Accelerated calculation method and device for geogas background light scattering
By preloading machine learning models and meteorological databases, using multi-threaded parallel ray tracing and cross-platform batch data processing, the problem of time-consuming traditional ground-atmosphere background light scattering calculations has been solved, achieving fast and accurate light scattering simulation.
Patent Information
- Application Number
- CN202511533529.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-25
- Publication Date
- 2026-01-30
AI Technical Summary
Traditional methods for calculating background light scattering from the Earth's atmosphere are too time-consuming and cannot achieve rapid calculations.
A parameterized model trained using machine learning and a meteorological database are preloaded into memory. Combined with multi-threaded parallel ray tracing and cross-platform batch data processing, radiance is calculated through cross-platform API calls.
It significantly improves the simulation accuracy and calculation speed of the distribution characteristics of ground-atmosphere light scattering, from hours to seconds, while maintaining the accuracy of the physical model.
Smart Images
Figure CN121435489A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of light scattering characteristic simulation and calculation technology, and in particular to an accelerated calculation method and apparatus for ground-atmosphere background light scattering. Background Technology
[0002] With the rapid development of space exploration technology, accurate simulation of the Earth-atmosphere background light scattering characteristics has become a key technology in fields such as space target detection, meteorological remote sensing, and environmental monitoring. In space exploration scenarios, the signals received by sensors contain complex solar radiation scattered from the Earth's surface, atmospheric aerosols, and clouds. The calculation is essentially a radiative transfer problem involving massive amounts of light transmission and interaction. To achieve accurate simulation of the Earth-atmosphere background radiation field, it is necessary to process tens of thousands of light rays and their scattering processes along the path from the sun to the sensor. This results in an extremely large computational scale, and traditional calculation methods are very time-consuming and cannot guarantee rapid computation.
[0003] Therefore, it is necessary to provide an accelerated calculation method for Earth-atmosphere background light scattering. Summary of the Invention
[0004] To address the problem that traditional calculation methods are extremely time-consuming and cannot guarantee rapid computation, this invention provides an accelerated calculation method for Earth-atmosphere background light scattering. The technical solution is as follows: On one hand, the present invention provides an accelerated calculation method for Earth-atmosphere background light scattering, characterized in that the method includes: The parameterized model and meteorological database trained using machine learning are preloaded into memory; A pre-compiled radiative transfer algorithm is used to track several probe rays within the detection field of view, and parameter data corresponding to each probe ray is extracted from a pre-saved meteorological database based on the tracking results. The parameter data of all probe rays are transmitted to the parameterized model at one time using a cross-platform call interface to calculate the radiance and obtain the calculation results; the parameterized model and the radiative transfer algorithm run in different software environments.
[0005] On the other hand, the present invention provides an accelerated computing device based on Earth-atmosphere background light scattering, characterized in that the device comprises: The preloading module is used to preload the parameterized model and meteorological database obtained by machine learning training into memory; The parameter extraction module is used to track several probe rays within the detection field of view using a pre-compiled radiative transfer algorithm, and extract the parameter data corresponding to each probe ray from a pre-saved meteorological database based on the tracking results. The radiance calculation module is used to transfer the parameter data of all probe rays to the parameterized model at one time using a cross-platform call interface to calculate the radiance and obtain the calculation results; the parameterized model and the radiative transfer algorithm run in different software environments.
[0006] On the other hand, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in any one of the present specification.
[0007] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method described in any one of the present specification.
[0008] On the other hand, the present invention provides a computer program product, characterized in that it includes a computer program, which, when executed by a processor, implements the steps of any of the methods described in this specification.
[0009] This invention provides an accelerated calculation method for ground-atmosphere background light scattering. The method first loads a parameterized model trained based on machine learning and a meteorological database into memory through a pre-loading step, establishing a high-speed data access environment. Then, a pre-compiled radiative transfer algorithm is used to track the probe rays within the detection field of view, and the corresponding parameter data is extracted from the meteorological database in memory based on the tracking results. Next, a cross-platform call interface is used to integrate the parameter data of all probe rays into a batch data array and transmit it all at once to the parameterized model running in different software environments for radiance calculation. This significantly improves the simulation accuracy and calculation speed of ground-atmosphere light scattering distribution characteristics. Attached Figure Description
[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 This is a flowchart of an accelerated calculation method for ground-atmosphere background light scattering provided by an example of the present invention; Figure 2 This is a hardware architecture diagram of an electronic device provided in an embodiment of the present invention; Figure 3 This is a structural diagram of an accelerated computing device for ground-atmosphere background light scattering provided in an embodiment of the present invention. Detailed Implementation
[0012] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0013] The inventive concept of this invention lies in: firstly, employing memory preloading technology to eliminate I / O bottlenecks and lay the foundation for high-speed computing; secondly, utilizing multi-threaded parallel ray tracing and mesh indexing mechanisms to achieve rapid parameter extraction; and finally, efficiently transferring data to the parameterized model for computation through cross-platform batch processing. This multi-layered collaborative acceleration design not only achieves an order-of-magnitude improvement in computing speed from traditional hours to seconds, but also maintains the accuracy level of the physical model.
[0014] The following describes the specific implementation of the above concept.
[0015] Please refer to Figure 1 This invention provides an accelerated calculation method for Earth-atmosphere background light scattering, the method comprising: Step 100: The parameterized model and meteorological database obtained by machine learning training are preloaded into memory; Step 102: Use a pre-compiled radiative transfer algorithm to track several probe rays within the detection field of view, and extract the parameter data corresponding to each probe ray from the pre-saved meteorological database based on the tracking results. Step 104: Use the cross-platform call interface to transfer all the parameter data of the probe rays to the parameterized model at once for radiance calculation and obtain the calculation results; the parameterized model and the radiative transfer algorithm run in different software environments.
[0016] In this invention, a pre-loading step is first used to load the parameterized model and meteorological database obtained from machine learning training into memory, establishing a high-speed data access environment. Then, a pre-compiled radiative transfer algorithm is used to track the probe rays within the detection field of view, and the corresponding parameter data is extracted from the meteorological database in memory based on the tracking results. Next, the parameter data of all probe rays are integrated into a batch data array through a cross-platform call interface and transferred at once to the parameterized model running in different software environments for radiance calculation, which significantly improves the simulation accuracy and calculation speed of the ground-atmosphere light scattering distribution characteristics.
[0017] The following description Figure 1 The execution method of each step is shown.
[0018] First, regarding step 100: In some implementations, this invention fundamentally solves the I / O bottleneck problem caused by frequent disk reads and writes in traditional computing by preloading the parameterized prediction model trained based on machine learning and the meteorological support database into memory. This strategy transforms subsequent calls to the model and database into high-speed memory-level access. Furthermore, since the parameterized model and meteorological database only need to be loaded once during initialization in the sequential calculations of multiple detection scenarios, repeated disk access overhead is effectively avoided.
[0019] Regarding step 102: In some implementations, a pre-compiled radiative transfer algorithm is used to track several probe rays within the detection field of view, and parameter data corresponding to each probe ray is extracted from a pre-saved meteorological database based on the tracking results. This may include: A local coordinate system is established based on the detector position, and the detector rays are discretized based on the detection direction to obtain several detector ray vectors in the Earth-fixed coordinate system. The spatial relationship between each probe ray vector and the Earth's surface is calculated and the detection mode is determined. Based on the detection mode, the first parameter data of the target location is extracted from the meteorological database. The first parameter data includes surface albedo, aerosol type, aerosol optical thickness, and cloud optical thickness. The detection modes include near-edge detection mode and non-near-edge detection mode. Based on the detection mode, the second parameter data required for the corresponding parameterized model is calculated, and the first and second parameter data are used as the parameter data corresponding to the detection ray.
[0020] Specifically, the local coordinate system has the detector's position as its origin, the Z-axis pointing to the Earth's center, the X-axis along the latitude line, and the Y-axis determined according to the right-handed coordinate system defined by the X and Z axes; the Earth-fixed coordinate system has the Earth's center as its origin, the X-axis pointing to the intersection of the prime meridian and the equator on the equatorial plane, the Z-axis perpendicular to the equatorial plane pointing to the North Pole, and the Y-axis determined according to the right-handed coordinate system defined by the X and Z axes.
[0021] In a preferred embodiment, the spatial relationship between each probe ray vector and the Earth's surface is calculated and the probe mode is determined. Based on the probe mode, the first parameter data of the target location is extracted from the meteorological database, including: The discriminant is calculated based on the detector position vector and the direction vector of the detector ray, and the detection mode is determined based on the value of the discriminant. When the discriminant is less than zero, the probe ray has no intersection with the Earth's surface, and it is determined to be an edge detection mode. The coordinates of the tangent point between the probe ray and the Earth's surface are calculated as the target position, so as to extract the first parameter data of the target position from the meteorological database. When the discriminant is greater than or equal to zero, the probe ray intersects with the Earth's surface, which is determined to be a non-edge detection mode. The coordinates of the intersection point between the probe ray and the Earth's surface are calculated as the target position, so as to extract the first parameter data of the target position from the meteorological database.
[0022] Specifically, the discriminant is determined by the following formula: in, p Let be the detector position vector. d To detect the direction vector of the light ray, R The radius is the Earth's radius.
[0023] The distance between the detector and the point where the probe ray intersects the Earth's surface is determined by the following formula: in, t int For detector p And to detect the distance between the point where the light rays intersect with the Earth's surface.
[0024] The coordinates of the intersection point in the Earth-fixed coordinate system are determined by the following formula: in, r int The coordinates of the intersection point in the Earth-fixed coordinate system.
[0025] The zenith angle and azimuth angle coordinates of the intersection point are determined by the following formula: in,( x int , y int , z int ) The coordinates of the intersection point in the Earth-fixed coordinate system are: The coordinates of the zenith angle and the coordinates of the azimuth angle of the intersection point.
[0026] The index coordinates of the intersection point on the Earth's surface latitude and longitude grid are determined by the following formula: in, The symbol for rounding down is ( m int , n int ) represents the index coordinates of the intersection point on the Earth's surface latitude and longitude grid.
[0027] In this invention, after obtaining the intersection point of the light rays with the Earth's surface through reverse ray tracing calculation, the latitude and longitude coordinates of the intersection point are quickly mapped to the integer index of the meteorological support database by rounding down based on the uniform latitude and longitude spherical grid positioning technology. This enables the efficient extraction of surface albedo, aerosol parameters, and cloud parameters. Since the division scale of the uniform latitude and longitude spherical grid is strictly consistent with the spatial resolution of the surface and the meteorological support database, the parameter extraction process can be accurately located without complex calculations, providing complete and accurate input data for subsequent model calculations.
[0028] In a preferred embodiment, based on the detection mode, the second parameter data required for inputting the corresponding parameterized model is calculated, including: When the detection mode of the probe ray vector is the edge detection mode, the solar zenith angle, edge cutting height and relative azimuth angle are calculated as the second parameter data; When the detection mode of the probe ray vector is the non-near edge detection mode, the solar zenith angle, the observed zenith angle, and the relative azimuth angle are calculated as the second parameter data; Specifically, the second parameter data is determined by the following formula: in, θ i The zenith angle of the sun. θ r To observe the zenith angle, The relative azimuth angle. e Let be the unit vector in the direction of detection. n To detect the normal vector at the point where the light ray intersects the Earth's surface, s n and e n These are the projection vectors onto the tangent plane of the unit vectors of the solar direction and the probe direction, respectively. s This is the direction vector of the sun.
[0029] Specifically, the solar direction vector is calculated using the following formula: Where n is the ordinal number day, B The intermediate angle variable introduced for calculating the time difference correction term. The solar declination, L To determine the geographical longitude of the subsolar point at a given time, For the detection time, For time zone correction, Standard meridian longitude, True solar time s This is the direction vector of the sun.
[0030] In some implementations, after obtaining several probe ray vectors, the parameter data of each probe ray vector is calculated in a multi-threaded parallel manner.
[0031] In this invention, multi-threaded parallel technology is applied to the detection mode determination of the probe ray and the calculation of the first and second parameter data. This enables multiple probe rays to perform multi-threaded parallel calculations of the required parameter data. When the number of discrete probe rays in the detection field of view reaches tens of thousands or more, the traditional sequential calculation method cannot meet the real-time response requirements. This invention dynamically allocates the calculation task of massive rays to multiple calculation threads for synchronous execution, making full use of the parallel computing capabilities of modern multi-core processors and significantly improving the overall calculation efficiency.
[0032] Regarding step 104: In some implementations, the parameterized model includes a near-edge detection parameterized model and a non-near-edge detection parameterized model; the parameterized model runs on the TensorFlow platform; The cross-platform API is used to transfer all probe ray parameter data to the parameterized model at once for radiance calculation, including: All probe ray parameter data are transferred to the TensorFlow platform at once using a cross-platform API call; Input the parameter data corresponding to the detection ray vector of the edge detection mode into the edge detection parameterization model to calculate the radiance; The parameter data corresponding to the probe ray vector of the non-proximal detection mode are input into the non-proximal detection parameterized model for radiance calculation.
[0033] In this invention, the parameterized model is trained and debugged on the TensorFlow platform using machine learning methods, enabling accurate simulation of complex radiative transfer physics. The ray tracing-based radiative transfer algorithm is compiled and optimized in a Fortran environment, fully leveraging its performance advantages in scientific computing. Addressing the traditional method where each probe ray requires separate parameter data input and triggers a cross-platform API call, causing the parameterized model to run once, resulting in I / O overhead that linearly increases with the number of rays and severely restricts computational speed, this invention employs a multi-level calling strategy: establishing a data channel between the Fortran program and the TensorFlow model through a cross-platform API written in C. In implementation, by integrating all probe ray parameter data into a batch data array in memory and transferring it to the parameterized model all at once, while simultaneously returning all prediction results, this batch processing mechanism completely changes the inefficient sequential processing mode of traditional methods, significantly reducing I / O overhead and inter-process communication frequency during computation. Furthermore, deep optimization of the compilation and linking process of the integrated development environment further improves the overall program execution efficiency, ultimately achieving a significant increase in computational speed.
[0034] like Figure 2 , Figure 3 As shown, this embodiment of the invention provides an accelerated computing device for ground-atmosphere background light scattering. The device embodiment can be implemented through software, hardware, or a combination of both. From a hardware perspective, such as... Figure 2 The diagram shown is a hardware architecture diagram of a computing device containing an accelerated computing device for ground-atmosphere background light scattering provided in an embodiment of the present invention, except for... Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, the computing device in the embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, such as... Figure 3 As shown, a device in a logical sense is formed by the CPU of the computing device in which it is located reading the corresponding computer program from the non-volatile memory into the memory for execution.
[0035] like Figure 3 As shown, an embodiment of the present invention provides an accelerated computing device for geothermal background light scattering, comprising: The preloading module 300 is used to preload the parameterized model and meteorological database obtained by machine learning training into memory; The parameter extraction module 302 is used to track several probe rays within the detection field of view using a pre-compiled radiative transfer algorithm, and extract the parameter data corresponding to each probe ray from a pre-saved meteorological database based on the tracking results. The radiance calculation module 304 is used to transfer the parameter data of all probe rays to the parameterized model at one time using a cross-platform call interface to calculate the radiance and obtain the calculation results; the parameterized model and the radiative transfer algorithm run in different software environments.
[0036] In some specific implementations, the preloading module 300 can be used to perform the above step 100, the parameter extraction module 302 can be used to perform the above step 102, and the radiance calculation module 304 can be used to perform the above step 104.
[0037] In some specific implementations, the parameter extraction module 302 is used to perform the following operations: A local coordinate system is established based on the detector position, and the detector rays are discretized based on the detection direction to obtain several detector ray vectors in the Earth-fixed coordinate system. The spatial relationship between each probe ray vector and the Earth's surface is calculated and the detection mode is determined. Based on the detection mode, the first parameter data of the target location is extracted from the meteorological database. The first parameter data includes surface albedo, aerosol type, aerosol optical thickness, and cloud optical thickness. The detection modes include near-edge detection mode and non-near-edge detection mode. Based on the detection mode, the second parameter data required for the corresponding parameterized model is calculated, and the first and second parameter data are used as the parameter data corresponding to the detection ray.
[0038] In some specific implementations, the parameter extraction module 302 is also used to perform the following operations: The discriminant is calculated based on the detector position vector and the direction vector of the detector ray, and the detection mode is determined based on the value of the discriminant. When the discriminant is less than zero, the probe ray has no intersection with the Earth's surface, and it is determined to be an edge detection mode. The coordinates of the tangent point between the probe ray and the Earth's surface are calculated as the target position, so as to extract the first parameter data of the target position from the meteorological database. When the discriminant is greater than or equal to zero, the probe ray intersects with the Earth's surface, which is determined to be a non-edge detection mode. The coordinates of the intersection point between the probe ray and the Earth's surface are calculated as the target position, so as to extract the first parameter data of the target position from the meteorological database.
[0039] In some specific implementations, the radiance calculation module 304 is used to perform the following operations: All probe ray parameter data are transferred to the TensorFlow platform at once using a cross-platform API call; Input the parameter data corresponding to the detection ray vector of the edge detection mode into the edge detection parameterization model to calculate the radiance; The parameter data corresponding to the probe ray vector of the non-proximal detection mode are input into the non-proximal detection parameterized model for radiance calculation.
[0040] In some specific implementations, the radiance calculation module 304 is also used to perform the following operations: When the detection mode of the probe ray vector is the edge detection mode, the solar zenith angle, edge cutting height and relative azimuth angle are calculated as the second parameter data; When the detection mode of the probe ray vector is the non-edge detection mode, the solar zenith angle, the observed zenith angle, and the relative azimuth angle are calculated as the second parameter data.
[0041] In some specific embodiments, the device further includes an operating device for performing the following operations: After obtaining several probe ray vectors, the parameter data of each probe ray vector is calculated in a multi-threaded parallel manner.
[0042] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on a computing device for accelerating ground-atmosphere background light scattering. In other embodiments of the present invention, a computing device for accelerating ground-atmosphere background light scattering may include more or fewer components than illustrated, or combine some components, or split some components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.
[0043] The information interaction and execution process between the modules in the above-mentioned device are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description of the method embodiment of the present invention, and will not be repeated here.
[0044] This invention also provides a computing device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements an accelerated calculation method for ground-atmosphere background light scattering according to any embodiment of this invention.
[0045] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform an accelerated calculation method for ground-atmosphere background light scattering according to any embodiment of this invention.
[0046] Embodiments of this application also provide a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium and executes the computer program, causing the computer device to perform an accelerated calculation method for ground-atmosphere background light scattering as described in any of the above embodiments.
[0047] Specifically, an apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the apparatus may read and execute the program code stored in the storage medium.
[0048] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.
[0049] Storage media embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.
[0050] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating devices on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the above embodiments.
[0051] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.
[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0053] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for accelerated computation of ground-gas background light scattering, characterized in that, The method comprises: loading a parameterized model trained by machine learning and a meteorological database into memory in advance; tracking a plurality of detection light rays in a detection field of view using a pre-compiled radiation transfer algorithm, and extracting parameter data corresponding to each of the detection light rays from a pre-stored meteorological database according to a tracking result; transferring the parameter data of all the detection light rays to the parameterized model at one time through a cross-platform calling interface for radiance calculation to obtain a calculation result; the parameterized model and the radiation transfer algorithm run in different software environments.
2. The method of claim 1, wherein, The tracking of the plurality of detection light rays in the detection field of view using the pre-compiled radiation transfer algorithm, and the extraction of the parameter data corresponding to each of the detection light rays from the pre-stored meteorological database according to the tracking result, comprises: establishing a local coordinate system based on a detector position, and discretizing the detector light rays based on a detection direction to convert a plurality of detection light ray vectors in a geostationary coordinate system; calculating a spatial relationship between each of the detection light ray vectors and the earth's surface and determining a detection mode, and extracting first parameter data of a target position from the meteorological database based on the detection mode; the first parameter data includes a surface albedo, an aerosol type, an aerosol optical thickness and a cloud optical thickness; the detection mode includes an edge detection mode and a non-edge detection mode; calculating second parameter data required for inputting into a corresponding parameterized model based on the detection mode, and taking the first parameter data and the second parameter data as the parameter data corresponding to the detection light ray.
3. The method of claim 2, wherein, The calculation of the spatial relationship between each of the detection light ray vectors and the earth's surface and the determination of the detection mode, and the extraction of the first parameter data of the target position from the meteorological database based on the detection mode, comprise: calculating a discriminant based on a detector position vector and a direction vector of the detection light ray, and determining the detection mode according to a value of the discriminant; when the discriminant is less than zero, the detection light ray has no intersection with the earth's surface, and the edge detection mode is determined, and a tangent point coordinate of the detection light ray and the earth's surface is calculated as the target position to extract the first parameter data of the target position from the meteorological database; when the discriminant is greater than or equal to zero, the detection light ray intersects with the earth's surface, and the non-edge detection mode is determined, and an intersection point coordinate of the detection light ray and the earth's surface is calculated as the target position to extract the first parameter data of the target position from the meteorological database.
4. The method of claim 2, wherein, The parameterized model comprises an edge detection parameterized model and a non-edge detection parameterized model; the parameterized model runs on a tensorflow platform; The transfer of the parameter data of all the detection light rays to the parameterized model at one time through the cross-platform calling interface for radiance calculation comprises: transferring the parameter data of all the detection light rays to the tensorflow platform through the cross-platform calling interface at one time; inputting the parameter data corresponding to the detection light ray vector of the edge detection mode into the edge detection parameterized model for radiance calculation; inputting the parameter data corresponding to the detection light ray vector of the non-edge detection mode into the non-edge detection parameterized model for radiance calculation.
5. The method of claim 2, wherein, Based on the detection mode, second parameter data required for inputting the parameterized model is calculated, including: When the detection mode of the detection light vector is the edge detection mode, the solar zenith angle, the edge tangent height, and the relative azimuth angle are calculated as the second parameter data; When the detection mode of the detection light vector is the non-edge detection mode, the solar zenith angle, the observation zenith angle, and the relative azimuth angle are calculated as the second parameter data.
6. The method according to any one of claims 2-5, characterized in that, After obtaining a plurality of detection light vectors, the parameter data of each detection light vector is calculated in a multi-thread parallel mode.
7. An accelerated computing device for atmospheric background light scattering, characterized in that, The device comprises: A preloading module for preloading a parameterized model trained by machine learning and a meteorological database into a memory; A parameter extraction module for tracking a plurality of detection light rays in a detection field of view by using a pre-compiled radiation transfer algorithm, and extracting parameter data corresponding to each detection light ray from a pre-stored meteorological database according to a tracking result; An irradiance calculation module for transmitting the parameter data of all detection light rays to the parameterized model at one time by using a cross-platform calling interface to perform irradiance calculation, and obtaining a calculation result; the parameterized model and the radiation transfer algorithm run in different software environments.
8. A computer device, comprising: The computer device comprises a memory and a processor, the memory is used to store a computer program, and the processor is used to execute the computer program stored on the memory to realize the steps of the method of any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is executed by the processor to realize the steps of the method of any one of claims 1-6.
10. A computer program product, characterised in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1-6.