on the physical model, multi-scene simulation data is generated to improve the accuracy and reliability of Rayleigh scattering proximity effect correction.

[0016] Secondly, machine learning method is introduced to realize automatic training and high adaptability of the correction model, which greatly improves the calculation efficiency under the premise of ensuring the calculation accuracy.

[0017] Thirdly, the water body type and morphological characteristics are considered to improve the correction accuracy of different water body regions.

[0018] Fourthly, an end-to-end remote sensing image automatic correction process is constructed to improve the processing efficiency and be applicable to various remote sensing image application scenarios.

[0019] The summary is provided to introduce selected concepts of the present disclosure in a simplified form, which will be further described below in the detailed description. The summary is not intended to identify key or essential features of the disclosure, nor is it intended to limit the scope of the disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which like reference characters refer to like parts throughout the figures, in exemplary embodiments of the present disclosure.

[0021] Figure 1 A schematic diagram of a remote sensing image water body Rayleigh scattering proximity effect correction method provided by an embodiment of the present application is shown; Figure 2 A structural block diagram of an apparatus for a remote sensing image water body Rayleigh scattering proximity effect correction method provided by an embodiment of the present application is shown; Figure 3 A schematic diagram of an electronic device for a remote sensing image water body Rayleigh scattering proximity effect correction method provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0022] Embodiments of the present disclosure will be described in more detail by making reference to the accompanying drawings. Although embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0023] As used herein, the term "including" and its variations represent open inclusion, i.e., "including but not limited to." Unless otherwise stated, the term "or" means "and / or." The term "based on" means "based at least in part on." The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment." The term "another embodiment" means "at least one additional embodiment." The terms "first," "second," etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0024] This application provides a method for correcting the proximity effect of Rayleigh scattering in water bodies in remote sensing images. Figure 1 , which is a schematic diagram of the first embodiment of this application. Figure 1 A method 100 for correcting Rayleigh scattering proximity effect in water bodies in remote sensing images is described in detail in the first embodiment of the present application.

[0025] Step S102: Simulating radiative transfer data, i.e., constructing a dataset for training a water Rayleigh scattering proximity effect correction model. The dataset includes parameters such as Rayleigh scattering optical depth, solar zenith angle, observation zenith angle, relative azimuth, wind speed, surface reflectivity, and water morphological parameters. Rayleigh scattering optical depth, solar zenith angle, observation zenith angle, relative azimuth, and wind speed are observation geometry and atmospheric parameters, while surface reflectivity is an environmental parameter. This allows the present invention to generate multi-scenario simulation data based on a physical model, improving the accuracy and reliability of water Rayleigh scattering proximity effect correction.

[0026] Specifically, the proximity effect can affect the accuracy of remote sensing inversion of water bodies in inland and coastal areas, where land and water interact. This effect is influenced by a variety of environmental factors, including Rayleigh scattering optical depth, solar-observation geometry, surface type, and water morphology. Therefore, this paper first constructs a radiance simulation dataset under a wide range of conditions based on the NUS-MC non-uniform surface water-air coupled radiative transfer model.

[0027] When simulating the Rayleigh scattering radiance of the underlying surface of a pure water body, different parameter combinations are set, including the step size change of the Rayleigh scattering optical depth, the solar zenith angle, the wind speed, the observation zenith angle, and the observation azimuth, and are input into the NUS-MC model to obtain the Rayleigh scattering radiance of the top of the atmosphere under the corresponding circumstances. Then, for the composite water scene with the land proximity effect, based on the above parameters, surface reflectance parameters and water body morphology parameters are further introduced for simulation. The water body morphology modeling method is as follows: 1) River type water body: Its spatial morphological characteristics are represented by river width, pixel distance from shore (i.e., the shortest distance from the pixel to the nearest shoreline), and river direction angle; 2) Lake type water body: divided into two types of long and round. Long lake adopts the same morphological parameter as river; round lake adopts lake radius and pixel distance from shore as main features; 3) Coastal zone nearshore water body, including: using distance from shore as a parameter to characterize water body morphology to simulate the change of the degree of interference from land radiation.

[0028] 4) Island or sea ice area, as the island and sea ice area show the structural characteristics of "water-enclosed island" or "water-enclosed ice". Therefore, some embodiments can include: using the distance from the water body pixel to the geometric center of the island or sea ice and the shortest distance to the boundary as the features for modeling the adjacent interference of Rayleigh scattering. In some embodiments, NDWI and other indicators can also be used to identify water body areas to form the required spatial feature combination for modeling. In this way, by extracting geometric features such as the distance from the water body pixel to the center of the target area and the distance to the boundary, the spatial distribution characteristics of the radial multi-directional interference are effectively characterized, and the physical rationality and prediction accuracy of the adjacent effect modeling are significantly improved.

[0029] Through the above parameter setting and multi-scene simulation, atmospheric top Rayleigh scattering radiance data covering a wide range of environmental changes can be systematically generated, providing a high reliability physical basis for subsequent model training and image correction.

[0030] Preferably, the present application first constructs a large-scale simulation data set covering various environments and observation conditions based on the NUS-MC non-uniform underlying surface water-air coupling radiation transfer model, which is used to train the adjacent effect correction model. The input parameters involved include (the range of each parameter is only an example, which can be set independently, and is not limited to the following range): (1) Rayleigh scattering optical thickness: [0.001, 1.0]; solar zenith angle: [0°, 75°]; observation zenith angle: [0°, 75°]; relative azimuth angle: [0°, 180°]; wind speed: [0 m / s, 30 m / s]; surface reflectivity: [0.01, 1.0]; (2) Water body morphological parameters ① River: river width: [10, 20000] meters; pixel distance from shore to river width ratio: [0, 1]; river direction angle: [0, 90] ② Lake: lake radius: [10, 10000] meters; pixel distance from shore to river width ratio: [0, 1]; ③ Coastal zone: distance from shore: [10, 20000] meters (3) Spatial structure parameters (applicable to "water-enclosed island / ice" scenarios): island / sea ice planar target area: [1, 1000] km 2; Water pixel to target region geometric centroid distance: [0, 50 km]; Shortest distance from water pixel to target boundary: [0, 10 km].

[0031] The Latin hypercube sampling method is used to generate about 50 million sets of simulation data (which can be set autonomously) to simulate the Rayleigh scattering radiance of pure water bodies and water-land interaction conditions according to the specified step length. The data set is not dependent on a specific sensor and has good universality and portability.

[0032] Step S104: training the machine learning model, i.e., generating a structured training sample based on the data set and the observation conditions, atmospheric and environmental parameters corresponding to the data set, training the preset at least one model, and setting the preset training end condition of the at least one model as follows: taking the radiance in the pure water body case as a reference, calculating the ratio of the radiance with the adjacent effect to the radiance in the pure water body case to construct a target correction factor until the ratio reaches a predetermined numerical range. In this way, the present application can realize automatic training and high adaptability of the correction model by introducing a machine learning method, and also achieve a balance between calculation accuracy and efficiency.

[0033] Specifically, the radiance data and the corresponding environmental parameters (water body type, morphological parameters, observation geometry, etc.) simulated in step S102 are arranged into a structured training sample. The radiance with the adjacent effect is calculated by taking the radiance in the pure water body case as a reference to construct a target correction factor.

[0034] On this basis, a variety of machine learning methods (including but not limited to XGBoost, CatBoost, lightGBM, random forest, support vector machine, neural network, etc.) are used to model the training data. The model performance is evaluated by cross-validation and other methods, and the model with the highest precision and the strongest generalization ability is finally selected as the Rayleigh scattering adjacent effect correction model. The model can automatically output the radiance correction coefficient for a specific scene according to the input parameters, and has good adaptability and real-time reasoning ability.

[0035] Preferably, after completing the radiation transfer simulation, the simulation data is converted into standardized training samples, including: 1) input feature vector: composed of the above-mentioned parameters (Rayleigh scattering optical thickness, solar zenith angle, observation zenith angle, relative azimuth angle, wind speed, surface reflectivity, water body type, water body morphological characteristics, etc.), and the above-mentioned parameter combinations can be subjected to feature engineering; 2) output target value: Rayleigh scattering radiance disturbance value caused by adjacent effect (i.e., the ratio of radiance in the mixed scene to that in the pure water body), which is also a correction coefficient for correcting the Rayleigh scattering adjacent effect The training process is as follows: Sample preprocessing: Normalize each input feature (e.g., min-max normalization), handle missing values and outliers, and group analysis according to water body types.

[0036] Model selection and parameter tuning: Try multiple machine learning models (including but not limited to XGBoost, CatBoost, lightGBM, random forest, support vector machine, neural network, etc.), use cross-validation method to evaluate model accuracy, and adjust hyperparameters. Finally, form a high-precision, high-robustness and high-adaptability proximity effect correction model, which can quickly infer the corresponding correction factor for input parameters.

[0037] Step S106: Extracting satellite image water body and morphological features, that is, collecting satellite remote sensing image data, obtaining Rayleigh scattering optical thickness of each band of the image, observation zenith angle, relative azimuth angle according to the information of the image, obtaining wind speed data from the observation data, and judging the water body region type of the satellite remote sensing image data, and extracting the corresponding water body morphological feature parameters, wherein the water body morphological feature parameters include: water body type, morphological parameter, spatial structure parameter. In this way, the present application can improve the correction accuracy of different water body regions by considering the water body type and morphological characteristics.

[0038] Specifically, the satellite remote sensing image to be processed is preprocessed, including steps such as radiometric calibration. Then, the water body recognition algorithm (such as based on NDWI, water body classification network, etc.) is used to automatically extract the water body region. For the extracted water body region, its type (coastal zone, inland river, lake) is further distinguished, and its corresponding morphological feature parameters are extracted.

[0039] Preferably, taking Sentinel-2 Level-1C image as an example, the remote sensing image data processing flow includes: Preprocessing: Radiometric calibration of the original image, and extraction of observation angle parameters and geographic coordinate information.

[0040] Water body extraction: Extract water body region using NDWI (Normalized Difference Water Index), MNDWI, etc.

[0041] Water body type discrimination and morphological feature extraction: classify water bodies according to DEM, water body distribution range, shoreline vector data or spatial morphological features (such as rivers, lakes, coastal zones), and calculate the following parameters: river width, lake radius; pixel distance from the shore; river or boundary trend angle, etc. These parameters will be important input features in the model inference stage.

[0042] In some embodiments, the target region can be identified and modeled: the sea island or sea ice region is extracted and converted to a vector; adjacent target space merging and gap filling are performed; the geometric center of the merged target region is calculated; the distance of each water body pixel to the center and to the boundary space feature is calculated; and the correction model input required feature vector is formed by combining the observation parameters and the above spatial structure information. This embodiment can train the model through simulation data, the input parameters are constructed based on the observation geometry and the target structure relationship, has good cross-sensor adaptability, and can be widely applied to water body correction tasks of Sentinel-2, Landsat series, GF and other medium and high resolution remote sensing images.

[0043] Step S1061: Model inference and correction application The parameters extracted in step S106 (including water body type, morphological features, observation geometric parameters, surface reflectivity estimated value, etc.) are input into the trained adjacent effect correction model on a pixel-by-pixel basis, and the specific process is as follows: Model input preparation: the complete parameter features of each water body pixel are combined into a vector and input into the machine learning model.

[0044] Correction factor prediction: the model outputs the adjacent effect correction factor (i.e., the correction coefficient that needs to be multiplied from the original radiance).

[0045] Radiance correction: the predicted correction factor is applied to the Rayleigh scattering radiance band of the original image, and adjusted on a pixel-by-pixel basis to obtain the correction result without the influence of adjacent effects.

[0046] Result output: the Rayleigh scattering corrected image is output, which can be directly used for subsequent remote sensing analysis tasks such as water color inversion, water body parameter extraction, and water quality evaluation.

[0047] Step S108: correcting satellite images, i.e., inputting the water body morphological feature parameters and observation conditions, atmospheric and environmental parameters as feature values into the at least one model to obtain the Rayleigh scattering adjacent effect correction factor of each pixel of the satellite remote sensing image data. In this way, the present application can improve the processing efficiency by constructing an end-to-end remote sensing image automatic correction process, and is applicable to various remote sensing image application scenarios.

[0048] Specifically, the water body type, morphological parameters, observation geometry and other information obtained in step S106 are input as features into the machine learning correction model trained in step S104 to obtain a Rayleigh scattering proximity effect correction factor for each pixel. According to the correction factor, the radiance values in the original satellite image are corrected pixel by pixel, thereby effectively eliminating or weakening the radiation deviation caused by the interference of land reflection, and realizing that the finally output corrected image under the interference of land proximity effect can be used as high-precision input for subsequent applications such as water body inversion, remote sensing classification, and water quality estimation, thereby improving the environmental adaptability and scientific value of remote sensing data.

[0049] In some embodiments, the data set for training the water body Rayleigh scattering proximity effect correction model is constructed, including: based on the data set, a plurality of parameter combinations are set and input into the NUS-MC model to obtain the Rayleigh scattering radiance at the top of the atmosphere under the corresponding situation; and for a composite water body scene with land proximity effect, a ground reflectivity parameter and a water body morphological parameter are further introduced for simulation.

[0050] In some embodiments, for a composite water body scene with land proximity effect, a ground reflectivity parameter and a water body morphological parameter are further introduced for simulation, including: for a river type water body, the spatial morphological characteristics thereof are represented by a river width, a pixel distance from the shore, and a river direction angle; or for a lake type water body, it is divided into two categories of long and narrow shape and circular shape, wherein the long and narrow lake adopts the same morphological parameters as the river to represent; and the circular lake adopts a lake radius and a pixel distance from the shore as main features; or for a coastal near-shore water body, the water body morphological characteristics are represented by a distance from the shore to simulate the change of the degree of interference of land radiation; or for an island or sea ice area, the distance from the water body pixel to the geometric center of the island or sea ice and the shortest distance to the boundary are used as features for modeling Rayleigh scattering proximity interference.

[0051] In some embodiments, the at least one model includes: XGBoost, CatBoost, lightGBM, random forest, support vector machine, and neural network.

[0052] In some embodiments, the satellite remote sensing image data is collected, including: pre-processing the satellite remote sensing image data to be processed, including radiation calibration.

[0053] In some embodiments, it further includes: correcting the radiance values in the original satellite image pixel by pixel according to the Rayleigh scattering proximity effect correction factor.

[0054] In some embodiments, it further includes: outputting the Rayleigh scattering corrected image for remote sensing analysis.

[0055] The application provides a device 200 for a remote sensing image water body Rayleigh scattering proximity effect correction method. The device 200 is applied to the remote sensing image water body Rayleigh scattering proximity effect correction method described above, and can include the following modules: a radiation transmission data simulation module 202, a machine learning model training module 204, a satellite image water body and morphological feature extraction module 206, and a satellite image correction module 208. Specifically, The radiation transmission data simulation module 202 is configured to construct a data set for training a remote sensing image water body Rayleigh scattering proximity effect correction model, wherein the data set includes parameters: Rayleigh scattering optical thickness, solar zenith angle, observation zenith angle, relative azimuth angle, wind speed, surface reflectivity, and water body morphological parameters.

[0056] For example, based on the NUS-MC non-uniform underlying surface water vapor coupling radiation transmission model, the module simulates the top atmospheric radiance generated by Rayleigh scattering under the condition of considering the land proximity effect. At the same time, the top atmospheric radiance of the pure water underlying surface is simulated under the same condition, so as to construct a radiation transmission data set covering various observation conditions and surface types. The machine learning model training module 204 is configured to generate a structured training sample based on the data set and the observation conditions, atmospheric and environmental parameters corresponding to the data set, to train at least one preset model, and set a preset training end condition of the at least one model as follows: taking the radiance under the pure water condition as a reference, calculating the ratio of the radiance with proximity effect to the radiance under the pure water condition to construct a target correction factor, until the ratio reaches a predetermined numerical range.

[0057] For example, the module pre-processes the simulated data and divides it into a training set and a test set, adopts various mainstream machine learning algorithms (including but not limited to XGBoost, CatBoost, lightGBM, random forest, support vector machine, neural network, etc.) for model training and verification, and finally selects the model with the best performance in terms of accuracy and generalization ability for subsequent image correction tasks.

[0058] The satellite image water body and morphological feature extraction module 206 is configured to collect satellite remote sensing image data, obtain Rayleigh scattering optical thickness, observation zenith angle, and relative azimuth angle of each waveband of the image according to information of the image, obtain wind speed data from the observation data, judge a water body region type of the satellite remote sensing image data, and extract corresponding water body morphological feature parameters therefrom, wherein the water body morphological feature parameters include water body type, morphological parameters, and spatial structure parameters.

[0059] For example, the module automatically extracts the water body area from the remote sensing satellite image, and distinguishes the water body type (such as the coastal zone, inland river, inland lake, etc.) based on the spatial distribution and morphological features. For different water body types, further morphological feature parameters are extracted, which provide important prior information for subsequent model input.

[0060] The satellite image correction module 208 is configured to input the water body morphological feature parameters and observation conditions, atmospheric and environmental parameters as feature values into the at least one model to obtain the Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data.

[0061] For example, the module inputs the water body type and feature information extracted from the satellite image as input, combines the constructed optimal machine learning model, and outputs the Rayleigh scattering radiance correction coefficient for proximity effect correction. Finally, the satellite water body radiance affected by the land proximity is automatically corrected, and the accuracy and reliability of the image in the water color remote sensing inversion are improved.

[0062] As shown in Figure 3 The electronic device can further include a transceiver. The processor and the memory and the transceiver can be connected through a communication bus. The memory stores computer readable instructions, and the computer readable instructions are executed by the processor to implement the steps of the remote sensing image water body Rayleigh scattering proximity effect correction method.

[0063] In a specific implementation, as an example, the processor 320 can include one or more CPUs.

[0064] Optionally, in a specific implementation, if the memory 310, the processor 320 and the communication interface 330 are integrated on a chip, the memory 310, the processor 320 and the communication interface 330 can complete the communication among each other through an internal interface.

[0065] In a specific implementation, as an example, the electronic device can also include multiple processors, for example, each of the processors can be a single-CPU or a multi-CPU. The processor can refer to one or more devices, circuits, and / or processing cores for processing data (for example, computer program instructions).

[0066] The memory is configured to store a software program for executing the scheme of the present application, and the processor is configured to control the execution. The specific implementation can refer to the above method embodiments, and will not be described here.

[0067] The transceiver is configured to communicate with a network device or a terminal device.

[0068] Optionally, the transceiver can include a receiver and a transmitter. Wherein the receiver is configured to implement the receiving function, and the transmitter is configured to implement the transmitting function.

[0069] Optionally, the transceiver can be integrated with the processor, or can exist independently and be coupled with the processor through the interface circuit of the electronic device, and the embodiments of the present application do not make specific limitations thereto.

[0070] It should be noted that the structure of the electronic device described above does not constitute a limitation on the electronic device, and the actual electronic device can include more or fewer components than the illustration, or combine certain components, or different component arrangements. In addition, the technical effects of the electronic device can refer to the technical effects of the above method embodiments, which will not be repeated here.

[0071] In the exemplary embodiments, the present application also provides a computer readable storage medium, and the computer readable storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the steps of the above-mentioned remote sensing image water body Rayleigh scattering proximity effect correction method. For example, the computer readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0072] The embodiments of the present application also provide an electronic device, which comprises: a processor; a memory, and the memory stores computer readable instructions, and the computer readable instructions are executed by the processor to implement the above-mentioned remote sensing image water body Rayleigh scattering proximity effect correction method.

[0073] The embodiments of the present application provide a computer readable storage medium, characterized in that the computer readable storage medium stores program codes, and the program codes can be called and executed by a processor to implement the above-mentioned remote sensing image water body Rayleigh scattering proximity effect correction method.

[0074] It should also be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM) or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM) and direct rambus RAM (DR RAM).

[0075] It should be understood that the term "and / or" herein only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after are an "or" relationship, but can also represent an "and / or" relationship, which can be understood in the context before and after.

[0076] It should be understood that in various embodiments of the present application, the size of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0077] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0078] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0079] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0080] The above has described the embodiments of the present disclosure, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical application or improvement of the technology in the market, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein.

Claims

1. A method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images, characterized in that: include: Constructing a data set for training a water Rayleigh scattering proximity effect correction model, wherein the data set includes parameters: Rayleigh scattering optical depth, solar zenith angle, observation zenith angle, relative azimuth, wind speed, surface reflectivity, and water body morphological parameters; generating structured training samples based on the data set and observation conditions, atmospheric and environmental parameters corresponding to the data set to train at least one preset model, and setting a preset training end condition for the at least one model to: using the radiance in a pure water case as a benchmark, calculating a ratio of the radiance including the proximity effect to the radiance, and constructing a target correction factor until the ratio reaches a predetermined value range; Collect satellite remote sensing image data, obtain the Rayleigh scattering optical depth, observation zenith angle, and relative azimuth of each band of the image based on the image information, obtain wind speed data from the observation data, and determine the water body area type of the satellite remote sensing image data, and extract corresponding water body morphological characteristic parameters therefrom, wherein the water body morphological characteristic parameters include: water body type, morphological parameters, and spatial structure parameters; The water body morphological characteristic parameters and observation conditions, atmospheric and environmental parameters are input into the at least one model as characteristic values ​​to obtain the Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data.

2. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 1, characterized in that: Construct a dataset for training the proximity effect correction model for Rayleigh scattering in water, including: Based on the data set, multiple parameter combinations are set and input into the NUS-MC model to obtain the Rayleigh scattering radiance at the top of the atmosphere under corresponding circumstances; For the complex water body scene with land proximity effect, surface reflectivity parameters and water body morphology parameters are further introduced for simulation.

3. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 2, characterized in that: For complex water body scenarios with land proximity effects, surface reflectance parameters and water body morphology parameters are further introduced for simulation, including: For river-type water bodies, the spatial morphological characteristics are represented by the width of the river channel, the distance from the pixel to the shore, and the angle of the river channel; or For lake-type water bodies, they are divided into two categories: long and narrow and circular. Long and narrow lakes are represented by the same morphological parameters as rivers; circular lakes use the lake radius and the pixel's distance from the shore as the main features; or For coastal waters, this includes: characterizing the water's morphology using offshore distance parameters to simulate changes in its degree of interference from land-based radiation; or For island or sea ice areas, this includes using the distance from the water pixel to the geometric centroid of the island or sea ice, and the shortest distance to the boundary, as features for modeling Rayleigh scattering neighboring interference.

4. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 1, characterized in that: The at least one model includes: XGBoost, CatBoost, lightGBM, random forest, support vector machine and neural network.

5. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 1, characterized in that: Collect satellite remote sensing image data, including: The satellite remote sensing image data to be processed is preprocessed, including radiometric calibration.

6. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 1, characterized in that: Also includes: According to the Rayleigh scattering proximity effect correction factor, the radiance value in the original satellite image is corrected pixel by pixel.

7. The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to claim 6, characterized in that: Also includes: Output Rayleigh scattering corrected images for remote sensing analysis.

8. A device for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images, characterized in that: The method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images as claimed in any one of claims 1 to 7 comprises: A radiation transfer data simulation module is used to construct a data set for training a water body Rayleigh scattering proximity effect correction model, wherein the data set includes parameters: Rayleigh scattering optical depth, solar zenith angle, observation zenith angle, relative azimuth, wind speed, surface reflectivity, and water body morphological parameters; a machine learning model training module, configured to generate structured training samples based on the data set and the observation conditions, atmospheric and environmental parameters corresponding to the data set, for training at least one preset model, and to set a preset training termination condition for the at least one model to: using the radiance in a pure water case as a benchmark, calculating a ratio of the radiance including the proximity effect to the radiance, and constructing a target correction factor until the ratio reaches a predetermined numerical range; The satellite image water body and morphological feature extraction module is used to collect satellite remote sensing image data, obtain the Rayleigh scattering optical thickness, observation zenith angle, and relative azimuth of each band of the image based on the image information, obtain wind speed data from the observation data, and determine the water body area type of the satellite remote sensing image data, and extract the corresponding water body morphological characteristic parameters therefrom, wherein the water body morphological characteristic parameters include: water body type, morphological parameters, and spatial structure parameters; The satellite image correction module is used to input the water body morphological characteristic parameters and observation conditions, atmospheric and environmental parameters as characteristic values ​​into the at least one model to obtain the Rayleigh scattering proximity effect correction factor of each pixel of the satellite remote sensing image data.

9. An electronic device, characterized in that: The electronic device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for correcting Rayleigh scattering proximity effect of water bodies in remote sensing images according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, and the program code can be called by a processor to execute the remote sensing image water body Rayleigh scattering proximity effect correction method according to any one of claims 1 to 7.

Images