Aerosol optical thickness inversion method, device and electronic equipment

Abstract

The application discloses an aerosol optical thickness inversion method and device and electronic equipment, and relates to the technical field of remote sensing. The method comprises the following steps: determining at least one highlight image block in a target remote sensing image to be processed; simulating simulated ground reflectivity of the highlight image block under a plurality of different aerosol optical thicknesses by using an atmospheric radiation transmission model; determining a target ground reflectivity curve of a target highlight ground block corresponding to the highlight image block from a ground reflectivity library of a highlight ground; determining a reference ground reflectivity of the highlight image block at a collection time point by using the target ground reflectivity curve; determining at least one target simulated ground reflectivity matched with the reference ground reflectivity; and fitting a target aerosol optical thickness corresponding to the highlight image block based on the aerosol optical thickness corresponding to the at least one target simulated ground reflectivity. The application can realize relatively accurate inversion of the aerosol optical thickness of the highlight ground.

Description

Technical Field

[0001] This application relates to the field of remote sensing technology, and in particular to an aerosol optical thickness inversion method, apparatus and electronic equipment. Background Technology

[0002] In the field of remote sensing technology, in order to eliminate the influence of atmospheric and lighting factors on the reflection of ground objects, atmospheric correction is required for remote sensing images to obtain true parameters such as surface reflectance. The key to atmospheric correction of remote sensing images is the accurate retrieval of aerosol optical thickness (AOT).

[0003] Currently, the dark pixel method (also known as the dark target method) can be used to invert aerosol optical thickness. However, while this algorithm has high accuracy in densely vegetated areas (i.e., low-brightness surface areas), its accuracy is lower in high-brightness surface areas such as urban areas. Therefore, how to accurately invert the aerosol optical thickness in high-brightness surface areas is a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0004] To address the above issues, this application provides an aerosol optical thickness inversion method, apparatus, and electronic device to achieve a more accurate inversion of the aerosol optical thickness of a bright surface.

[0005] On the one hand, this application provides a method for inverting aerosol optical thickness, including:

[0006] Identify at least one bright image block in the target remote sensing image that belongs to a bright surface;

[0007] For each bright image block, based on the pixel value of each pixel in the bright image block, the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses is simulated using the atmospheric radiative transfer model.

[0008] Based on the latitude and longitude information corresponding to the bright image block, the target surface reflectance curve of the target bright surface block corresponding to the bright image block is determined from the surface reflectance database of bright surfaces. The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time point is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time point.

[0009] Based on the acquisition time point corresponding to the bright image block, the reference surface reflectance of the bright image block at the acquisition time point is determined using the target surface reflectance curve;

[0010] From the simulated surface reflectivities corresponding to the plurality of different aerosol optical thicknesses, determine at least one target simulated surface reflectivity that matches the reference surface reflectivity;

[0011] Based on the aerosol optical thickness corresponding to the simulated surface reflectance of the at least one target, the target aerosol optical thickness corresponding to the high-brightness image block is fitted.

[0012] In one possible implementation, determining at least one target simulated surface reflectance that matches the reference surface reflectance from the plurality of simulated surface reflectances corresponding to different aerosol optical thicknesses includes:

[0013] For each aerosol optical thickness, the spectral angle is calculated based on the simulated surface reflectance and the reference surface reflectance of the high-brightness image block under the aerosol optical thickness, and the spectral angle corresponding to the aerosol optical thickness is obtained.

[0014] From the spectral angles corresponding to multiple different aerosol optical thicknesses, determine the smallest first spectral angle, and the second and third spectral angles with the smallest angle difference from the first spectral angle.

[0015] In another possible implementation, fitting the target aerosol optical thickness corresponding to the high-brightness image block based on the aerosol optical thickness corresponding to the at least one target simulated surface reflectance includes:

[0016] Based on the first aerosol optical thickness AOT1 corresponding to the first spectral angle α1, the second aerosol optical thickness AOT2 corresponding to the second spectral angle α2, and the third aerosol optical thickness AOT3 corresponding to the third spectral angle α3, the first thickness parameter a and the second thickness parameter b are fitted using the following formula:

[0017]

[0018] Based on the first thickness parameter a and the second thickness parameter b, the target aerosol optical thickness corresponding to the high-brightness image block is determined by the following formula:

[0019]

[0020] Wherein, AOT4 is the target aerosol optical thickness corresponding to the high-brightness image block.

[0021] In another possible implementation, the step of simulating the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses based on the pixel value of each pixel in the bright image block using an atmospheric radiative transfer model includes:

[0022] Based on the latitude and longitude information corresponding to the bright image block, the target aerosol mode and target atmospheric mode corresponding to the bright image block are determined.

[0023] Based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block, the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses is simulated using an atmospheric radiative transfer model.

[0024] In another possible implementation, the surface reflectance library of the bright surface includes the surface reflectance curve of at least one bright surface block;

[0025] The surface reflectance curve of the high-brightness surface patch was obtained in the following way:

[0026] An image time series is obtained, which includes reference remote sensing images acquired at multiple reference time points. The reference remote sensing images are remote sensing images after atmospheric correction, and each pixel in the reference remote sensing images has its own surface reflectance.

[0027] For each frame of the reference remote sensing image, remove the pixels that are covered by clouds or snow to obtain the processed reference remote sensing image.

[0028] For each processed baseline remote sensing image, at least one reference image block in the baseline remote sensing image is identified as belonging to at least one bright surface block, wherein each reference image block belongs to one bright surface block.

[0029] For each bright surface patch in the reference remote sensing image corresponding to a reference image block, the average surface reflectance of the bright surface patch at the reference time point corresponding to the reference remote sensing image is determined based on the surface reflectance of each pixel in the reference image block corresponding to the bright surface patch.

[0030] For each bright surface patch, a surface reflectance curve for the bright surface patch is fitted based on the average surface reflectance of the bright surface patch at the multiple reference time points.

[0031] In another possible implementation, determining at least one reference image block in the baseline remote sensing image that belongs to at least one highlighted surface block includes:

[0032] Identify at least one candidate reference image patch in the reference remote sensing image that belongs to a bright surface;

[0033] From the at least one candidate reference image block, at least one reference image block is determined where the proportion of effective pixels exceeds a set threshold. Each reference image block corresponds to a bright surface block, wherein the effective pixel is a pixel with a pixel value.

[0034] In another possible implementation, after fitting the surface reflectance curve of the high-brightness surface patch, the method further includes:

[0035] For each bright surface block, determine the rate of change of surface reflectance at each unit time point in the surface reflectance curve of the bright surface block;

[0036] Based on the rate of change at each unit time point in the surface reflectance curve, determine the maximum and minimum rates of change;

[0037] Based on the maximum and minimum rates of change, a threshold for the maximum rate of change is determined;

[0038] Remove the surface reflectance from the surface reflectance curve that corresponds to the target unit time point that meets the conditions, wherein the rate of change of the surface reflectance corresponding to the target unit time point that meets the conditions exceeds the maximum rate of change threshold, and the rate of change of the surface reflectance corresponding to two adjacent unit time points adjacent to the target unit time point does not exceed the maximum rate of change threshold.

[0039] In yet another possible implementation, the aerosol optical thickness inversion method further includes:

[0040] Identify at least one low-brightness image block in the target remote sensing image that belongs to a low-brightness surface;

[0041] The aerosol thickness of each low-brightness image block was determined using the dark pixel method.

[0042] Atmospheric correction is performed on the target remote sensing image based on the target aerosol thickness in each high-brightness image block and the aerosol thickness in each low-brightness image block.

[0043] In another aspect, this application also provides an aerosol optical thickness inversion device, comprising:

[0044] The bright patch determination unit is used to determine at least one bright image patch in the target remote sensing image to be processed that belongs to a bright surface.

[0045] The reflectance simulation unit is used to simulate the surface reflectance of each bright image block under multiple different aerosol optical thicknesses, based on the pixel value of each pixel in the bright image block and using an atmospheric radiative transfer model.

[0046] The database query unit is used to determine the target surface reflectance curve of the target bright surface block corresponding to the bright image block from the surface reflectance database of bright surfaces based on the latitude and longitude information corresponding to the bright image block. The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time points is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time points.

[0047] The reflectance query unit is used to determine the reference surface reflectance of the high-brightness image block at the acquisition time point based on the acquisition time point corresponding to the high-brightness image block and using the target surface reflectance curve.

[0048] A reflectance matching unit is used to determine at least one target simulated surface reflectance that matches the reference surface reflectance from the plurality of simulated surface reflectances corresponding to different aerosol optical thicknesses.

[0049] The aerosol determination unit is used to fit the target aerosol optical thickness corresponding to the high-brightness image block based on the aerosol optical thickness corresponding to the at least one target simulated surface reflectance.

[0050] In another aspect, this application also provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0051] The memory is used to store computer programs;

[0052] The processor is used to execute the computer program to enable the electronic device to implement the aerosol optical thickness inversion method as described in any of the above.

[0053] In another aspect, this application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the aerosol optical thickness inversion methods provided in the embodiments of this application.

[0054] In another aspect, this application also provides a computer-readable storage medium carrying one or more computer programs that, when executed by an electronic device, enable the electronic device to implement any of the aerosol optical thickness inversion methods provided in the embodiments of this application.

[0055] As described above, this application constructs a surface reflectance database for bright surfaces. For each bright image block in the target remote sensing image that belongs to a bright surface, the target surface reflectance curve of the target bright surface block corresponding to that bright image block is determined from the surface reflectance database. Since the target surface reflectance curve of the target bright surface block is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points, by fitting the curve to the average surface reflectance at multiple reference time points, the influence of noise on surface reflectance can be effectively eliminated, so that the fitted target surface reflectance curve can accurately reflect the surface reflectance changes of the bright surface block. Based on this, after simulating the simulated surface reflectance of the bright image block under different aerosol optical thicknesses using the atmospheric radiative transfer model, and combining the reference surface reflectance corresponding to the acquisition time point of the bright image area determined from the target surface reflectance curve, the aerosol optical height corresponding to the bright image block can be determined more accurately, thus improving the accuracy of determining the aerosol optical thickness of the bright surface area. Attached Figure Description

[0056] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0057] Figure 1 A schematic flowchart of the aerosol optical thickness inversion method provided in this application;

[0058] Figure 2 This is a schematic diagram illustrating one implementation process for determining the surface reflectance curves of each bright surface block in the surface reflectance library for bright surfaces in this application.

[0059] Figure 3 This is another flowchart illustrating the aerosol optical thickness inversion method provided in this application.

[0060] Figure 4 A schematic diagram of one component structure of the aerosol optical thickness inversion device provided in this application;

[0061] Figure 5 A schematic diagram of the component architecture of the electronic device provided in this application. Detailed Implementation

[0062] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is only for explaining specific embodiments and is not intended to limit the application. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0063] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0064] like Figure 1 This embodiment illustrates a flowchart of the aerosol optical thickness inversion method provided in this application. Example The methods may include:

[0065] S101, Identify at least one bright image block in the target remote sensing image to be processed that belongs to a bright surface.

[0066] In this application, for ease of distinction, the remote sensing image to be processed is referred to as the target remote sensing image.

[0067] High-brightness surfaces refer to surface areas with relatively high brightness values ​​in the pixels presented in remote sensing images. It's generally understood that high-brightness surfaces also refer to the surfaces of urban areas. Because urban areas have relatively high reflectivity compared to other land cover types, they appear brighter in remote sensing images; therefore, urban areas are also called high-reflectivity areas or high-brightness surface areas. Accordingly, each high-brightness image block can be a remote sensing image block corresponding to an urban area.

[0068] In this application, at least one bright image block belonging to a bright surface can be determined based on the brightness values ​​of each pixel in the target remote sensing image. For example, the brightness values ​​of pixels exceeding a set percentage or all pixels in each bright image block exceed a set brightness threshold. There are no restrictions on the specific implementation for determining each bright image block in the target remote sensing image.

[0069] S102, For each bright image block, based on the pixel value of each pixel in the bright image block, the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses is simulated using the atmospheric radiative transfer model.

[0070] The header file of the target remote sensing image includes the pixel values ​​of each pixel. The pixel values ​​of a pixel may include, but are not limited to: the pixel's brightness value, angle information, and the sensor position and attitude information corresponding to the pixel.

[0071] The atmospheric radiative transfer model can be the 6S (Second Simulation of Satellite Signal in the Solar Spectrum) radiative transfer model, or any other atmospheric radiative transfer model; there are no restrictions on this.

[0072] In this application, the number of different aerosol optical thicknesses that need to be simulated using the atmospheric radiative transfer model, as well as the specific values ​​of the aerosol optical thicknesses, can be set according to actual needs. For example, the atmospheric radiative transfer model can be used to simulate the surface reflectance of the high-brightness image block when the values ​​of the aerosol optical thicknesses are 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, and 0.9.

[0073] For ease of distinction, the surface reflectance simulated using the atmospheric radiative transfer model is called simulated surface reflectance.

[0074] There are many possible implementations of the surface reflectivity simulation based on the atmospheric radiative transfer model, and this application does not impose any restrictions on them.

[0075] For example, in one possible implementation, the target aerosol model and target atmospheric model corresponding to the bright image block can be determined based on the latitude and longitude information of the bright image block. Accordingly, based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block, an atmospheric radiative transfer model can be used to simulate the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses. For instance, based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block, a 6S radiative transfer model can be used to simulate the surface reflectance of the bright image block under different aerosol optical thicknesses.

[0076] The latitude and longitude information of the highlighted image blocks can also be stored in the image header file of the target remote sensing image.

[0077] Based on the latitude and longitude information corresponding to the bright image block, the corresponding aerosol model and atmospheric model can be determined. Among them, the aerosol model can be divided into model types such as particulate model and desert model, while the atmospheric model can be divided into mid-latitude summer model and mid-latitude autumn model, etc., without restriction.

[0078] In this application, for ease of distinction, the atmospheric mode corresponding to the bright image block is referred to as the target atmospheric mode, and the aerosol mode corresponding to the bright image block is referred to as the target aerosol mode.

[0079] S103, Based on the latitude and longitude information corresponding to the bright image block, determine the target surface reflectance curve of the target bright surface block corresponding to the bright image block from the surface reflectance database of bright surfaces.

[0080] In this application, the surface reflectance library of high-brightness surfaces is pre-constructed, which includes surface reflectance curves of multiple high-brightness surface blocks.

[0081] The surface reflectance curve of the bright surface patch is a curve fitted based on the average surface reflectance of the bright surface patch at multiple reference time points. The average surface reflectance of the bright surface patch at the reference time point is the average of the surface reflectance of each pixel belonging to the bright surface patch in the reference remote sensing image acquired at that reference time point.

[0082] For ease of distinction, the bright surface patch corresponding to the bright image patch is referred to as the target bright surface patch, and the surface reflectance curve of the target bright surface patch is referred to as the target surface reflectance curve. The target bright surface patch refers to the surface patch whose latitude and longitude information matches that of the bright image patch; therefore, the bright image patch is the remote sensing image patch corresponding to the target bright surface patch. Accordingly, the target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface patch at multiple reference time points. The average surface reflectance of the target bright surface patch at the reference time points is the average of the surface reflectance of each pixel belonging to the target bright surface patch in the reference remote sensing image acquired at that reference time point.

[0083] The reference remote sensing image is an atmospherically corrected image; therefore, the surface reflectance of each pixel in the reference remote sensing image is a known and relatively accurate surface reflectance. The specific implementation of atmospheric correction for the reference remote sensing image can be chosen as needed and is not restricted.

[0084] In this application, a surface reflectance database is constructed specifically for high-brightness surfaces. This allows the surface reflectance curves of each high-brightness surface block in the database to more accurately reflect the changes in surface reflectance of the high-brightness surface block at different time points. Therefore, based on this surface reflectance database, reliable data support can be provided for retrieving the aerosol optical thickness of each high-brightness image block, which is beneficial for accurately determining the aerosol optical thickness of each high-brightness image block in the future.

[0085] Furthermore, the surface reflectance database in this application stores the surface reflectance curves of each bright surface block. These surface reflectance curves are fitted based on the average surface reflectance of the bright surface block at multiple reference time points. Curve fitting can effectively eliminate the influence of noise on surface reflectance. Compared to simply listing the discrete surface reflectance of bright surface blocks at different time points, the surface reflectance curves can more accurately represent the true surface reflectance of bright surface blocks.

[0086] S104. Based on the acquisition time point corresponding to the bright image block, the reference surface reflectance of the bright image block at the acquisition time point is determined using the target surface reflectance curve.

[0087] For example, the reference surface reflectance can be obtained by querying the surface reflectance corresponding to the collection time point from the surface reflectance curve of the target.

[0088] It should be noted that steps S103 and S104 can be executed simultaneously with step S102, or step S102 can be executed first and then steps S103 and S104 can be executed; of course, steps S103 and S104 can also be executed first and then step S102, and there are no restrictions on this.

[0089] S105, determine at least one target simulated surface reflectance that matches the reference surface reflectance from multiple simulated surface reflectances corresponding to different aerosol optical thicknesses.

[0090] For example, if there exists a simulated surface reflectance that is identical to the reference surface reflectance, that simulated surface reflectance can be determined as the target simulated surface reflectance.

[0091] For example, considering that in most cases there may not be a simulated surface reflectance that is exactly the same as the reference surface reflectance, in order to accurately determine the aerosol optical thickness of the bright image block, the simulated surface reflectances of two or three targets that are closest to the reference surface reflectance can be determined.

[0092] S106, based on the aerosol optical thickness corresponding to the surface reflectance of at least one target, fit the target aerosol optical thickness corresponding to the high-brightness image block.

[0093] Understandably, if the target simulates only one surface reflectance, then the target simulated surface reflectance and the parameter... The surface reflectivity is the same. In this case, the aerosol optical thickness corresponding to the simulated surface reflectivity of the target is... That's why it should be high. The aerosol optical thickness corresponding to the bright image area, and correspondingly, the fitted target aerosol optical thickness is actually the simulated surface reflectance of the target.

[0094] If there are multiple simulated surface reflectivities for the target, it means that the aerosol optical thickness corresponding to these multiple simulated surface reflectivities is close to the actual aerosol optical thickness corresponding to the bright image block, but none of them are the actual aerosol optical thickness of the bright image block. Based on this, by fitting the aerosol optical thickness corresponding to the simulated surface reflectivities of multiple targets, the target aerosol optical thickness corresponding to the bright image block can be more accurately fitted.

[0095] In this application, there are no restrictions on the specific implementation process of fitting the aerosol optical thickness corresponding to the surface reflectance of at least one target.

[0096] As described above, this application constructs a surface reflectance database for bright surfaces. For each bright image block in the target remote sensing image that belongs to a bright surface, the target surface reflectance curve of the target bright surface block corresponding to that bright image block is determined from the surface reflectance database. Since the target surface reflectance curve of the target bright surface block is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points, by fitting the curve to the average surface reflectance at multiple reference time points, the influence of noise on surface reflectance can be effectively eliminated, so that the fitted target surface reflectance curve can accurately reflect the surface reflectance changes of the bright surface block. Based on this, after simulating the simulated surface reflectance of the bright image block under different aerosol optical thicknesses using the atmospheric radiative transfer model, and combining the reference surface reflectance corresponding to the acquisition time point of the bright image area determined from the target surface reflectance curve, the aerosol optical height corresponding to the bright image block can be determined more accurately, thus improving the accuracy of determining the aerosol optical thickness of the bright surface area.

[0097] It is understood that, simultaneously with or after determining each bright image block, this application may also determine at least one low-brightness image block in the target remote sensing image that belongs to a low-brightness surface, and use the dark pixel method to invert the aerosol thickness of each low-brightness image block. Specifically, the proportion of pixels with brightness values ​​below a set brightness threshold in the low-brightness image block exceeds a set proportion.

[0098] Based on this, this application can perform atmospheric correction on the remote sensing image of the target based on the target aerosol thickness in each high-brightness image block and the aerosol thickness in each low-brightness image block.

[0099] For example, based on the target aerosol thickness in each high-brightness image block and the aerosol thickness in each low-brightness image block, the aerosol thickness of each pixel in the target remote sensing image is obtained through spatial interpolation. Then, atmospheric correction is performed on the target remote sensing image based on the aerosol thickness of each pixel. This application does not limit the specific implementation process of atmospheric correction.

[0100] To facilitate understanding of the scheme in this application, the following section will first introduce one implementation process of constructing the surface reflectance curves of each bright surface block in the surface reflectance library of bright surfaces in this application.

[0101] like Figure 2 This illustration shows a schematic diagram of an implementation process for determining the surface reflectance curve of a bright surface patch according to an embodiment of this application. The process may include:

[0102] S301, obtain image time series.

[0103] The image time series includes reference remote sensing images acquired at multiple reference time points. The reference remote sensing images are atmospherically corrected images, and each pixel in the reference remote sensing images has its own surface reflectance.

[0104] In this application, multiple reference time points for acquiring remote sensing images can be determined according to actual needs, and the designated areas for acquiring remote sensing images can also be identified. For example, if subsequent remote sensing image analysis of the Beijing-Tianjin-Hebei map is required, then the designated area can be the Beijing-Tianjin-Hebei region.

[0105] S302. For each frame of the reference remote sensing image, remove pixels that belong to clouds or snow from the reference remote sensing image to obtain the processed reference remote sensing image.

[0106] Among them, removing pixels belonging to clouds or snow in the reference remote sensing image can be achieved by using a cloud and snow detection algorithm to detect pixels belonging to clouds or snow in the reference remote sensing image and then removing the corresponding pixels.

[0107] In this application, there are no restrictions on the specific implementation of detecting pixels that belong to clouds or snow.

[0108] For example, let's take identifying pixels that belong to the cloud:

[0109] For each frame of reference remote sensing image, this application can first employ an object-oriented cloud and cloud shadow detection (Fmask) algorithm (also known as a functional masking algorithm) to perform potential cloud detection on the reference remote sensing image, thereby identifying potential cloud pixels in the reference remote sensing image. Potential cloud pixels may be pixels belonging to clouds or pixels with clear skies. Conversely, other pixels that do not belong to potential cloud pixels are considered as pixels with absolutely clear skies.

[0110] Potential cloud area detection includes basic testing, whiteness testing, HOT (Haze Optimized Transformation) testing, and band testing.

[0111] Based on the bright and cold characteristics of clouds, potential cloud areas can be distinguished from snow-covered areas and some vegetation through basic testing. The basic test determines potential cloud areas using the following formula: pixels belonging to potential cloud areas must meet the conditions shown in formula 1.

[0112] Band7>0.03 and BT<27 and NDSI<0.8 and NDVI<0.8 (Formula 1);

[0113] Here, "Band" represents a frequency band, which refers to a specific frequency range or wavelength range in the electromagnetic spectrum. Remote sensing images can correspond to multiple bands, and the number of bands can vary depending on the specific application scenario. In this application, "Band" represents the surface reflectance of a pixel in a remote sensing image (such as a reference remote sensing image) within that band. Correspondingly, in Formula 1, Band7 represents the surface reflectance of a pixel in the reference remote sensing image within band 7.

[0114] BT is the brightness temperature band of a pixel. BT < 27 means that the brightness temperature band is less than 27 degrees.

[0115] NDSI, or Normalized Differential Snow Cover Index, is an index used in remote sensing to detect snow cover. The NDSI value for a pixel can be calculated using the following formula:

[0116]

[0117] Wherein, Band3 is the surface reflectance of the pixel in band 3; Band6 is the surface reflectance of the pixel in band 6.

[0118] NDVI is the Normalized Difference Vegetation Index. The NDVI value corresponding to a pixel can be calculated using the following formula:

[0119]

[0120] Wherein, Band4 is the surface reflectance of the pixel in band 4; Band5 is the surface reflectance of the pixel in band 5.

[0121] In Formula 1, "and" means that a pixel must simultaneously meet the conditions corresponding to the above four parameters to be identified as a pixel belonging to a potential cloud area.

[0122] Building upon the above, the whiteness test and HOT test are further employed to separate pixels that are clearly not clouds. The whiteness test identifies cloud areas based on the white appearance of clouds in remote sensing images, while the HOT test separates smoke, thin clouds, and clear sky pixels. The HOT test can also identify some bright pixels, such as rocks, murky water, or ice and snow surfaces.

[0123] For example, in a whiteness test, if a pixel belonging to a cloud area has a whiteness of less than 0.7, the whiteness of the pixel can be obtained using the following formula:

[0124]

[0125] Where Bandj is the surface reflectance of the pixel in the band corresponding to band number j (i.e., band j); during the whiteness test, j can be an integer from 1 to 3.

[0126] MeanVis is an intermediate parameter, which can be calculated using the following formula:

[0127]

[0128] Here, Band2 represents the surface reflectance of the pixel in band 2, and the meanings of Band3 and Band4 are explained above.

[0129] In the HOT test, a pixel is considered to belong to a cloud area if its HOT value is greater than zero. The HOT value of a pixel can be calculated using the following formula:

[0130] HOT = (Band2 - 0.5) * (Band4 - 0.08) (Formula 6);

[0131] Through the above tests, pixels belonging to potential cloud areas can be identified. For pixels belonging to potential cloud areas, pixels that meet the conditions in Formula 7 below are further identified as clouds over water or land, so as to ultimately identify clouds over land and water:

[0132] (NDVI < 0.01 and Band5 < 0.11) or (NDVI < 0.1 and Band5 < 0.05) (Formula 7);

[0133] The relevant parameters in Formula 7 can be found in the previous introduction. In Formula 7, "and" means "and". It can be seen that if a cell belonging to a potential cloud area satisfies the condition before or after "or" in Formula 7, then the cell can be determined to be a cell covered by cloud.

[0134] It's understandable that the above explanation uses the detection of pixels belonging to clouds as an example, but the Fmask algorithm is also used for snow detection, and the specifics are similar, so I won't elaborate further. Of course, there are other ways to identify pixels belonging to clouds and snow cover, and there are no restrictions on this.

[0135] Furthermore, in order to improve the accuracy of cloud detection, this application can also use the Fmask algorithm to detect pixels belonging to clouds and snow, and then use the quality assessment (QA) band of the Landsat 8 Operational Land Imager (OLI) image tool to perform a second screening of the reference remote sensing image pixel by pixel. The specific implementation process is not limited.

[0136] After identifying the pixels belonging to clouds and snow in the reference remote sensing image, this application can remove the pixels belonging to clouds and snow in the reference remote sensing image by means of masking, such as using the Fmask algorithm to mask the reference remote sensing image, without any limitation.

[0137] S303, for each processed reference remote sensing image, determine at least one reference image block in the reference remote sensing image that belongs to at least one bright surface block.

[0138] Each reference image block is a remote sensing image block belonging to a bright surface in the baseline remote sensing image. Since each reference image block is a remote sensing image of a bright surface block, each reference image block corresponds to a bright surface block.

[0139] Correspondingly, the latitude and longitude information of the highlighted surface patch is the same as that of the reference image patch corresponding to the highlighted surface patch.

[0140] For example, a reference image block can be determined as a continuous image block composed of pixels whose brightness values ​​exceed a set brightness threshold, thus obtaining at least one reference image block, and each reference image area corresponds to a bright surface block, without any specific restrictions.

[0141] In one possible implementation, at least one candidate reference image block belonging to a bright surface can be identified in the baseline remote sensing image after removing clouds and snow. Specifically, the proportion of pixels in the candidate reference image block whose brightness values ​​exceed a set brightness threshold exceeds a set percentage. Correspondingly, from these candidate reference image blocks, at least one reference image block whose effective pixel proportion exceeds the set threshold is determined, with each reference image block corresponding to a bright surface block.

[0142] In this context, a valid pixel refers to a pixel that has a pixel value. It can be understood that if pixels belonging to clouds or snow are removed from the processed reference remote sensing image, then these pixels no longer have pixel values, resulting in gaps at those pixel locations in the reference remote sensing image.

[0143] Therefore, if there are many missing pixels in a reference image block, the surface reflectance of each pixel in the reference image block cannot accurately reflect the average surface reflectance of the high-brightness surface block corresponding to the reference image block. Therefore, it is necessary to filter out each reference image block whose proportion of valid pixels exceeds a set threshold.

[0144] S304, for each bright surface patch in the reference remote sensing image corresponding to a reference image patch, based on the surface reflectance of each pixel in the reference image patch corresponding to the bright surface patch, determine the average surface reflectance of the bright surface patch at the reference time point corresponding to the reference remote sensing image.

[0145] It is understandable that each reference remote sensing image corresponds to a reference time point when that reference remote sensing image was acquired. Therefore, the average surface reflectance of each pixel in the reference image block is the average surface reflectance of the bright surface block corresponding to that reference image block at that reference time point.

[0146] This application uses the average surface reflectance of a bright surface patch to replace the surface reflectance of a single image cloud. This can effectively measure the number of effective surface reflectances and reduce the inaccuracy of the data in the constructed surface reflectance library caused by the inaccuracy of the surface reflectance of a single pixel.

[0147] S305, For each bright surface patch, based on the average surface reflectance of the bright surface patch at multiple reference time points, fit the surface reflectance curve of the bright surface patch.

[0148] It is understandable that a bright surface patch at the same location in different reference remote sensing images represents the same surface patch. The only difference between the two images is the pixel information contained in them. Accordingly, the average surface reflectance of the bright surface patch in each reference remote sensing image is the average surface reflectance of the bright surface patch at each reference time point.

[0149] In this application, the specific implementation of fitting the surface reflectance curve based on the average surface reflectance of the bright surface patch at multiple reference time points can employ various fitting algorithms, and there are no restrictions on this.

[0150] For example, considering that the average surface reflectance of a bright surface patch at multiple reference time points is a time series, this application can use any one of the three time series models proposed by Zhu and Woodcock et al. for fitting. These three time series models can be divided into simple models, advanced models, and complete models. The simple model, advanced model, and complete model are represented by the following formulas 8, 9, and 10, respectively:

[0151]

[0152] in, This represents the function expression of the average surface reflectance curve fitted using a simple model. This represents the function expression of the average surface reflectance curve fitted using an advanced model; This represents the function expression of the average surface reflectance curve fitted using the complete model.

[0153] Understandably, remote sensing images can correspond to multiple bands, and the surface reflectance of remote sensing images will differ across different bands. Therefore, it is necessary to determine the surface reflectance corresponding to each remote sensing image for each band. Based on this, in formulas eight, nine, and ten above, i is the band number, and i is an integer from 1 to n. b This represents the total number of bands corresponding to the baseline remote sensing image.

[0154] Among them, a 0,i a 1,i b 1,i and c 1,i These represent different fitting parameters corresponding to band i in the simple simulation model;

[0155] χ represents the cumulative days of the year, which is the total number of days from the first reference time point to the last reference time point among multiple reference time points. T represents the number of annual cycles, which can take the value of 365.25.

[0156] a 2,i and b 2,i These are the two fitting parameters corresponding to band i in the advanced model.

[0157] a 3,i and b 3,i These are the two fitting parameters corresponding to band i in the complete model.

[0158] Taking a simple model as an example, the average surface reflectance of the bright surface patch at multiple reference time points is used as... Given multiple specific values, we can solve for a. 0,i a 1,i b 1,i and c 1,i The specific value can then be obtained. The corresponding function representation.

[0159] Of course, the above is just an example. This embodiment is also applicable to cases where other methods are used to fit the surface reflectance curve.

[0160] It is understandable that, in addition to the noise interference from clouds and snow on the baseline remote sensing imagery, other noise factors may also exist, leading to outliers in the fitted surface reflectance curve. Therefore, after fitting the surface reflectance curve for the bright surface patch, any outliers that may exist in the curve can be removed.

[0161] Specifically, for each bright surface area, the rate of change of surface reflectance at each unit time point in the surface reflectance curve of that bright surface area can be determined. Then, based on the rate of change at each unit time point in the surface reflectance curve, the maximum and minimum rates of change are determined, and based on these maximum and minimum rates of change, a maximum rate of change threshold is determined. Correspondingly, the surface reflectance corresponding to the target unit time point that meets the conditions is removed from the surface reflectance curve.

[0162] Among them, the rate of change of surface reflectance corresponding to the target unit time point that meets the condition exceeds the maximum rate of change threshold, and the rate of change of surface reflectance corresponding to the two adjacent unit time points adjacent to the target unit time point does not exceed the maximum rate of change threshold.

[0163] The size of the unit of time can be set according to actual needs. For example, it can be one hour, one day, or one month as a unit of time, without any restrictions.

[0164] Wherein, the rate of change of surface reflectance θ at unit time point t t It can be calculated using the following formula eleven:

[0165]

[0166] Where, N t N represents the surface reflectance at time t. t+1 The surface reflectance is the value at time t+1.

[0167] The maximum rate of change threshold can be determined in several ways. For example, in one possible implementation, the maximum rate of change threshold can be three times the difference between the maximum and minimum rate of change, i.e., the maximum rate of change threshold θ. ε It can be represented by the following formula twelve:

[0168]

[0169] Where, N Max For the maximum rate of change, N Min The minimum rate of change; t(N) Max ) represents the unit time point corresponding to the maximum rate of change, t(N) Min () represents the unit time point corresponding to the minimum rate of change.

[0170] Based on this, if the rate of change at a certain unit time point exceeds the maximum rate of change threshold, and the rate of change at other unit time points after that unit time point also exceeds the maximum rate of change threshold, it indicates that there is a new change in the curve as a whole after that unit time point; if only the rate of change at that unit time point exceeds the maximum rate of change threshold, it indicates that the surface reflectance at that unit time point is an anomaly, and therefore, the surface reflectance at that unit time point needs to be removed from the surface reflectance curve.

[0171] Removing the surface reflectance corresponding to a specific time point can be achieved by marking that the surface reflectance at that time point as an outlier, so that it is not subsequently used as the reference surface reflectance for atmospheric correction. Of course, other methods can also be used to remove outliers, and there are no restrictions on this.

[0172] The following describes one implementation example to illustrate the aerosol optical thickness inversion method provided in this application. Figure 3 This illustration shows another implementation flowchart of the aerosol optical thickness inversion method provided in this application. The method in this embodiment may include:

[0173] S401, Identify at least one bright image block in the target remote sensing image to be processed that belongs to a bright surface.

[0174] S402, for each bright image block, based on the latitude and longitude information corresponding to the bright image block, determine the target aerosol mode and target atmospheric mode corresponding to the bright image block.

[0175] For each high-brightness image block, steps S402 to S408 are performed respectively to determine the aerosol optical thickness corresponding to the high-brightness image block.

[0176] S403 uses an atmospheric radiative transfer model to simulate the surface reflectance of the bright image block under multiple different aerosol optical thicknesses, based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block.

[0177] It is understood that steps S402 and S403 are illustrated using one method of simulating simulated surface reflectance. However, if other methods are used to simulate the surface reflectance of bright image blocks under multiple different aerosol optical thicknesses, this embodiment is also applicable, and there is no limitation thereto.

[0178] S404. Based on the latitude and longitude information corresponding to the bright image block, determine the target surface reflectance curve of the target bright surface block corresponding to the bright image block from the surface reflectance database of bright surfaces.

[0179] The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time points is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time points.

[0180] S405, based on the acquisition time point corresponding to the high-brightness image block, the reference surface reflectance of the high-brightness image block at the acquisition time point is determined using the target surface reflectance curve.

[0181] Steps S404 and S405 above can be found in the relevant descriptions of the previous embodiments, and will not be repeated here.

[0182] S406. For each aerosol optical thickness, the spectral angle is calculated based on the simulated surface reflectance and reference surface reflectance of the high-brightness image block under the aerosol optical thickness, and the spectral angle corresponding to the aerosol optical thickness is obtained.

[0183] The spectral angle is the generalized angle between the bright image area and the bright surface area. A smaller spectral angle indicates that the simulated surface reflectance is closer to the reference surface reflectance.

[0184] This application does not impose any restrictions on the specific calculation process for calculating the spectral angle.

[0185] For example, in one possible implementation, the spectral angle α for each aerosol optical thickness can be calculated using the following formula:

[0186]

[0187] Where, r i s represents the simulated surface reflectance of the highlighted image block in the i-th band (i.e., the band with band number i); i The reference ground for the bright surface block corresponding to the bright image block in the i-th band (that is, the band with band number i). Surface reflection Rate. i and n b The meaning of is explained above and will not be repeated here.

[0188] S407, from the spectral angles corresponding to multiple different aerosol optical thicknesses, determine the smallest first spectral angle and the second and third spectral angles with the smallest angle difference from the first spectral angle.

[0189] For example, the second spectral angle is the spectral angle that is smaller than the first spectral angle and closest to the first spectral angle among the spectral angles corresponding to multiple different aerosol optical thicknesses; while the third spectral angle is the spectral angle that is larger than the first spectral angle and closest to the first spectral angle among the multiple spectral angles.

[0190] It is understandable that the simulated surface reflectances corresponding to these three spectral angles are also the three simulated surface reflectances that are closest to the reference surface reflectance.

[0191] Of course, this embodiment uses three spectral angles as an example. In practical applications, the smallest two or four spectral angles can also be selected, and there is no limitation on this.

[0192] S408, based on the aerosol optical thickness corresponding to the first spectral angle, the second spectral angle and the third spectral angle respectively, fits the target aerosol optical thickness corresponding to the high brightness image block.

[0193] There are multiple ways to fit the target aerosol optical thickness corresponding to the high-brightness image block by combining the aerosol optical thicknesses corresponding to these three spectral angles, and there are no restrictions on this.

[0194] In one possible implementation, the first thickness parameter a and the second thickness parameter b can be fitted using the following formula fourteen based on the first aerosol optical thickness AOT1 corresponding to the first spectral angle α1, the second aerosol optical thickness AOT2 corresponding to the second spectral angle α2, and the third aerosol optical thickness AOT3 corresponding to the third spectral angle α3:

[0195]

[0196] Based on the first thickness parameter a and the second thickness parameter b, the target aerosol optical thickness AOT4 corresponding to the high-brightness image block is determined by the following formula fifteen:

[0197]

[0198] Of course, the above is only one way to fit the target aerosol optical thickness corresponding to the bright image block, and other implementation methods are also applicable to this embodiment.

[0199] Corresponding to the aerosol optical thickness inversion method of this application, this application also provides an aerosol optical thickness inversion device.

[0200] like Figure 4 This diagram illustrates a structural composition of the aerosol optical thickness inversion device provided in this application. The device in this embodiment may include:

[0201] Brightness block determination unit 501 is used to determine at least one bright image block in the target remote sensing image to be processed that belongs to a bright surface.

[0202] The reflectance simulation unit 502 is used to simulate the surface reflectance of each bright image block under multiple different aerosol optical thicknesses based on the pixel value of each pixel in the bright image block using an atmospheric radiative transfer model.

[0203] The database query unit 503 is used to retrieve data from the surface of the highlighted land area based on the latitude and longitude information corresponding to the highlighted image block. In the reflectance database, determine the target surface reflectance curve of the target bright surface block corresponding to the bright image block. The purpose The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time point is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time point.

[0204] The reflectance query unit 504 is used to determine the reference surface reflectance of the high-brightness image block at the acquisition time point based on the acquisition time point corresponding to the high-brightness image block and using the target surface reflectance curve.

[0205] The reflectance matching unit 505 is used to determine at least one target simulated surface reflectance that matches the reference surface reflectance from the simulated surface reflectances corresponding to the plurality of different aerosol optical thicknesses.

[0206] The aerosol determination unit 506 is used to fit the target aerosol optical thickness corresponding to the high-brightness image block based on the aerosol optical thickness corresponding to the at least one target simulated surface reflectance.

[0207] In one possible implementation, the reflectivity matching unit includes:

[0208] The spectral angle calculation subunit is used to calculate the spectral angle for each aerosol optical thickness based on the simulated surface reflectance and the reference surface reflectance of the high-brightness image block under the aerosol optical thickness, and obtain the spectral angle corresponding to the aerosol optical thickness.

[0209] The spectral angle selection subunit is used to determine the smallest first spectral angle and the second and third spectral angles with the smallest angle difference from the first spectral angle from multiple spectral angles corresponding to different aerosol optical thicknesses.

[0210] In yet another possible implementation, the aerosol determining unit includes:

[0211] The parameter calculation subunit is used to fit the first thickness parameter a and the second thickness parameter b based on the first aerosol optical thickness AOT1 corresponding to the first spectral angle α1, the second aerosol optical thickness AOT2 corresponding to the second spectral angle α2, and the third aerosol optical thickness AOT3 corresponding to the third spectral angle α3, using the following formula:

[0212]

[0213] The thickness calculation subunit is used to determine the target aerosol optical thickness corresponding to the high-brightness image block based on the first thickness parameter a and the second thickness parameter b, using the following formula:

[0214]

[0215] Wherein, AOT4 is the target aerosol optical thickness corresponding to the high-brightness image block.

[0216] In yet another possible implementation, the reflectivity simulation unit includes:

[0217] The mode determination subunit is used to determine the target aerosol mode and target atmospheric mode corresponding to the high-brightness image block based on the latitude and longitude information corresponding to the high-brightness image block.

[0218] The reflectance simulation subunit is used to simulate the surface reflectance of the bright image block under multiple different aerosol optical thicknesses based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block, using an atmospheric radiative transfer model.

[0219] In another possible implementation, the surface reflectance library of the bright surface includes the surface reflectance curve of at least one bright surface block;

[0220] The device also includes a curve construction unit for determining the surface reflectance curves of each high-brightness surface patch in the surface reflectance library in the following manner:

[0221] An image time series is obtained, which includes reference remote sensing images acquired at multiple reference time points. The reference remote sensing images are remote sensing images after atmospheric correction, and each pixel in the reference remote sensing images has its own surface reflectance.

[0222] For each frame of the reference remote sensing image, remove the pixels that are covered by clouds or snow to obtain the processed reference remote sensing image.

[0223] For each processed baseline remote sensing image, at least one reference image block in the baseline remote sensing image is identified as belonging to at least one bright surface block, wherein each reference image block belongs to one bright surface block.

[0224] For each bright surface patch in the reference remote sensing image corresponding to a reference image block, the average surface reflectance of the bright surface patch at the reference time point corresponding to the reference remote sensing image is determined based on the surface reflectance of each pixel in the reference image block corresponding to the bright surface patch.

[0225] For each bright surface patch, a surface reflectance curve for the bright surface patch is fitted based on the average surface reflectance of the bright surface patch at the multiple reference time points.

[0226] In one possible implementation, the curve construction unit, when determining at least one reference image block in the baseline remote sensing image that belongs to at least one highlighted surface block, is specifically used for:

[0227] Identify at least one candidate reference image patch in the reference remote sensing image that belongs to a bright surface;

[0228] From the at least one candidate reference image block, at least one reference image block is determined where the proportion of effective pixels exceeds a set threshold. Each reference image block corresponds to a bright surface block, wherein the effective pixel is a pixel with a pixel value.

[0229] In yet another possible implementation, the device further includes:

[0230] The rate of change determination unit is used to determine the rate of change of surface reflectance at each unit time point in the surface reflectance curve of the high-brightness surface block after the curve construction unit fits the surface reflectance curve of the high-brightness surface block.

[0231] The rate of change screening unit is used to determine the maximum and minimum rate of change based on the rate of change corresponding to each unit time point in the surface reflectance curve;

[0232] A threshold determination unit is used to determine a maximum change rate threshold based on the maximum change rate and the minimum change rate;

[0233] An anomaly removal unit is used to remove the surface reflectance corresponding to the target unit time point that meets the conditions in the surface reflectance curve, wherein the rate of change of the surface reflectance corresponding to the target unit time point that meets the conditions exceeds the maximum rate of change threshold, and the rate of change of the surface reflectance corresponding to two adjacent unit time points adjacent to the target unit time point does not exceed the maximum rate of change threshold.

[0234] In yet another possible implementation, the device further includes:

[0235] The low-brightness block determination unit is used to determine at least one low-brightness image block in the target remote sensing image that belongs to a low-brightness surface.

[0236] The low-brightness block inversion unit is used to invert the aerosol thickness of each low-brightness image block using the dark pixel method.

[0237] An atmospheric correction unit is used to perform atmospheric correction on the target remote sensing image based on the target aerosol thickness of each high-brightness image block and the aerosol thickness of each low-brightness image block.

[0238] This application also provides an electronic device in its embodiments. For example... Figure 5 As shown, it illustrates a schematic diagram of the composition structure of the electronic device, which includes at least one processor 601 and a memory 602 connected to the processor 601.

[0239] Memory 602 is used to store computer programs;

[0240] Processor 601 is configured to execute the computer program to enable the electronic device to implement the aerosol optical thickness inversion method as described in any of the above embodiments.

[0241] It is understood that the electronic device may also include a display unit 603 and an input unit 604.

[0242] Of course, the electronic device can also have more than Figure 5 There are no restrictions on the number of components, whether more or fewer.

[0243] This application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the aerosol optical thickness inversion methods provided in this application.

[0244] This application also provides a computer-readable storage medium carrying one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the aerosol optical thickness inversion methods provided in this application.

[0245] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0246] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0247] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0248] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for inverting aerosol optical thickness, characterized in that, include: Identify at least one bright image block in the target remote sensing image that belongs to a bright surface; For each bright image block, based on the pixel value of each pixel in the bright image block, the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses is simulated using the atmospheric radiative transfer model. Based on the latitude and longitude information corresponding to the bright image block, the target surface reflectance curve of the target bright surface block corresponding to the bright image block is determined from the surface reflectance database of bright surfaces. The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time point is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time point. Based on the acquisition time point corresponding to the bright image block, the reference surface reflectance of the bright image block at the acquisition time point is determined using the target surface reflectance curve; For each aerosol optical thickness, the spectral angle is calculated based on the simulated surface reflectance and the reference surface reflectance of the high-brightness image block under the aerosol optical thickness, and the spectral angle corresponding to the aerosol optical thickness is obtained. From the spectral angles corresponding to multiple different aerosol optical thicknesses, determine the smallest first spectral angle, and the second and third spectral angles with the smallest angle difference from the first spectral angle; Based on the first spectral angle The corresponding first aerosol optical thickness Second spectral angle Corresponding second aerosol optical thickness and the third spectral angle Corresponding third aerosol optical thickness The first thickness parameter is fitted using the following formula. Second thickness parameter : ； Based on the first thickness parameter Second thickness parameter The optical thickness of the target aerosol corresponding to the high-brightness image block is determined by the following formula: ； in, The target aerosol optical thickness corresponding to the bright image block.

2. The aerosol optical thickness inversion method according to claim 1, characterized in that, The simulation of the surface reflectance of the high-brightness image block under multiple different aerosol optical thicknesses, based on the pixel value of each pixel in the high-brightness image block and using an atmospheric radiative transfer model, includes: Based on the latitude and longitude information corresponding to the bright image block, the target aerosol mode and target atmospheric mode corresponding to the bright image block are determined. Based on the target aerosol model, the target atmospheric model, and the pixel value of each pixel in the bright image block, the simulated surface reflectance of the bright image block under multiple different aerosol optical thicknesses is simulated using an atmospheric radiative transfer model.

3. The aerosol optical thickness inversion method according to claim 1, characterized in that, The surface reflectance library of the high-brightness surface includes the surface reflectance curve of at least one high-brightness surface block; The surface reflectance curve of the high-brightness surface patch was obtained in the following way: An image time series is obtained, which includes reference remote sensing images acquired at multiple reference time points. The reference remote sensing images are remote sensing images after atmospheric correction, and each pixel in the reference remote sensing images has its own surface reflectance. For each frame of the reference remote sensing image, remove the pixels that belong to clouds or snow from the reference remote sensing image to obtain the processed reference remote sensing image. For each processed baseline remote sensing image, at least one reference image block in the baseline remote sensing image is identified as belonging to at least one bright surface block, wherein each reference image block belongs to one bright surface block. For each bright surface patch in the reference remote sensing image corresponding to a reference image block, the average surface reflectance of the bright surface patch at the reference time point corresponding to the reference remote sensing image is determined based on the surface reflectance of each pixel in the reference image block corresponding to the bright surface patch. For each bright surface patch, a surface reflectance curve for the bright surface patch is fitted based on the average surface reflectance of the bright surface patch at the multiple reference time points.

4. The aerosol optical thickness inversion method according to claim 3, characterized in that, The process of determining at least one reference image block belonging to at least one bright surface block in the baseline remote sensing image includes: Identify at least one candidate reference image patch in the reference remote sensing image that belongs to a bright surface; From the at least one candidate reference image block, at least one reference image block is determined where the proportion of effective pixels exceeds a set threshold. Each reference image block corresponds to a bright surface block, wherein the effective pixel is a pixel with a pixel value.

5. The aerosol optical thickness inversion method according to claim 3, characterized in that, After fitting the surface reflectance curve of the high-brightness surface patch, the process also includes: For each bright surface block, determine the rate of change of surface reflectance at each unit time point in the surface reflectance curve of the bright surface block; Based on the rate of change at each unit time point in the surface reflectance curve, determine the maximum and minimum rates of change; Based on the maximum and minimum rates of change, a threshold for the maximum rate of change is determined; Remove the surface reflectance from the surface reflectance curve that corresponds to the target unit time point that meets the conditions, wherein the rate of change of the surface reflectance corresponding to the target unit time point that meets the conditions exceeds the maximum rate of change threshold, and the rate of change of the surface reflectance corresponding to two adjacent unit time points adjacent to the target unit time point does not exceed the maximum rate of change threshold.

6. The aerosol optical thickness inversion method according to claim 1, characterized in that, Also includes: Identify at least one low-brightness image block in the target remote sensing image that belongs to a low-brightness surface; The aerosol thickness of each low-brightness image block was determined using the dark pixel method. Atmospheric correction is performed on the target remote sensing image based on the target aerosol thickness in each high-brightness image block and the aerosol thickness in each low-brightness image block.

7. An aerosol optical thickness inversion device, characterized in that, include: The bright patch determination unit is used to determine at least one bright image patch in the target remote sensing image to be processed that belongs to a bright surface. The reflectance simulation unit is used to simulate the surface reflectance of each bright image block under multiple different aerosol optical thicknesses, based on the pixel value of each pixel in the bright image block and using an atmospheric radiative transfer model. The database query unit is used to determine the target surface reflectance curve of the target bright surface block corresponding to the bright image block from the surface reflectance database of bright surfaces based on the latitude and longitude information corresponding to the bright image block. The target surface reflectance curve is a curve fitted based on the average surface reflectance of the target bright surface block at multiple reference time points. The average surface reflectance of the target bright surface block at the reference time points is the average surface reflectance of each pixel belonging to the target bright surface block in the reference remote sensing image acquired at the reference time points. The reflectance query unit is used to determine the reference surface reflectance of the high-brightness image block at the acquisition time point based on the acquisition time point corresponding to the high-brightness image block and using the target surface reflectance curve. A reflectance matching unit is used to determine at least one target simulated surface reflectance that matches the reference surface reflectance from the plurality of simulated surface reflectances corresponding to different aerosol optical thicknesses. The aerosol determination unit is used to fit the target aerosol optical thickness corresponding to the high-brightness image block based on the aerosol optical thickness corresponding to the at least one target simulated surface reflectance. The reflectivity matching unit includes: The spectral angle calculation subunit is used to calculate the spectral angle for each aerosol optical thickness based on the simulated surface reflectance and the reference surface reflectance of the high-brightness image block under the aerosol optical thickness, and obtain the spectral angle corresponding to the aerosol optical thickness. The spectral angle selection subunit is used to determine the smallest first spectral angle and the second and third spectral angles with the smallest angle difference from the first spectral angle from multiple spectral angles corresponding to different aerosol optical thicknesses. The aerosol determination unit includes: Parameter calculation subunit, used to calculate based on the first spectral angle The corresponding first aerosol optical thickness Second spectral angle Corresponding second aerosol optical thickness and the third spectral angle Corresponding third aerosol optical thickness The first thickness parameter is fitted using the following formula. Second thickness parameter : ； Thickness calculation subunit, used to calculate based on the first thickness parameter Second thickness parameter The optical thickness of the target aerosol corresponding to the high-brightness image block is determined by the following formula: ； in, The target aerosol optical thickness corresponding to the bright image block.

8. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program to enable the electronic device to implement the aerosol optical thickness inversion method as described in any one of claims 1 to 6.

Images