Land surface temperature inversion method and land surface temperature inversion device

By combining the atmospheric correction regression network and atmospheric radiation transmission model, the spectral data in surface temperature inversion are processed, and the problems of surface temperature inversion accuracy and efficiency are solved, achieving higher computational accuracy and interpretability.

CN118761330BActive Publication Date: 2024-11-19AEROSPACE INFORMATION RES INST CAS +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411239332.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-11-19
Estimated Expiration
2044-09-05

AI Technical Summary

Technical Problem

The prior art has poor inversion accuracy and long calculation time during the inversion process of surface temperature inversion, especially the problem of few channels of multispectral surface temperature inversion methods and low inversion accuracy under specific conditions.

Method used

The combination of atmospheric correction regression network and atmospheric radiation transmission model is used to extract the radiation observations at the pupil from the target spectral image, and the atmospheric spectral parameters are obtained, and the spectral smoothing is generated through the ground-off spectral radiation estimate and the surface temperature candidate value. Loss value, determine the estimated surface temperature.

Benefits of technology

The calculation accuracy and calculation efficiency of surface temperature are improved, while the interpretability of inversion is enhanced, and the problem of insufficient accuracy and efficiency of traditional methods is overcome.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118761330B_ABST
    Figure CN118761330B_ABST
Patent Text Reader

Abstract

The present invention provides a surface temperature inversion method and a surface temperature inversion device. The method comprises extracting a radiation observation value at an entrance pupil from a target spectral image, processing the radiation observation value at the entrance pupil using an atmospheric correction regression network corresponding to a target load, obtaining a set of atmospheric spectral parameters of each pixel in a target area, processing the set of atmospheric spectral parameters corresponding to the pixel using an atmospheric radiation transmission model for any pixel in the target area, generating an off-ground spectral radiation estimation value of the pixel, generating a spectral smoothing loss value corresponding to each surface temperature candidate value based on the off-ground spectral radiation estimation value and a plurality of surface temperature candidate values ​​corresponding to the pixel, so as to determine the surface temperature estimation value of the pixel in the target area.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of remote sensing technology, and more specifically, to a surface temperature inversion method and a surface temperature inversion device. Background Art

[0002] Surface temperature is an important characteristic physical quantity of the land surface system. It can provide information on the spatiotemporal changes of the surface energy balance state, and comprehensively reflects the results of energy flow and material exchange in the soil-vegetation-atmosphere system. It plays a vital role in research in the fields of environmental monitoring, drought monitoring, frost damage monitoring, agricultural water resources management, forest fire prevention, etc., and is listed as one of the priority parameters for measurement by the International Geosphere-Biosphere Program.

[0003] In the process of realizing the concept of the present invention, it was found that in the process of surface temperature inversion in the related technology, the inversion accuracy was poor and the calculation time was long. Summary of the invention

[0004] In view of this, embodiments of the present invention provide a surface temperature inversion method and surface temperature inversion device, electronic equipment, computer-readable storage medium, and computer program product.

[0005] An aspect of an embodiment of the present invention provides a method for inverting land surface temperature, comprising:

[0006] In response to the temperature inversion instruction, a target spectrum image collected by the target payload is acquired, and a radiation observation value at the entrance pupil is extracted from the target spectrum image, wherein the target spectrum image is an image under the target environmental state obtained after the target payload passes through the target area;

[0007] Processing the radiation observation value at the entrance pupil by using the atmospheric correction regression network corresponding to the target load to obtain a set of atmospheric spectrum parameters of each pixel in the target area, wherein the set of atmospheric spectrum parameters includes a plurality of parameters characterizing the radiation and transmittance of the target area;

[0008] For any pixel in the target area, use the atmospheric radiation transfer model to process the atmospheric spectral parameter set corresponding to the pixel to generate an off-ground spectral radiation estimation value of the pixel;

[0009] Generate a spectral smoothing loss value corresponding to each of the above-mentioned surface temperature candidate values ​​according to the above-mentioned off-ground spectral radiation estimation value and a plurality of surface temperature candidate values ​​corresponding to the above-mentioned pixel;

[0010] The surface temperature candidate value corresponding to the minimum value among the plurality of spectral smoothing loss values ​​is determined as the surface temperature estimation value of the pixel in the target area.

[0011] According to an embodiment of the present invention, the above-mentioned atmospheric correction regression network is trained in the following manner:

[0012] Acquire a surface temperature training set, wherein the surface temperature training set includes a plurality of training samples corresponding to the target load, and the training samples include an entrance pupil radiation training value corresponding to at least one training area type and an atmospheric radiation label parameter corresponding to each entrance pupil radiation training value;

[0013] For each of the above training samples, the above pupil radiation training value is input into the deep residual network to obtain the atmospheric radiation prediction parameters;

[0014] Input the atmospheric radiation prediction parameter and the atmospheric radiation label parameter corresponding to the atmospheric radiation prediction parameter into the target loss function, and output the target loss value;

[0015] The deep residual network is iteratively adjusted according to the target loss value to obtain a trained atmospheric correction regression network.

[0016] According to an embodiment of the present invention, the above training samples are generated in the following manner:

[0017] For each of the above training area types, obtain the atmospheric profile data corresponding to the above training area type and the payload observation geometry of the above target payload;

[0018] Based on the spectral response function, the above-mentioned entrance pupil radiation training value and the above-mentioned atmospheric radiation label parameters are generated according to the atmospheric profile data and the above-mentioned payload observation geometry.

[0019] According to an embodiment of the present invention, before generating the above-mentioned spectrum smoothing loss value, the method further includes:

[0020] For each of the above-mentioned pixel, the above-mentioned off-ground spectral radiation estimation value of the above-mentioned pixel is denoised using an outlier detection method, so that the above-mentioned off-ground spectral radiation estimation value becomes smooth in the target spectral range, and a filtered off-ground spectral radiation estimation value is obtained, so as to calculate the above-mentioned spectral smoothing loss value using the above-mentioned filtered off-ground spectral radiation estimation value.

[0021] According to an embodiment of the present invention, generating a spectral smoothing loss value corresponding to each of the above-mentioned surface temperature candidate values ​​according to the above-mentioned off-ground spectral radiation estimation value and a plurality of surface temperature candidate values ​​corresponding to the above-mentioned pixel includes:

[0022] For each of the above-mentioned pixels, generating a plurality of the above-mentioned surface temperature candidate values ​​corresponding to the above-mentioned pixel according to the above-mentioned off-ground spectral radiation estimation value;

[0023] For each of the above-mentioned surface temperature candidate values, the above-mentioned spectral smoothing loss value corresponding to the above-mentioned surface temperature candidate value is obtained according to the above-mentioned surface temperature candidate value, the above-mentioned off-ground spectral radiation estimation value and the above-mentioned atmospheric spectral parameter set.

[0024] According to an embodiment of the present invention, generating a plurality of candidate ground surface temperature values ​​corresponding to the pixel based on the above-mentioned off-ground spectral radiation estimation value includes:

[0025] Determine an initial estimate of the surface temperature based on the above-mentioned estimated spectral radiation from the ground;

[0026] Based on the candidate temperature formula, according to the above-mentioned initial estimated value of the surface temperature, a candidate number of candidate surface temperature values ​​are generated, wherein any of the above-mentioned candidate surface temperature values ​​is located in a candidate interval, and the above-mentioned candidate interval is determined according to the above-mentioned initial estimated value of the surface temperature and the length of the candidate interval.

[0027] According to an embodiment of the present invention, obtaining the spectral smoothing loss value corresponding to the above-mentioned surface temperature candidate value according to the above-mentioned surface temperature candidate value, the above-mentioned off-ground spectral radiation estimation value and the above-mentioned atmospheric spectral parameter set includes:

[0028] Generate a spectral emissivity estimation value of any spectral band under a target spectrum according to the surface temperature candidate value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set, wherein the target spectrum includes a plurality of the spectral bands;

[0029] For any of the above spectral bands, a smoothed estimate of the off-ground spectral radiation is generated according to the spectral emissivity estimate of the above spectral band, the spectral emissivity estimate of the associated band and the above atmospheric spectral parameter set;

[0030] The spectral smoothing loss value corresponding to the ground surface temperature candidate value is generated based on the above-mentioned above-ground spectral radiation smoothing estimation value and the above-ground spectral radiation estimation value.

[0031] According to an embodiment of the present invention, the above-mentioned atmospheric spectral parameter set includes atmospheric downward spectral radiation prediction values.

[0032] According to an embodiment of the present invention, generating an off-ground spectral radiation smoothing estimate value according to the spectral emissivity estimate value of the spectral band, the spectral emissivity estimate value of the associated band and the atmospheric spectral parameter set includes:

[0033] Based on the spectral smoothing principle, a spectral emissivity smoothing estimate is generated according to the spectral emissivity estimate of the above spectral band and the spectral emissivity estimate of the associated band;

[0034] Based on the thermal infrared transmission equation, the above-mentioned smoothed estimate of the off-ground spectral radiation is generated according to the above-mentioned smoothed estimate of the spectral emissivity, the above-mentioned candidate value of the surface temperature and the above-mentioned predicted value of the atmospheric downlink spectral radiation.

[0035] According to an embodiment of the present invention, the above-mentioned atmospheric spectral parameter set includes an atmospheric upward spectral radiation prediction value and an atmospheric spectral transmittance prediction value, and the above-mentioned atmospheric radiation transmission model includes a thermal infrared transmission model.

[0036] According to an embodiment of the present invention, the atmospheric spectral parameter set corresponding to the pixel is processed using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel, including:

[0037] The thermal infrared transmission model is used to process the radiation observation value at the entrance pupil, the atmospheric upward spectral radiation prediction value and the atmospheric spectral transmittance prediction value to generate the above-ground spectral radiation estimation value.

[0038] Another aspect of an embodiment of the present invention provides a surface temperature inversion device, comprising:

[0039] An acquisition module is used to acquire a target spectrum image collected by a target payload in response to a temperature inversion instruction, and extract a radiation observation value at an entrance pupil from the target spectrum image, wherein the target spectrum image is an image under a target environmental state obtained after the target payload passes through a target area;

[0040] A first processing module is used to process the radiation observation value at the entrance pupil by using the atmospheric correction regression network corresponding to the target load to obtain a set of atmospheric spectrum parameters of each pixel in the target area, wherein the set of atmospheric spectrum parameters includes a plurality of parameters characterizing the radiation and transmittance of the target area;

[0041] A second processing module is used to process the atmospheric spectral parameter set corresponding to any pixel in the target area using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel;

[0042] A generating module, configured to generate a spectral smoothing loss value corresponding to each of the above-mentioned surface temperature candidate values ​​according to the above-mentioned off-ground spectral radiation estimation value and a plurality of surface temperature candidate values ​​corresponding to the above-mentioned pixel;

[0043] A determination module is used to determine the surface temperature candidate value corresponding to the minimum value among the plurality of spectral smoothing loss values ​​as the surface temperature estimation value of the pixel in the target area.

[0044] Another aspect of an embodiment of the present invention provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described above.

[0045] Another aspect of an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the method described above when executed.

[0046] Another aspect of an embodiment of the present invention provides a computer program product, wherein the computer program product includes computer executable instructions, and the instructions are used to implement the method described above when executed.

[0047] According to an embodiment of the present invention, by extracting the radiation observation value at the entrance pupil from the target spectral image, the radiation observation value at the entrance pupil is processed by the atmospheric correction regression network corresponding to the target load, and the atmospheric spectral parameter set of each pixel in the target area is obtained. For any pixel in the target area, the atmospheric spectral parameter set corresponding to the pixel is processed by the atmospheric radiation transfer model to generate the off-ground spectral radiation estimation value of the pixel. According to the off-ground spectral radiation estimation value and multiple surface temperature candidate values ​​corresponding to the pixel, a spectral smoothing loss value corresponding to each surface temperature candidate value is generated to determine the surface temperature estimation value of the pixel in the target area. Since the present application uses a combination of the atmospheric correction regression network and the atmospheric radiation transfer model to calculate the surface temperature estimation value, it overcomes the problems of the traditional multi-spectral surface temperature inversion method with few channels and low inversion accuracy under specific conditions, and can improve the calculation accuracy and efficiency of the surface temperature. At the same time, the inversion of the surface temperature has high interpretability. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0049] Figure 1 An exemplary system architecture to which a land surface temperature inversion method according to an embodiment of the present invention can be applied is shown;

[0050] Figure 2 A flow chart of a method for inverting land surface temperature according to an embodiment of the present invention is shown;

[0051] Figure 3 A flow chart of a method for surface temperature inversion according to another embodiment of the present invention is shown;

[0052] Figure 4 A block diagram of a ground surface temperature inversion device according to an embodiment of the present invention is shown;

[0053] Figure 5 A block diagram of a training device for an atmospheric correction regression network according to an embodiment of the present invention is shown;

[0054] Figure 6 A block diagram of an electronic device suitable for implementing the method described above according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0055] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0056] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0057] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0058] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0059] Thermal infrared remote sensing is an effective means to obtain global or regional target surface temperature. After several years of vigorous development, the technology of obtaining surface temperature products using multispectral thermal infrared remote sensing has achieved fruitful results, and surface temperature inversion algorithms such as single-channel algorithm, multi-channel algorithm, multi-angle method, temperature and emissivity separation method have been developed. However, due to the limited amount of information in traditional multispectral data, these inversion algorithms rely on the approximation of the radiation transfer equation and different assumptions and restrictions to solve the pathological problems of the inversion itself.

[0060] Hyperspectral thermal infrared sensors have the characteristics of narrow channel width and high spectral resolution. Thousands of spectral channel data contain rich and delicate surface / atmosphere spectral information, which provides additional physical constraints for effectively realizing the separation of surface and atmospheric information and the decoupling of surface temperature and emissivity, breaking through the limitation of insufficient observation information of remote sensing inversion of temperature and emissivity of multispectral thermal infrared data. With the advent of hyperspectral infrared sensors, such as IASI, CrIS, and HIRAS, domestic and foreign researchers have conducted exploratory research on remote sensing inversion of surface temperature from hyperspectral data, using the delicate spectral resolution of hyperspectral infrared data to solve the equation ill-conditioned problem in the process of separating temperature and emissivity by refining constraints with physical significance. Existing hyperspectral thermal infrared surface temperature inversion is divided into two categories: physical methods and artificial intelligence methods. The physical method is based on the thermal infrared radiation transfer equation, with clear physical mechanism and strong interpretability, but there are problems such as underdetermined equation solution, strict assumptions and heavy computational burden; the artificial intelligence method has strong learning ability for nonlinear relationships and high computational efficiency, but there are problems such as insufficient generalization and migration ability and insufficient interpretability.

[0061] In view of this, an embodiment of the present invention provides a surface temperature inversion method and a surface temperature inversion device. The method includes extracting a radiation observation value at the entrance pupil from a target spectral image, processing the radiation observation value at the entrance pupil using an atmospheric correction regression network corresponding to the target load, obtaining a set of atmospheric spectral parameters of each pixel in the target area, and for any pixel in the target area, processing the set of atmospheric spectral parameters corresponding to the pixel using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel, and generating a spectral smoothing loss value corresponding to each surface temperature candidate value based on the off-ground spectral radiation estimation value and multiple surface temperature candidate values ​​corresponding to the pixel, so as to determine the surface temperature estimation value of the pixel in the target area.

[0062] Figure 1 An exemplary system architecture 100 to which the surface temperature inversion method can be applied according to an embodiment of the present invention is shown. It should be noted that: Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present invention may be applied, to help those skilled in the art understand the technical content of the present invention, but do not mean that the embodiments of the present invention may not be used in other devices, systems, environments or scenarios.

[0063] like Figure 1As shown, the system architecture 100 of this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, a server 105, and a thermal infrared load 106. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, the server 105, and the thermal infrared load 106. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0064] The user may use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only for example).

[0065] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0066] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0067] The thermal infrared payload 106 may be a satellite-borne hyperspectral thermal infrared payload, which may collect hyperspectral images.

[0068] It should be noted that the land surface temperature inversion method provided in the embodiment of the present invention can be executed by one of the server 105, the first terminal device 101, the second terminal device 102, the third terminal device 103, and other servers or server clusters that can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Correspondingly, the land surface temperature inversion device provided in the embodiment of the present invention can generally be set in one of the server 105, the first terminal device 101, the second terminal device 102, the third terminal device 103, and other servers or server clusters that can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0069] It should be understood that Figure 1 The number of terminal devices, networks, servers and thermal infrared loads in the embodiment is only . According to the implementation requirements, there can be any number of terminal devices, networks, servers and thermal infrared loads.

[0070] Figure 2 A flow chart of a method for land surface temperature inversion according to an embodiment of the present invention is shown.

[0071] like Figure 2 As shown, the surface temperature inversion method includes operations S201 to S205.

[0072] In operation S201, in response to a temperature inversion instruction, a target spectrum image collected by a target payload is acquired, and a radiation observation value at an entrance pupil is extracted from the target spectrum image, wherein the target spectrum image is an image under a target environmental state obtained after the target payload passes through a target area;

[0073] In operation S202, the radiation observation value at the entrance pupil is processed using an atmospheric correction regression network corresponding to the target load to obtain an atmospheric spectral parameter set of each pixel in the target area, wherein the atmospheric spectral parameter set includes multiple parameters characterizing the radiation and transmittance of the target area;

[0074] In operation S203, for any pixel in the target area, an atmospheric spectral parameter set corresponding to the pixel is processed using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel;

[0075] In operation S204, a spectral smoothing loss value corresponding to each ground surface temperature candidate value is generated according to the off-ground spectral radiation estimation value and a plurality of ground surface temperature candidate values ​​corresponding to the pixel;

[0076] In operation S205, a candidate ground surface temperature value corresponding to a minimum value among the plurality of spectral smoothing loss values ​​is determined as a ground surface temperature estimation value of a pixel in the target area.

[0077] According to an embodiment of the present invention, the temperature inversion instruction can be input by a staff member through a corresponding operation on an electronic device such as a computer, and the electronic device automatically generates the instruction in response to the operation.

[0078] According to an embodiment of the present invention, the target payload may be any type of hyperspectral thermal infrared payload. The target environment state may refer to a cloudless state or other types of weather states. The target area may select a typical target area covering an underlying surface such as soil, vegetation, or water. The target spectral image may be a hyperspectral image. The atmospheric spectral parameter set may include an atmospheric uplink spectral radiation prediction value, an atmospheric downlink spectral radiation prediction value, and an atmospheric spectral transmittance prediction value.

[0079] According to an embodiment of the present invention, the cloudless hyperspectral images of the target area where the target payload passes through are downloaded and screened, and the cloud pixels are removed to obtain the cloudless data of the target area, and the radiation observation value at the entrance pupil of the target area is extracted. The radiation observation value at the entrance pupil of the target area is input into the constructed atmospheric correction regression network to obtain the predicted value of the atmospheric parameters of the target area, that is, the set of atmospheric spectral parameters of each pixel in the target area. At this time, the radiation observation value at the entrance pupil of the target area can be standardized and then input into the atmospheric correction regression network for processing.

[0080] According to an embodiment of the present invention, for any pixel in the target area, an atmospheric spectral parameter set corresponding to the pixel is processed using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel. A spectral smoothing loss value corresponding to each of the surface temperature candidate values ​​is generated based on the off-ground spectral radiation estimation value and multiple surface temperature candidate values ​​corresponding to the pixel. The surface temperature candidate value corresponding to the minimum value among the multiple spectral smoothing loss values ​​is determined as the surface temperature estimation value of the pixel in the target area.

[0081] According to an embodiment of the present invention, by extracting the radiation observation value at the entrance pupil from the target spectral image, the radiation observation value at the entrance pupil is processed by the atmospheric correction regression network corresponding to the target load, and the atmospheric spectral parameter set of each pixel in the target area is obtained. For any pixel in the target area, the atmospheric spectral parameter set corresponding to the pixel is processed by the atmospheric radiation transfer model to generate the off-ground spectral radiation estimation value of the pixel. According to the off-ground spectral radiation estimation value and multiple surface temperature candidate values ​​corresponding to the pixel, a spectral smoothing loss value corresponding to each surface temperature candidate value is generated to determine the surface temperature estimation value of the pixel in the target area. Since the present application uses a combination of the atmospheric correction regression network and the atmospheric radiation transfer model to calculate the surface temperature estimation value, it overcomes the problems of the traditional multi-spectral surface temperature inversion method with few channels and low inversion accuracy under specific conditions, and can improve the calculation accuracy and efficiency of the surface temperature. At the same time, the inversion of the surface temperature has high interpretability.

[0082] Figure 3 A flow chart of a method for ground surface temperature inversion according to another embodiment of the present invention is shown.

[0083] According to an embodiment of the present invention, Figure 3 As shown in Figure 1, the atmospheric correction regression network is trained as follows:

[0084] Acquire a land surface temperature training set, wherein the land surface temperature training set includes a plurality of training samples corresponding to the target load, and the training samples include an entrance pupil radiation training value corresponding to at least one training area type and an atmospheric radiation label parameter corresponding to each entrance pupil radiation training value;

[0085] For each training sample, the pupil radiation training value is input into the deep residual network to obtain the atmospheric radiation prediction parameters;

[0086] Inputting the atmospheric radiation prediction parameters and the atmospheric radiation label parameters corresponding to the atmospheric radiation prediction parameters into the target loss function, and outputting the target loss value;

[0087] The deep residual network is iteratively adjusted according to the target loss value to obtain a trained atmospheric correction regression network.

[0088] According to an embodiment of the present invention, a satellite-borne hyperspectral thermal infrared payload that meets the spectral resolution requirements, has high calibration accuracy and stable performance is selected as the target payload. The atmospheric radiation prediction parameters may include atmospheric uplink radiation data, atmospheric downlink radiation data and atmospheric transmittance data.

[0089] According to an embodiment of the present invention, a deep residual network is composed of a plurality of residual blocks, each of which is composed of a plurality of fully connected layers, batch normalization layers, and activation layers, and a reasonable network depth and width are set, wherein the depth refers to the number of fully connected layers, and the width refers to the number of neurons.

[0090] According to an embodiment of the present invention, the entrance pupil radiation training value in the training sample is input into the deep residual network to obtain the atmospheric radiation prediction parameter, the atmospheric radiation prediction parameter and the atmospheric radiation label parameter corresponding to the atmospheric radiation prediction parameter are input into the target loss function, and the target loss value is output, wherein the target loss function may be mean square error (MSE), mean absolute error (MAE), SmoothL1_loss loss function, etc.

[0091] According to an embodiment of the present invention, the deep residual network is iteratively adjusted according to the target loss value to obtain a trained atmospheric correction regression network.

[0092] According to an embodiment of the present invention, during iterative training, if the number of iterations does not reach the iteration number threshold T, the deep residual network corresponding to the minimum value of the adjacent preset number of target loss value fluctuations is determined as the atmospheric correction regression network.

[0093] According to an embodiment of the present invention, a test set may be set, and the test set is the same as the surface temperature training set, which will not be described in detail herein. The test set is used to evaluate the performance of the atmospheric correction regression network.

[0094] According to an embodiment of the present invention, the training samples are generated in the following manner:

[0095] For each training area type, obtain the atmospheric profile data and the payload observation geometry of the target payload corresponding to the training area type;

[0096] Based on the spectral response function, the entrance pupil radiation training value and atmospheric radiation label parameters are generated according to the atmospheric profile data and payload observation geometry.

[0097] According to an embodiment of the present invention, a suitable atmospheric profile database is selected, such as ECMWF and TIGR, and relative humidity is used as a judgment index. When there is a layer of atmospheric relative humidity greater than 90% or two consecutive layers of atmospheric relative humidity greater than 85% in the profile, it is judged to be a cloud profile, that is, the atmospheric profile data corresponding to the training area type. Based on this, the sunny and cloudless atmospheric profile is screened out, and the entrance pupil radiation training value and atmospheric radiation label parameter (including atmospheric upward radiation, atmospheric downward radiation and atmospheric transmittance) data set of the target payload under different underlying surfaces and different surface temperature conditions are simulated by combining the payload observation geometry, spectral response function, emissivity spectrum library, etc.

[0098] According to an embodiment of the present invention, the atmospheric spectral parameter set includes an atmospheric upward spectral radiation prediction value and an atmospheric spectral transmittance prediction value, and the atmospheric radiation transmission model includes a thermal infrared transmission model.

[0099] According to an embodiment of the present invention, an atmospheric spectral parameter set corresponding to a pixel is processed using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel, including:

[0100] The thermal infrared transmission model is used to process the radiation observations at the entrance pupil, the predicted values ​​of the atmospheric upwelling spectral radiation and the predicted values ​​of the atmospheric spectral transmittance to generate the estimated value of the off-ground spectral radiation.

[0101] According to an embodiment of the present invention, 800-980cm is selected -1 The thermal infrared atmospheric window is used as the target spectral interval for surface temperature inversion. According to the thermal infrared radiation transmission theory, at wavelength The thermal infrared radiation reaching the sensor entrance pupil can be expressed as follows:

[0102] (1)

[0103] (2)

[0104] in and The wavelengths are The radiation observation value at the sensor entrance pupil and the radiation off the ground, is the atmospheric spectral transmittance, is the surface spectral emissivity, is the Planck function, is the true value of the surface temperature, and They are respectively the upward and downward spectral radiation of the atmosphere.

[0105] According to an embodiment of the present invention, for each pixel in the target area, the spectral band The corresponding off-ground spectral radiation can be expressed as formula (3):

[0106] (3)

[0107] According to an embodiment of the present invention, the radiation observation value at the entrance pupil, the predicted value of the atmospheric upward spectral radiation, and the predicted value of the atmospheric spectral transmittance are substituted into the above formula, where the radiation observation value at the entrance pupil is used as , the predicted values ​​of atmospheric upward spectral radiation and atmospheric spectral transmittance are respectively and , calculate the estimated value of the spectral radiation off the ground .

[0108] According to an embodiment of the present invention, before generating the spectrum smoothing loss value, the method further includes:

[0109] For each pixel, the outlier detection method is used to denoise the off-ground spectral radiation estimation value of the pixel, so that the off-ground spectral radiation estimation value becomes smooth in the target spectral range, and the filtered off-ground spectral radiation estimation value is obtained, and the spectral smoothing loss value is calculated using the filtered off-ground spectral radiation estimation value.

[0110] According to an embodiment of the present invention, the outlier detection method may be a Z-score test, a Tukey test, a Cook's distance test, a Mahalanobis distance test, a Hampel test, a LOF (local outlier factor) test, or the like.

[0111] In a specific embodiment, an outlier detection method is used to remove noise from the calculated off-ground spectral radiation estimation value of each pixel in the target area, such as Hampel filtering. The Hampel filter is a filter based on the median and the median absolute deviation, which can identify and remove outliers in the sequence data. Compared with the traditional mean and standard deviation method, the Hampel filter is more robust to outliers. The Hampel filter is used to smooth the off-ground spectral radiation estimation value in the target spectral range to obtain the off-ground spectral radiation filtered estimation value. , which is the estimated value of the off-ground spectral radiation after filtering.

[0112] According to an embodiment of the present invention, a spectral smoothing loss value corresponding to each ground surface temperature candidate value is generated according to an off-ground spectral radiation estimation value and a plurality of ground surface temperature candidate values ​​corresponding to a pixel, including:

[0113] For each pixel, a plurality of candidate ground surface temperature values ​​corresponding to the pixel are generated according to the estimated value of the off-ground spectral radiation;

[0114] For each candidate value of the land surface temperature, a spectral smoothing loss value corresponding to the candidate value of the land surface temperature is obtained according to the candidate value of the land surface temperature, an estimated value of the off-ground spectral radiation and a set of atmospheric spectral parameters.

[0115] According to an embodiment of the present invention, a plurality of ground surface temperature candidate values ​​corresponding to a pixel are generated according to an estimated value of off-ground spectral radiation, including:

[0116] Determine an initial estimate of the surface temperature based on the estimated spectral radiation off the ground;

[0117] Based on the candidate temperature formula, according to the initial estimated value of the surface temperature, a candidate number of surface temperature candidate values ​​are generated, wherein any surface temperature candidate value is located in a candidate interval, and the candidate interval is determined according to the initial estimated value of the surface temperature and the length of the candidate interval.

[0118] According to an embodiment of the present invention, for each pixel in the target area, the off-ground spectral radiation estimation value or off-ground spectral radiation filtering estimation value of all bands in the target spectral interval is calculated. The corresponding brightness temperature mean is used as the initial estimate of the surface temperature. .

[0119] According to an embodiment of the present invention, based on the candidate temperature formula shown in formula (4), according to the initial estimated value of the surface temperature , generate candidate number of surface temperature candidate values :

[0120] (4)

[0121] Among them, t is the length of the candidate interval and n is the number of candidates.

[0122] According to an embodiment of the present invention, the candidate interval is [T0-t / 2, T0+t / 2], and the initial estimated value of the surface temperature is and the candidate interval length t can be set according to actual needs, such as the initial estimate of the surface temperature is 290K, and the candidate interval length t is 40K.

[0123] According to an embodiment of the present invention, a spectrum smoothing loss value corresponding to the candidate value of the ground temperature is obtained according to the candidate value of the ground temperature, the estimated value of the off-ground spectral radiation and the set of atmospheric spectral parameters, including:

[0124] Generate a spectral emissivity estimation value of any spectral band under a target spectrum according to a surface temperature candidate value, an off-ground spectral radiation estimation value and an atmospheric spectral parameter set, wherein the target spectrum includes multiple spectral bands;

[0125] For any spectral band, a smoothed estimate of the off-ground spectral radiation is generated according to the spectral emissivity estimate of the spectral band, the spectral emissivity estimate of the associated band and a set of atmospheric spectral parameters;

[0126] A spectral smoothing loss value corresponding to the surface temperature candidate value is generated based on the off-ground spectral radiation smoothing estimate value and the off-ground spectral radiation estimate value.

[0127] According to the embodiment of the present invention, according to formula (5), in the spectral band , spectral emissivity estimate It can be expressed as the surface temperature Function:

[0128] (5)

[0129] According to an embodiment of the present invention, the surface temperature candidate value, the off-ground spectral radiation filtering estimation value of each spectral band in the target spectral interval, and the atmospheric downlink spectral radiation prediction value are substituted into the above formula, where the surface temperature candidate value is As , spectral band Filtered estimate of the off-ground spectral radiation As , or the estimated spectral radiation above ground as , spectral band The predicted value of atmospheric downward spectral radiation is taken as , calculate the estimated spectral emissivity of the band .

[0130] According to an embodiment of the present invention, for any spectral band, a smoothed estimate of the off-ground spectral radiation is generated according to the spectral emissivity estimate of the spectral band, the spectral emissivity estimate of the associated band and a set of atmospheric spectral parameters.

[0131] According to an embodiment of the present invention, for each pixel in the target area, when the surface temperature candidate value is The spectrum smoothing loss value at can be calculated by formula (6):

[0132] (6)

[0133] Where N is the number of bands.

[0134] According to an embodiment of the present invention, the spectral smoothing loss value can be obtained by substituting the off-ground spectral radiation smoothing estimation value and the off-ground spectral radiation filtering estimation value (or the off-ground spectral radiation estimation value) into the above formula.

[0135] According to an embodiment of the present invention, for each pixel in the target area, all the candidate ground temperature values ​​under the pixel are calculated by spectral smoothing loss, and n spectral smoothing loss values ​​are obtained. The candidate ground temperature value taken when the spectral smoothing loss value is the minimum is used as the estimated ground temperature value of the pixel. ,Right now .

[0136] According to an embodiment of the present invention, the atmospheric spectral parameter set includes an atmospheric downwelling spectral radiation prediction value.

[0137] According to an embodiment of the present invention, generating an off-ground spectral radiation smoothing estimate value according to the spectral emissivity estimate value of the spectral band, the spectral emissivity estimate value of the associated band and the atmospheric spectral parameter set includes:

[0138] Based on the spectrum smoothing principle, a spectrum emissivity smoothing estimation value is generated according to the spectrum emissivity estimation value of the spectrum band and the spectrum emissivity estimation value of the associated band;

[0139] Based on the thermal infrared transmission equation, a smoothed estimate of the off-ground spectral radiation is generated according to the smoothed estimate of the spectral emissivity, the candidate value of the surface temperature and the predicted value of the atmospheric downlink spectral radiation.

[0140] According to an embodiment of the present invention, for each spectral band in the target spectral range, based on the spectral smoothing principle, the average of the spectral emissivity estimation values ​​of the spectral band and two adjacent spectral bands (i.e., associated bands) is used as the spectral emissivity smoothing estimation value. , to ensure that the emissivity range is within [0,1], if If it is greater than 1, update the value to 1; If it is less than 0, update the value to 0. Then Substituting the candidate value of surface temperature and the predicted value of atmospheric downward spectral radiation into formula (2), the smoothed estimated value of the above-ground spectral radiation of this band is calculated. .

[0141] According to an embodiment of the present invention, the surface temperature inversion method further includes: based on the measured surface temperature data of appropriate stations in the target area or reliable high-precision surface temperature products, extracting the ideal true value of the surface temperature that matches the estimated surface temperature value obtained by the inversion in time and space, and using the mean error (MAE), standard deviation (STD), root mean square error (RMSE) and correlation coefficient (R 2 ) and other indicators are used to verify and evaluate the accuracy of the surface temperature inversion method proposed in this invention.

[0142] According to the embodiments of the present invention, the surface temperature inversion method of the present invention overcomes the problems of the traditional multi-spectral surface temperature inversion method, such as the small number of channels and low inversion accuracy under specific conditions, and makes full use of the spectral information of the hyperspectral itself to promote the application of hyperspectral thermal infrared satellite data. At the same time, the present invention combines the advantages of strong interpretability of physical methods and strong learning ability and high computing efficiency of artificial intelligence methods, and can achieve a robust model, high-precision and interpretable surface temperature inversion.

[0143] Figure 4 A block diagram of a ground surface temperature inversion device according to an embodiment of the present invention is shown.

[0144] like Figure 4 As shown, the surface temperature inversion device 400 includes an acquisition module 410 , a first processing module 420 , a second processing module 430 , a generation module 440 , and a determination module 450 .

[0145] The acquisition module 410 is used to acquire the target spectrum image collected by the target payload in response to the temperature inversion instruction, and extract the radiation observation value at the entrance pupil from the target spectrum image, wherein the target spectrum image is an image under the target environmental state obtained after the target payload passes through the target area;

[0146] The first processing module 420 is used to process the radiation observation value at the entrance pupil by using the atmospheric correction regression network corresponding to the target load to obtain the atmospheric spectrum parameter set of each pixel in the target area, wherein the atmospheric spectrum parameter set includes multiple parameters characterizing the radiation and transmittance of the target area;

[0147] The second processing module 430 is used to process the atmospheric spectral parameter set corresponding to any pixel in the target area using the atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel;

[0148] A generating module 440, configured to generate a spectral smoothing loss value corresponding to each ground surface temperature candidate value according to the off-ground spectral radiation estimation value and a plurality of ground surface temperature candidate values ​​corresponding to the pixel;

[0149] The determination module 450 is used to determine the ground surface temperature candidate value corresponding to the minimum value among the multiple spectral smoothing loss values ​​as the ground surface temperature estimation value of the pixel in the target area.

[0150] According to an embodiment of the present invention, by extracting the radiation observation value at the entrance pupil from the target spectral image, the radiation observation value at the entrance pupil is processed by the atmospheric correction regression network corresponding to the target load, and the atmospheric spectral parameter set of each pixel in the target area is obtained. For any pixel in the target area, the atmospheric spectral parameter set corresponding to the pixel is processed by the atmospheric radiation transfer model to generate the off-ground spectral radiation estimation value of the pixel. According to the off-ground spectral radiation estimation value and multiple surface temperature candidate values ​​corresponding to the pixel, a spectral smoothing loss value corresponding to each surface temperature candidate value is generated to determine the surface temperature estimation value of the pixel in the target area. Since the present application uses a combination of the atmospheric correction regression network and the atmospheric radiation transfer model to calculate the surface temperature estimation value, it overcomes the problems of the traditional multi-spectral surface temperature inversion method with few channels and low inversion accuracy under specific conditions, and can improve the calculation accuracy and efficiency of the surface temperature. At the same time, the inversion of the surface temperature has high interpretability.

[0151] Figure 5 A block diagram of a training device for an atmospheric correction regression network according to an embodiment of the present invention is shown.

[0152] like Figure 5 As shown, the training device 500 for the atmospheric correction regression network includes a second acquisition module 510 , a third processing module 520 , a calculation module 530 , and a training module 540 .

[0153] A second acquisition module 510 is used to acquire a land surface temperature training set, wherein the land surface temperature training set includes a plurality of training samples corresponding to the target load, and the training samples include an entrance pupil radiation training value corresponding to at least one training area type and an atmospheric radiation label parameter corresponding to each entrance pupil radiation training value;

[0154] The third processing module 520 is used to input the entrance pupil radiation training value into the deep residual network for each training sample to obtain the atmospheric radiation prediction parameter;

[0155] A calculation module 530, used for inputting the atmospheric radiation prediction parameter and the atmospheric radiation label parameter corresponding to the atmospheric radiation prediction parameter into the target loss function, and outputting the target loss value;

[0156] The training module 540 is used to iteratively adjust the deep residual network according to the target loss value to obtain a trained atmospheric correction regression network.

[0157] According to an embodiment of the present invention, the training samples are generated in the following manner:

[0158] An acquisition unit, for acquiring, for each training area type, atmospheric profile data corresponding to the training area type and a payload observation geometry of the target payload;

[0159] The first generating unit is used to generate entrance pupil radiation training values ​​and atmospheric radiation label parameters based on the spectral response function, according to the atmospheric profile data and the payload observation geometry.

[0160] According to an embodiment of the present invention, the ground surface temperature inversion device 500 further includes:

[0161] The denoising module is used to denoise the off-ground spectral radiation estimation value of each pixel by using an outlier detection method, so that the off-ground spectral radiation estimation value becomes smooth in the target spectral interval, and the filtered off-ground spectral radiation estimation value is obtained, so as to calculate the spectral smoothing loss value by using the filtered off-ground spectral radiation estimation value.

[0162] According to an embodiment of the present invention, the generation module 440 includes:

[0163] The first generation submodule is used to generate, for each pixel, a plurality of surface temperature candidate values ​​corresponding to the pixel according to the off-ground spectral radiation estimation value;

[0164] The second generation submodule is used to obtain, for each candidate surface temperature value, a spectral smoothing loss value corresponding to the candidate surface temperature value according to the candidate surface temperature value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set.

[0165] According to an embodiment of the present invention, the first generation submodule includes:

[0166] A determination unit, used for determining an initial estimated value of ground surface temperature according to the estimated value of off-ground spectral radiation;

[0167] The second generating unit is used to generate a candidate number of surface temperature candidate values ​​based on the candidate temperature formula and the initial estimated value of the surface temperature, wherein any surface temperature candidate value is located in a candidate interval, and the candidate interval is determined based on the initial estimated value of the surface temperature and the length of the candidate interval.

[0168] According to an embodiment of the present invention, the second generation submodule includes:

[0169] A third generating unit is used to generate a spectral emissivity estimation value of any spectral band under a target spectrum according to the surface temperature candidate value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set, wherein the target spectrum includes a plurality of spectral bands;

[0170] A fourth generating unit is used to generate, for any spectral band, a smoothed estimated value of the off-ground spectral radiation according to the spectral emissivity estimated value of the spectral band, the spectral emissivity estimated value of the associated band and a set of atmospheric spectral parameters;

[0171] The fifth generating unit is used to generate a spectral smoothing loss value corresponding to the candidate value of the ground surface temperature according to the off-ground spectral radiation smoothing estimation value and the off-ground spectral radiation estimation value.

[0172] According to an embodiment of the present invention, the atmospheric spectral parameter set includes an atmospheric downwelling spectral radiation prediction value.

[0173] According to an embodiment of the present invention, the fourth generating unit includes:

[0174] A first generating subunit is used to generate a smoothed spectral emissivity estimate value according to the spectral emissivity estimate value of the spectral band and the spectral emissivity estimate value of the associated band based on the spectral smoothing principle;

[0175] The second generating subunit is used to generate a smoothed estimate of the off-ground spectral radiation based on the thermal infrared transmission equation, the smoothed estimate of the spectral emissivity, the candidate value of the surface temperature and the predicted value of the atmospheric downlink spectral radiation.

[0176] According to an embodiment of the present invention, the atmospheric spectral parameter set includes an atmospheric upward spectral radiation prediction value and an atmospheric spectral transmittance prediction value, and the atmospheric radiation transmission model includes a thermal infrared transmission model.

[0177] According to an embodiment of the present invention, the second processing module 430 includes:

[0178] The third generation submodule is used to use the thermal infrared transmission model to process the radiation observation value at the entrance pupil, the atmospheric upwelling spectral radiation prediction value and the atmospheric spectral transmittance prediction value to generate the off-ground spectral radiation estimation value.

[0179] According to the embodiments of the present invention, any one or more of the modules, units, and sub-units, or at least part of the functions of any one of them, can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, units, and sub-units can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, units, and sub-units can be at least partially implemented as a hardware circuit, such as a field programmable gate array (Field Programmable Gate Array, FPGA), a programmable logic array (Programmable Logic Arrays, PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuit, or in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, units, and sub-units can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0180] For example, any of the acquisition module 410, the first processing module 420, the second processing module 430, the generation module 440, the determination module 450, or the second acquisition module 510, the third processing module 520, the calculation module 530, and the training module 540 can be combined in one module / unit / sub-unit for implementation, or any of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present invention, at least one of the acquisition module 410, the first processing module 420, the second processing module 430, the generation module 440, the determination module 450, or the second acquisition module 510, the third processing module 520, the calculation module 530, and the training module 540 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware or in an appropriate combination of any of them. Alternatively, at least one of the acquisition module 410, the first processing module 420, the second processing module 430, the generation module 440, the determination module 450, or the second acquisition module 510, the third processing module 520, the calculation module 530, and the training module 540 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.

[0181] It should be noted that the surface temperature inversion device part in the embodiment of the present invention corresponds to the surface temperature inversion method part in the embodiment of the present invention, and the description of the surface temperature inversion device part specifically refers to the surface temperature inversion method part, which will not be repeated here. Similarly, the training device part of the atmospheric correction regression network in the embodiment of the present invention corresponds to the training method part of the atmospheric correction regression network in the embodiment of the present invention, and the description of the training device part of the atmospheric correction regression network specifically refers to the training method part of the atmospheric correction regression network, which will not be repeated here.

[0182] Figure 6 A block diagram of an electronic device suitable for implementing the method described above according to an embodiment of the present invention is shown. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0183] like Figure 6As shown, the electronic device 600 according to an embodiment of the present invention includes a processor 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 to the random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0184] In RAM 603, various programs and data required for the operation of electronic device 600 are stored. Processor 601, ROM 602 and RAM 603 are connected to each other via bus 604. Processor 601 performs various operations of the method flow according to the embodiment of the present invention by executing the program in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 can also perform various operations of the method flow according to the embodiment of the present invention by executing the program stored in the one or more memories.

[0185] According to an embodiment of the present invention, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The system 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0186] According to an embodiment of the present invention, the method flow according to an embodiment of the present invention can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system of the embodiment of the present invention are executed. According to an embodiment of the present invention, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0187] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiment; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present invention is implemented.

[0188] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM (Erasable Programmable Read Only Memory, EPROM) or flash memory), a portable compact disk read-only memory (Computer Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device.

[0189] For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 602 and / or the RAM 603 described above and / or one or more memories other than the ROM 602 and the RAM 603 .

[0190] An embodiment of the present invention also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method provided by the embodiment of the present invention. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the method provided by the embodiment of the present invention.

[0191] When the computer program is executed by the processor 601, the above functions defined in the system / device of the embodiment of the present invention are executed. According to the embodiment of the present invention, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0192] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0193] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0194] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways, even if such a combination or combination is not explicitly recorded in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways. All these combinations and / or combinations fall within the scope of the present invention.

[0195] The embodiments of the present invention have been described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment has been described above, this does not mean that the measures in each embodiment cannot be advantageously used in combination. The scope of the present invention is limited by the embodiments and their equivalents. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A surface temperature inversion method, characterized in that: include: In response to the temperature inversion instruction, a target spectrum image collected by the target payload is acquired, and a radiation observation value at the entrance pupil is extracted from the target spectrum image, wherein the target spectrum image is an image under the target environmental state obtained after the target payload passes through the target area; Processing the radiation observation value at the entrance pupil by using an atmospheric correction regression network corresponding to the target load to obtain an atmospheric spectral parameter set of each pixel in the target area, wherein the atmospheric spectral parameter set includes a plurality of parameters characterizing the radiation and transmittance of the target area; For any pixel in the target area, use an atmospheric radiation transfer model to process the atmospheric spectral parameter set corresponding to the pixel to generate an off-ground spectral radiation estimation value of the pixel; Generate a spectral smoothing loss value corresponding to each of the ground surface temperature candidate values ​​according to the off-ground spectral radiation estimation value and a plurality of ground surface temperature candidate values ​​corresponding to the pixel; Determine the candidate ground temperature value corresponding to the minimum value among the plurality of spectral smoothing loss values ​​as the ground temperature estimation value of the pixel in the target area; The atmospheric correction regression network is trained in the following way: Acquire a land surface temperature training set, wherein the land surface temperature training set includes a plurality of training samples corresponding to the target load, and the training samples include an entrance pupil radiation training value corresponding to at least one training area type and an atmospheric radiation label parameter corresponding to each entrance pupil radiation training value; For each of the training samples, the entrance pupil radiation training value is input into a deep residual network to obtain atmospheric radiation prediction parameters; Inputting the atmospheric radiation prediction parameter and the atmospheric radiation label parameter corresponding to the atmospheric radiation prediction parameter into a target loss function, and outputting a target loss value; Iteratively adjusting the deep residual network according to the target loss value to obtain a trained atmospheric correction regression network; Before generating the spectrum smoothing loss value, the method further includes: For each of the pixels, the off-ground spectral radiation estimation value of the pixel is denoised using an outlier detection method, so that the off-ground spectral radiation estimation value becomes smooth in a target spectral interval, and a filtered off-ground spectral radiation estimation value is obtained, so as to calculate the spectral smoothing loss value using the filtered off-ground spectral radiation estimation value; The step of generating a spectral smoothing loss value corresponding to each of the ground surface temperature candidate values ​​according to the off-ground spectral radiation estimation value and a plurality of ground surface temperature candidate values ​​corresponding to the pixel comprises: For each of the pixel, generating a plurality of candidate ground surface temperature values ​​corresponding to the pixel according to the off-ground spectral radiation estimation value; For each of the surface temperature candidate values, the spectrum smoothing loss value corresponding to the surface temperature candidate value is obtained according to the surface temperature candidate value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set.

2. The method according to claim 1, characterized in that: The training samples are generated in the following way: For each of the training area types, obtaining atmospheric profile data corresponding to the training area type and the payload observation geometry of the target payload; Based on the spectral response function, the entrance pupil radiation training value and the atmospheric radiation label parameter are generated according to the atmospheric profile data and the payload observation geometry.

3. The method according to claim 1, characterized in that Generating a plurality of ground surface temperature candidate values ​​corresponding to the pixel according to the off-ground spectral radiation estimation value, comprising: Determining an initial estimate of ground surface temperature based on the estimated off-ground spectral radiation value; Based on the candidate temperature formula, a candidate number of surface temperature candidate values ​​are generated according to the initial surface temperature estimate, wherein any of the surface temperature candidate values ​​is located in a candidate interval, and the candidate interval is determined according to the initial surface temperature estimate and the candidate interval length.

4. The method according to claim 1, characterized in that: Obtaining the spectral smoothing loss value corresponding to the candidate ground temperature value according to the candidate ground temperature value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set, comprising: Generate a spectral emissivity estimation value of any spectral band under a target spectrum according to the candidate ground temperature value, the off-ground spectral radiation estimation value and the set of atmospheric spectral parameters, wherein the target spectrum includes a plurality of the spectral bands; For any of the spectral bands, generating a smoothed estimate of the off-ground spectral radiation according to the spectral emissivity estimate of the spectral band, the spectral emissivity estimate of the associated band and the set of atmospheric spectral parameters; The spectral smoothing loss value corresponding to the ground surface temperature candidate value is generated according to the off-ground spectral radiation smoothing estimation value and the off-ground spectral radiation estimation value.

5. The method according to claim 4, characterized in that The atmospheric spectral parameter set includes the atmospheric downlink spectral radiation prediction value; The method of generating a smoothed estimate of the off-ground spectral radiation according to the spectral emissivity estimate of the spectral band, the spectral emissivity estimate of the associated band and the set of atmospheric spectral parameters comprises: Based on the spectral smoothing principle, generating a smoothed spectral emissivity estimate value according to the spectral emissivity estimate value of the spectral band and the spectral emissivity estimate value of the associated band; Based on the thermal infrared transmission equation, the off-ground spectral radiation smoothed estimate value is generated according to the spectral emissivity smoothed estimate value, the surface temperature candidate value and the atmospheric downlink spectral radiation prediction value.

6. The method according to claim 1, characterized in that The atmospheric spectral parameter set includes an atmospheric upward spectral radiation prediction value and an atmospheric spectral transmittance prediction value, and the atmospheric radiation transmission model includes a thermal infrared transmission model; The step of processing the atmospheric spectral parameter set corresponding to the pixel using the atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel includes: The thermal infrared transmission model is used to process the radiation observation value at the entrance pupil, the atmospheric upward spectral radiation prediction value and the atmospheric spectral transmittance prediction value to generate the off-ground spectral radiation estimation value.

7. A surface temperature inversion device, characterized in that: include: An acquisition module, configured to acquire a target spectrum image collected by a target payload in response to a temperature inversion instruction, and extract a radiation observation value at an entrance pupil from the target spectrum image, wherein the target spectrum image is an image under a target environmental state obtained after the target payload passes through a target area; A first processing module is used to process the radiation observation value at the entrance pupil by using the atmospheric correction regression network corresponding to the target load to obtain an atmospheric spectral parameter set of each pixel in the target area, wherein the atmospheric spectral parameter set includes a plurality of parameters characterizing the radiation and transmittance of the target area; A second processing module is used to process the atmospheric spectral parameter set corresponding to any pixel in the target area using an atmospheric radiation transfer model to generate an off-ground spectral radiation estimation value of the pixel; A generating module, configured to generate a spectral smoothing loss value corresponding to each of the candidate ground temperature values ​​according to the off-ground spectral radiation estimation value and a plurality of candidate ground temperature values ​​corresponding to the pixel; A determination module, configured to determine a surface temperature candidate value corresponding to a minimum value among a plurality of the spectral smoothing loss values ​​as a surface temperature estimation value of the pixel in the target area; Among them, the training device of the atmospheric correction regression network includes: A second acquisition module is used to acquire a surface temperature training set, wherein the surface temperature training set includes a plurality of training samples corresponding to the target load, and the training samples include an entrance pupil radiation training value corresponding to at least one training area type and an atmospheric radiation label parameter corresponding to each entrance pupil radiation training value; The third processing module is used to input the entrance pupil radiation training value into the deep residual network for each training sample to obtain the atmospheric radiation prediction parameter; A calculation module, used for inputting the atmospheric radiation prediction parameters and the atmospheric radiation label parameters corresponding to the atmospheric radiation prediction parameters into the target loss function, and outputting the target loss value; A training module, for iteratively adjusting the deep residual network according to the target loss value to obtain a trained atmospheric correction regression network; Wherein, the surface temperature inversion device further comprises: A denoising module is used for denoising the off-ground spectral radiation estimation value of each pixel by using an outlier detection method, so that the off-ground spectral radiation estimation value becomes smooth in the target spectral interval, and a filtered off-ground spectral radiation estimation value is obtained, so as to calculate a spectral smoothing loss value by using the filtered off-ground spectral radiation estimation value; Wherein, the generating module comprises: The first generation submodule is used to generate, for each pixel, a plurality of surface temperature candidate values ​​corresponding to the pixel according to the off-ground spectral radiation estimation value; The second generation submodule is used to obtain, for each candidate surface temperature value, a spectral smoothing loss value corresponding to the candidate surface temperature value according to the candidate surface temperature value, the off-ground spectral radiation estimation value and the atmospheric spectral parameter set.