Hyperspectral remote sensing image generation method and device based on three-dimensional scene radiation simulation

Abstract

The invention provides a hyperspectral remote sensing image generation method and device based on three-dimensional scene radiation simulation, relates to the technical field of remote sensing data processing, and aims at solving the technical problems that in the prior art, hyperspectral remote sensing image data acquisition cost is too high, data sources are few, and the resolution is low. The method comprises the following steps: determining three-dimensional scene data according to a user demand; inputting the three-dimensional scene data into a pre-constructed hyperspectral imaging simulation model, and executing the following operations: based on a global coordinate system, starting from an original point, and emitting randomly distributed light to a three-dimensional scene object through a virtual imaging plane, the three-dimensional scene object being used for simulating optical response of a real ground object structure after light incidence; determining an intersection point of each light ray and the three-dimensional scene object; according to the geometric information and the material information corresponding to the intersection point, combining atmospheric radiation parameters, calculating radiation information of light and radiation data of an entrance pupil of the remote sensor; and generating a hyperspectral remote sensing picture according to the radiation data at the entrance pupil.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing data processing, and more specifically, to a method and device for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation. Background Art

[0002] Hyperspectral remote sensing is a technology that acquires reflectance or radiation information from ground objects within a continuous, narrow wavelength band. It can simultaneously record both the spatial and continuous spectral information of a target, enabling precise identification and quantitative analysis of ground objects. Hyperspectral remote sensing technology plays a vital role in environmental monitoring, precision agriculture, mineral and geological exploration, ecological research, and disaster management. However, in practical applications, hyperspectral remote sensing technology is affected by factors such as observation geometry, imaging conditions, and atmospheric variations, resulting in limitations in its data acquisition capabilities and consistency. To address these issues, data acquisition is currently primarily based on hyperspectral imaging simulation technology. This technology integrates physical mechanisms with computer simulation, creating a controllable digital experimental platform for remote sensing research and providing efficient solutions for algorithm testing, data expansion, and remote sensor performance optimization.

[0003] Current hyperspectral imaging simulations primarily utilize two methods: two-dimensional image fusion and three-dimensional scene simulation. Furthermore, in terms of hardware and software acceleration, with the advancement of computers, the latest graphics processing units (GPUs) now offer dedicated ray tracing support. Model architectures based on GPU computing cores are the future direction of hyperspectral remote sensing imaging simulation.

[0004] Three-dimensional scene simulation methods, such as ray tracing algorithms, require significant computational resources, meaning they must use a large number of rays to achieve the required accuracy. The resulting long runtimes and resource consumption hinder the practical application of hyperspectral imaging simulation models. Existing hyperspectral remote sensing image data suffers from technical challenges such as high acquisition costs, limited data sources, and low resolution. Summary of the Invention

[0005] In view of this, the present invention provides a method and device for generating hyperspectral remote sensing images based on three-dimensional scene radiation simulation, aiming to solve technical problems in the prior art such as high cost of acquiring hyperspectral remote sensing image data, few data sources, and low resolution.

[0006] One aspect of the present invention provides a method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation, comprising: determining three-dimensional scene data according to user needs; inputting the three-dimensional scene data into a pre-constructed hyperspectral imaging simulation model, wherein the hyperspectral imaging simulation model is configured with a global coordinate system and a three-dimensional scene object for matching the three-dimensional scene data, and performing the following operations: based on the global coordinate system, starting from the origin, emitting randomly distributed light rays to the three-dimensional scene object through a virtual imaging plane, wherein the virtual imaging plane is used to simulate the imaging surface of the remote sensor, and the three-dimensional scene object is used to simulate the optical response of the real ground structure after the light is incident; determining the intersection point of each light ray with the three-dimensional scene object; calculating the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and in combination with atmospheric radiation parameters; and generating a hyperspectral remote sensing image based on the radiation data at the entrance pupil.

[0007] According to an embodiment of the present invention, a hyperspectral imaging simulation model includes multiple triangular element faces, wherein determining the intersection point of each light ray with a three-dimensional scene object includes: determining whether the light ray intersects with the triangular element face; in response to the light ray intersecting with the triangular element face, determining whether the intersection point is located within the boundary range of the triangular element face; in response to the intersection point being located within the boundary range of the triangular element face, obtaining geometric information and material information of the intersection point.

[0008] According to an embodiment of the present invention, based on the geometric information and material information corresponding to the intersection point and combined with the atmospheric radiation parameters, the radiation data at the entrance pupil of the remote sensor is calculated, including: calculating the first surface reflected radiation parameter of the intersection point from direct solar radiation; calculating the second surface reflected radiation parameter of the intersection point from solar scattered radiation; calculating the third surface reflected radiation parameter of the intersection point from background reflected radiation; based on the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter, the total radiation data at the entrance pupil of the remote sensor is calculated.

[0009] According to an embodiment of the present invention, based on the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter, the total radiation data at the entrance pupil of the remote sensor is calculated, including: calculating the sum of the first surface reflected radiation parameter and the second surface reflected radiation parameter, or calculating the sum of the first surface reflected radiation parameter and the third surface reflected radiation parameter to obtain the surface zero-line-of-sight reflection radiance parameter of each light ray; superimposing the surface zero-line-of-sight reflection radiance parameters of all light rays and performing average calculation to obtain the surface zero-line-of-sight reflection radiance parameter of each pixel; based on the surface zero-line-of-sight reflection radiance parameter of each pixel, fusing the path radiation parameter and the cross radiation to obtain the total radiation data at the entrance pupil of the remote sensor.

[0010] According to an embodiment of the present invention, pre-constructing a hyperspectral imaging simulation model includes: establishing a local coordinate system with the optical center of the remote sensor as the origin, and transforming the position and direction of the remote sensor from the local coordinate system to the global coordinate system according to the specified remote sensor observation geometric information; based on the global coordinate system, starting from the origin, emitting randomly distributed light rays to the three-dimensional scene object through the virtual imaging plane; determining the intersection of each light ray and the three-dimensional scene object; according to the geometric information and material information corresponding to the intersection point, combined with the atmospheric radiation parameters, calculating the radiation data at the entrance pupil of the remote sensor, wherein the radiation data at the entrance pupil represents the actual radiation brightness received by each pixel; and constructing a hyperspectral imaging simulation model based on the generation path of the radiation data at the entrance pupil.

[0011] According to an embodiment of the present invention, a hyperspectral imaging simulation model is implemented based on a graphics processor, wherein the rendering pipeline corresponding to the graphics application programming interface of the graphics processor includes: a path tracing renderer, which is used to calculate the light transmission path result using a ray tracing model based on the input three-dimensional scene data, and output the light transmission path result through a pixel shader and encode it to a texture buffer for storage; a radiometry renderer, which is arranged in series with the path tracing renderer, and is used to calculate the radiation data at the entrance pupil of the remote sensor based on the light transmission path result in combination with spectral radiation parameters.

[0012] According to an embodiment of the present invention, a path tracing renderer is configured to accelerate the processing of a ray tracing model by means of Monte Carlo random sampling, wherein the path tracing renderer includes: a hierarchical bounding box model for accelerating the detection of intersection features between light and three-dimensional scene objects, wherein the hierarchical bounding box model is configured to search for the intersection features between light and the bounding boxes of each node in sequence, starting from the root node, according to a depth-first search strategy; an importance sampling module for preferentially screening out directions or targets that have the greatest impact on the results for sampling during the Monte Carlo random sampling process; a low-discrepancy sequence sampling module for generating a uniformly distributed set of points in the sample space through mathematical construction to improve the convergence speed of the Monte Carlo integral; a simulation accuracy and simulation efficiency balancing module for determining the optimal number of rays based on the simulation accuracy and simulation efficiency.

[0013] Another aspect of the present invention provides a hyperspectral remote sensing image generation device based on three-dimensional scene radiation simulation, including: a first determination module, used to determine three-dimensional scene data according to user needs; an input module, used to input the three-dimensional scene data into a pre-built hyperspectral imaging simulation model, wherein the hyperspectral imaging simulation model is configured with a global coordinate system and a three-dimensional scene object for matching the three-dimensional scene data, and executes the following modules: a transmission module, used to emit randomly distributed light from the origin through a virtual imaging plane to the three-dimensional scene object based on the global coordinate system, wherein the virtual imaging plane is used to simulate the imaging surface of the remote sensor, and the three-dimensional scene object is used to simulate the optical response of the real ground structure after the light is incident; a second determination module, used to determine the intersection of each light ray and the three-dimensional scene object; a calculation module, used to calculate the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point, combined with the atmospheric radiation parameters; a generation module, used to generate hyperspectral remote sensing image data based on the radiation data at the entrance pupil.

[0014] Another aspect 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 above method.

[0015] Another aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above method when executed.

[0016] Another aspect of the present invention provides a computer program product, which includes computer executable instructions. When the instructions are executed, they are used to implement the above method.

[0017] Compared with the prior art, the hyperspectral remote sensing image generation and device based on three-dimensional scene radiation simulation provided by the embodiments of the present invention has at least the following beneficial effects:

[0018] (1) The hyperspectral remote sensing image generation method and device based on three-dimensional scene radiation simulation provided by the embodiment of the present invention, based on the derivation of the ground background and target space-based imaging process, uses a ray tracing algorithm to construct a hyperspectral imaging simulation model. Through this hyperspectral imaging simulation model, starting from the remote sensor, light is emitted into the three-dimensional scene in reverse order, pixel by pixel, and then the propagation path and radiation information source of each light ray are traced to calculate the radiance information of each pixel, and finally generate hyperspectral remote sensing image data. Because the hyperspectral imaging simulation model is equipped with a three-dimensional scene object for simulation, it can quickly simulate scenes under various observation geometric conditions, solar conditions, and atmospheric conditions according to user needs, generate qualified hyperspectral simulation data and images, meet the needs of various fields, and significantly reduce the economic and time costs of obtaining hyperspectral remote sensing data.

[0019] (2) The hyperspectral remote sensing image generation method and device based on three-dimensional scene radiation simulation provided by the embodiments of the present invention utilizes a GPU to accelerate the hyperspectral imaging simulation model. Specifically, by improving the rendering pipeline corresponding to the graphics application programming interface of the graphics processor, a path tracing renderer and a radiosity renderer are introduced to achieve hyperspectral rendering, thereby solving the data input and output problems, quantification problems, and multi-spectral rendering problems of the original rendering pipeline. Because the improved rendering pipeline splits the original ray tracing renderer into two serially executed renderers, the computational complexity of radiation calculations is greatly reduced, thereby improving the efficiency of large-scale hyperspectral simulations.

[0020] (3) The hyperspectral remote sensing image generation method and device based on three-dimensional scene radiation simulation provided by the embodiment of the present invention adopts hierarchical bounding boxes to accelerate the intersection of light and scene, and uses importance sampling and low-discrepancy sequences to accelerate the convergence of Monte Carlo integration. By exploring the relationship between the number of light rays and simulation accuracy, the optimal number of light rays is selected for simulation experiments, achieving a significant improvement in simulation efficiency at a very low cost in accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] 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:

[0022] Figure 1 The flowchart of the method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation according to an embodiment of the present invention is schematically shown.

[0023] Figure 2 The coordinate system diagram of the hyperspectral imaging simulation model according to an embodiment of the present invention is schematically shown.

[0024] Figure 3 The figure schematically shows a display diagram of a sampling point receiving direct solar radiation according to an embodiment of the present invention.

[0025] Figure 4 The diagram schematically shows a display in which a sampling point does not receive direct solar radiation according to an embodiment of the present invention.

[0026] Figure 5 The figure schematically shows a display diagram of solar scattered radiation received by a sampling point according to an embodiment of the present invention.

[0027] Figure 6 The figure schematically shows a display diagram of background reflected radiation received by a sampling point according to an embodiment of the present invention.

[0028] Figure 7 The flowchart of the preliminary improved graphics application programming interface ray tracing RGB rendering pipeline according to an embodiment of the present invention is schematically shown.

[0029] Figure 8 The flowchart of the secondary improved graphics application programming interface ray tracing hyperspectral rendering pipeline according to an embodiment of the present invention is schematically shown.

[0030] Figure 9 The flowchart and functional module block diagram of the hyperspectral remote sensing imaging simulation software according to an embodiment of the present invention are schematically shown.

[0031] Figure 10 The structure block diagram of an electronic device suitable for implementing a method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation according to an embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0032] Hereinafter, 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 the concept of the present invention.

[0033] 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 presence of the features, steps, operations and / or components, but do not exclude the presence or addition of one or more other features, steps, operations or components.

[0034] All terms used herein (including technical and scientific terms) 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.

[0035] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning 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.).

[0036] In the embodiments of the present invention, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of all data involved (including, but not limited to, user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken to prevent unauthorized access to user personal information data and maintain the security of user personal information and network security.

[0037] Current hyperspectral imaging simulation mainly uses two methods: two-dimensional image fusion and three-dimensional scene simulation.

[0038] Research on two-dimensional image fusion methods mainly focuses on the fusion and spectral reconstruction of existing multi-source data. However, since the essence of two-dimensional image fusion methods is a two-dimensional image-to-image mapping process, there are limitations in viewing angles and they cannot meet the needs of multi-angle and multi-condition hyperspectral imaging simulation.

[0039] The three-dimensional scene simulation method can simulate scenes under arbitrary observation geometric conditions, solar conditions and atmospheric conditions, and has more advantages in scenes that pursue more quantitative radiation calculations.

[0040] In terms of hardware and software acceleration, with the development of computers, the latest graphics processing units (GPUs) now have dedicated ray tracing support. However, most of the more mature hyperspectral imaging simulation models are currently implemented on CPUs and lack GPU acceleration.

[0041] In general, existing technologies for hyperspectral remote sensing image data have technical problems such as high acquisition cost, few data sources and low resolution.

[0042] Based on this, an embodiment of the present invention provides a method for generating hyperspectral remote sensing images based on three-dimensional scene radiation simulation, aiming to solve technical problems in the prior art such as high cost of acquiring hyperspectral remote sensing image data, few data sources, and low resolution.

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0044] Figure 1 The flowchart of the method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation according to an embodiment of the present invention is schematically shown.

[0045] like Figure 1 As shown, the method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation in this embodiment may include operations S1 to S5.

[0046] In operation S1 , three-dimensional scene data is determined according to user requirements.

[0047] Input the three-dimensional scene data into the pre-built hyperspectral imaging simulation model, and perform the following operations S2 to S5.

[0048] The hyperspectral imaging simulation model is configured with a global coordinate system and a three-dimensional scene object for matching three-dimensional scene data.

[0049] In operation S2, based on the global coordinate system, starting from the origin, randomly distributed light is emitted to the three-dimensional scene object through the virtual imaging plane, where the virtual imaging plane is used to simulate the imaging surface of the remote sensor, and the three-dimensional scene object is used to simulate the optical response of the real ground structure after the light is incident.

[0050] In operation S3 , the intersection point of each ray with the three-dimensional scene object is determined.

[0051] In operation S4, the radiation data at the entrance pupil of the remote sensor is calculated based on the geometric information and material information corresponding to the intersection point and the atmospheric radiation parameters.

[0052] In operation S5 , hyperspectral remote sensing image data is generated according to the radiation data at the entrance pupil.

[0053] In this embodiment, three-dimensional scene data is first determined according to user requirements. The three-dimensional scene data may include, for example, geometric information and material information (such as reflectivity, scattering properties, etc.) of the surface of the object.

[0054] The three-dimensional scene data is then input into a pre-built hyperspectral imaging simulation model for simulation. Specifically, light is projected into the scene (simulating the perspective of the remote sensor) and the interaction between the light and the three-dimensional scene objects (such as ground objects) in the scene is calculated. At the same time, the influence of the atmosphere, including atmospheric scattering and absorption, is considered, and the changes in light during propagation are calculated. Then, based on the situation of the light reaching the remote sensor after the influence of the atmosphere, the radiation information of the light is calculated and the radiation data at the entrance pupil of the remote sensor, that is, the radiance of each pixel, is obtained. Finally, based on the radiance calculation result, the radiance of each pixel in different bands is combined into a hyperspectral image.

[0055] The hyperspectral remote sensing image generation method based on three-dimensional scene radiation simulation provided by an embodiment of the present invention uses a ray tracing algorithm to construct a hyperspectral imaging simulation model based on the derivation of the ground background and target space-based imaging process. Through this hyperspectral imaging simulation model, starting from the remote sensor, light is emitted into the three-dimensional scene in reverse, pixel by pixel. The propagation path and radiation information source of each light ray are then traced, and the radiance information of each pixel is calculated, ultimately generating hyperspectral remote sensing image data. Because the hyperspectral imaging simulation model is equipped with a three-dimensional scene object for simulation, it can quickly simulate scenes under various observation geometric conditions, solar conditions, and atmospheric conditions according to user needs, generate qualified hyperspectral simulation data and images, meet the needs of various fields, and significantly reduce the economic and time costs of acquiring hyperspectral remote sensing data.

[0056] According to an embodiment of the present invention, pre-building a hyperspectral imaging simulation model may specifically include:

[0057] A local coordinate system is established with the optical center of the remote sensor as the origin, and the position and orientation of the remote sensor are transformed from the local coordinate system to the global coordinate system according to the specified remote sensor observation geometry information;

[0058] Based on the global coordinate system, starting from the origin, randomly distributed light rays are emitted through the virtual imaging plane toward the three-dimensional scene object; and the intersection point of each light ray with the three-dimensional scene object is determined;

[0059] According to the geometric information and material information corresponding to the intersection point, combined with the atmospheric radiation parameters, the radiation data at the entrance pupil of the remote sensor is calculated. The radiation data at the entrance pupil represents the actual radiation brightness received by each pixel.

[0060] Based on the generation path of radiation data at the entrance pupil, a hyperspectral imaging simulation model is constructed.

[0061] In this embodiment, building a hyperspectral imaging simulation model can be specifically divided into the following four steps:

[0062] (1) Establish a coordinate system, such as Figure 2 shown.

[0063] Figure 2 The coordinate system diagram of the hyperspectral imaging simulation model according to an embodiment of the present invention is schematically shown.

[0064] like Figure 2 As shown, first, establish the sensor's local coordinate system, the position of the pinhole in the remote sensor optical system (that is, the optical center of the remote sensor) p o Located at the center of the coordinate system, the direction opposite to the observation direction of the remote sensor is defined as e of the local coordinate system z Axis, the horizontal and vertical directions of the remote sensor imaging plane are defined as the e of the local coordinate system x axis and e y axis.

[0065] Secondly, according to the given remote sensor observation geometry information (such as the remote sensor observation zenith angle θ o , the remote sensor observation azimuth φ o , remote sensor observation distance r o ), defines the position of the remote sensor in the global coordinate system o , e of the local coordinate system of the remote sensor z Axis, e x axis and e y axis:

[0066]

[0067]

[0068]

[0069] e y =e z ×e x

[0070] (2) Project light from the remote sensor.

[0071] Based on the position and orientation of the remote sensor, a ray is projected from the remote sensor through each grid on the virtual imaging plane toward the 3D scene object. The ray can be expressed as follows:

[0072]

[0073] Among them, r represents the current position of the light; o represents the starting point of the light, which is the position of the aperture stop in the remote sensor imaging system, that is, p o; t represents the distance that the light travels; d represents the unit vector along the direction of light propagation, which can be calculated from the position of the pixel point on the remote sensor and the position of the remote sensor imaging point. Suppose the position of a pixel of the remote sensor is p pix , then d can be expressed as:

[0074]

[0075] Based on the random sampling principle of the Monte Carlo method, a large number of randomly distributed sampling rays are emitted from the optical center of the remote sensor to each pixel grid in the virtual imaging plane to obtain a sufficient number of samples to ensure the accuracy of the Monte Carlo integration. pix They are randomly and evenly distributed across the pixel grid.

[0076] This method can also realistically simulate the mixed pixel effect in the remote sensing imaging process.

[0077] (3) Calculation of the intersection between light and scene.

[0078] According to an embodiment of the present invention, the hyperspectral imaging simulation model includes a plurality of triangular surfaces, wherein determining the intersection point of each ray with the three-dimensional scene object may specifically include:

[0079] Determine whether the ray intersects the triangle face;

[0080] In response to the ray intersecting the triangle face, determining whether the intersection point is within the boundary of the triangle face;

[0081] In response to the intersection point being within the boundary of the triangle face, geometric information and material information of the intersection point are obtained.

[0082] In this embodiment, the intersection calculation between the light and the scene is divided into the following two steps (31) and (32):

[0083] (31) Ray-plane intersection detection and intersection point calculation.

[0084] First, we need to determine whether the ray intersects the plane containing the triangle. This is accomplished by solving the ray expression equation simultaneously with the plane equation. If the simultaneous equations have a real solution, we calculate the coordinates of the intersection of the ray and the plane and verify that the intersection lies in the effective propagation direction of the ray.

[0085] (32) Intersection-triangle face containment test.

[0086] After confirming that a ray intersects a plane, it's necessary to determine whether the intersection lies within the boundaries of the triangle. By calculating the position of the intersection point relative to the triangle's edges, a geometric criterion is established to determine its inclusion. If the intersection satisfies all inclusion criteria, the 3D coordinates, normal, and other geometric information of the intersection are returned. The radiometric characteristics of the triangle are then determined based on the optical properties (material information) of the triangle.

[0087] (4) Calculation of radiation at the entrance pupil of the remote sensor.

[0088] According to an embodiment of the present invention, the calculation of the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and the atmospheric radiation parameters may specifically include:

[0089] Calculate the first surface reflected radiation parameter from direct solar radiation at the intersection point;

[0090] Calculate the second surface reflected radiation parameter from the sun's scattered radiation at the intersection point;

[0091] Calculate the third surface reflected radiation parameter from the background reflected radiation at the intersection point;

[0092] Based on the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter, the total radiation data at the entrance pupil of the remote sensor is calculated.

[0093] In this embodiment, after obtaining the intersection of the light and the scene, the surface reflected radiance received by the remote sensor can be calculated based on the geometric information of the intersection and the light (exit angle, incident angle, intersection normal) and the material information of the intersection, combined with the atmospheric radiation parameters during simulation (the irradiance of direct solar radiation reaching the surface, the radiance of indirect solar radiation reaching the surface, and the atmospheric transmittance from the surface to the remote sensor).

[0094] For each ray, the atmospheric attenuation during its upward journey from the Earth's surface to the sensor is the same, meaning the atmospheric transmittance is the same. Therefore, for a single ray, only the radiance reflected from the Earth's surface at zero line of sight needs to be calculated. The radiance at the sensor's entrance pupil is calculated by adding the statistical average of the zero line of sight reflected radiation from the Earth's surface calculated for each fiber to the atmospheric transmittance.

[0095] From the rendering equation, we know that to accurately solve the outgoing radiation brightness at a point P1 on the surface, we need to calculate all the incident radiation at that point. The incident radiation at this intersection point has three sources: direct solar radiation is the direct radiation received by the intersection point, which can be calculated by sampling a direct radiation sampling ray from the intersection point to the sun; solar scattered radiation and background radiation are indirect radiation received by the intersection point, because they come from the hemispherical space ξ defined by the normal direction of the intersection point. 2 All incident directions ω i, it is necessary to calculate the incident radiation integral in the hemispherical space.

[0096] 1. Calculation of surface reflected radiation from direct solar radiation (also known as the first surface reflected radiation parameter).

[0097] After obtaining the geometric information and material information of the intersection point P1 of the light and the 3D scene object, it is used as a sampling point. A direct radiation sampling light is emitted from the sampling point P1 toward the sun to sample and calculate the direct solar radiation. The specific method is: continue to track the intersection of the direct radiation sampling light and the 3D scene object. If there is no intersection with the scene, it is determined that the sampling point P1 receives direct solar radiation, such as Figure 3 shown.

[0098] Figure 3 The figure schematically shows a display diagram of a sampling point receiving direct solar radiation according to an embodiment of the present invention.

[0099] like Figure 3 As shown, at this time, the zero-view distance reflected radiance L of the sampling point P1 to direct solar radiation is d (P1,ω0) can be calculated as follows:

[0100]

[0101] Among them, E sd It indicates the irradiance perpendicular to the radiation direction when direct solar radiation reaches the earth's surface, which is calculated by atmospheric radiation transfer software. i,d Indicates the direction of direct solar radiation source, that is, the direction of the sun. d Indicates the angle between the sun direction and the normal direction of the sampling point P1. i,d , ω0 ) represents the bidirectional reflectance distribution function (BRDF) of the material at the sampling point P1.

[0102] If the direct radiation sampling light intersects with the three-dimensional scene object, it is determined that the sampling point P1 does not receive direct solar radiation, that is, the point is in the shadow. Figure 4 shown.

[0103] Figure 4 The diagram schematically shows a display in which a sampling point does not receive direct solar radiation according to an embodiment of the present invention.

[0104] like Figure 4 As shown in the figure, the zero-viewing distance reflected radiance of the sampling point P1 to direct solar radiation is 0, that is:

[0105]

[0106] Second, the calculation of surface reflected radiation from solar scattered radiation (also known as the second surface reflected radiation parameter).

[0107] From the sampling point P1 normal hemisphere space ξ 2 After emitting an indirect radiation sampling ray in a random direction, the intersection with the 3D scene object is traced. If there is no intersection with the 3D scene, it is determined that the indirect radiation received by the sampling point P1 comes from the sun's scattered radiation, such as Figure 5 shown.

[0108] Figure 5 The figure schematically shows a display diagram of solar scattered radiation received by a sampling point according to an embodiment of the present invention.

[0109] like Figure 5 As shown, at this time, the zero-viewing distance reflected radiance L of the sampling point P1 for indirect radiation id (P1,ω0) can be calculated as follows:

[0110]

[0111] Among them, L ss (P1,ω i,ss ) is the radiance of the sun’s scattered radiation reaching the Earth’s surface, which is calculated by atmospheric radiation transfer software. i,ss is the direction of the solar scattered radiation source. ss is the angle between the source direction of the solar scattered radiation and the normal direction of the sampling point P1. pdf P1 (ω i,ss ) is the sampling probability density function of the sampling point P1.

[0112] 3. Calculation of surface reflected radiation from background reflected radiation (also known as the third surface reflected radiation parameter).

[0113] If the indirect radiation sampling light intersects with the three-dimensional scene object, it is determined that the indirect radiation received by the sampling point P1 comes from the background reflected radiation, such as Figure 6 shown.

[0114] Figure 6 The figure schematically shows a display diagram of background reflected radiation received by a sampling point according to an embodiment of the present invention.

[0115] like Figure 6 As shown, at this time, the zero-viewing distance reflected radiance L of the sampling point P1 for indirect radiation id (P1,ω0) can be calculated as follows:

[0116]

[0117] Among them, ω i,br is the direction of the background reflected radiation source. br L is the angle between the background reflected radiation source and the normal direction of the sampling point P1. br (P1,ω i ) is the background reflected radiation radiance of the surrounding background to the sampling point P1. br (P1,ω i ) cannot be obtained directly, and it is necessary to use the idea of ​​recursion to take the new intersection point P2 of the indirect radiation sampling ray and the three-dimensional scene object as a new sampling point for radiation sampling calculation. The sampling of the new sampling point P2 also includes direct radiation sampling and indirect radiation sampling. If its indirect radiation sampling ray intersects with the three-dimensional scene object again, the new intersection point P2 is again used as a new sampling point for radiation sampling calculation. This recursion is repeated until the indirect radiation sampling ray of the sampling point does not intersect with the three-dimensional scene object, or the recursion is stopped after the number of recursions reaches a predetermined maximum threshold, and the radiation is calculated along the radiation transmission path obtained by the recursive tracing. It is generally believed that when the number of recursions reaches 4 times, the contribution of the background reflection radiation received by the last sampling point to the first sampling point P2 can be ignored. At this time, the background reflection radiation received by the last sampling point is considered to be 0.

[0118] After calculating the incident radiation of the lower intersection of the above three sources (that is, after calculating the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter), the total radiation data at the entrance pupil of the remote sensor is calculated based on the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter.

[0119] According to an embodiment of the present invention, calculating the total radiation data at the entrance pupil of the remote sensor based on the first surface reflected radiation parameter, the second surface reflected radiation parameter, and the third surface reflected radiation parameter may specifically include:

[0120] Calculating the sum of the first surface reflected radiation parameter and the second surface reflected radiation parameter, or calculating the sum of the first surface reflected radiation parameter and the third surface reflected radiation parameter, to obtain the surface zero-line-of-sight reflected radiance parameter of each ray;

[0121] Superimpose the zero-view-distance reflected radiance parameters of the surface of all light rays and calculate the average to obtain the zero-view-distance reflected radiance parameter of the surface of each pixel;

[0122] Based on the zero-viewing distance reflection radiance parameter of each pixel surface, the fusion radiation parameter and cross radiation are used to obtain the total radiation data at the entrance pupil of the remote sensor.

[0123] In this embodiment, for a single ray, the zero-viewing-distance reflected radiance L of the ground surface to direct radiation is d(P1,ω0) (the first surface reflected radiation parameter), which is related to the surface zero-line-of-sight reflected radiance L of the direct radiation id By adding (P1,ω0) (the second surface reflected radiation parameter or the third surface reflected radiation parameter), we can get the surface zero-view distance reflected radiance L zero (P1, ω0), that is:

[0124]

[0125] Among them, when the sampling light at the first intersection position of the light and the three-dimensional scene object intersects with the sky, the surface zero-view distance reflected radiance L of the surface to direct radiation id (P1,ω0) is the solar scattered radiation, which is also the second surface reflected radiation parameter. When the light intersects with other objects in the three-dimensional scene object, the surface zero-view distance reflected radiance L of the direct radiation id (P1,ω0) is the background reflected radiation, which is also the third surface reflected radiation parameter.

[0126] The statistical average of the zero-line-of-sight surface reflection radiance calculated for each ray can be used to obtain an unbiased estimate of the zero-line-of-sight surface reflection radiance of each pixel. In addition, during its transmission to the entrance pupil of the remote sensor, it will be affected by atmospheric attenuation, which is manifested as the upward transmittance τ(ω0) from the surface point P1 to the remote sensor entrance pupil point P0, where τ(ω0) can be calculated by atmospheric radiation transmission software.

[0127] The radiation received at the entrance pupil of the remote sensor includes not only the radiation reflected from the ground but also the path radiation L path (P1) and cross radiation L cross (ω0), the radiance of these two radiations can also be calculated by atmospheric radiation transfer software. Therefore, the total radiation L received at the entrance pupil of the remote sensor is ob (ω0) can be calculated by the following formula:

[0128]

[0129] Among them, L zero,j (P1,ω0) represents the zero-viewing-distance reflected radiance of the ground surface for the jth ray.

[0130] At this point, the hyperspectral imaging simulation model is constructed. By tracing back the path of step (1) establishing the coordinate system, step (2) projecting light from the remote sensor, step (3) calculating the intersection of the light and the scene, and step (4) calculating the radiation at the remote sensor entrance pupil, the hyperspectral imaging simulation model can be obtained.

[0131] After the hyperspectral imaging simulation model is constructed, in the application of using the hyperspectral imaging simulation model to generate hyperspectral remote sensing image data, it is only necessary to input the three-dimensional scene data determined according to user needs as the model input and input it into the hyperspectral imaging simulation model to generate hyperspectral remote sensing image data.

[0132] In this embodiment, the three-dimensional scene data determined according to user needs is used as the model input, and the detailed operations of inputting the hyperspectral imaging simulation model to generate hyperspectral remote sensing image data can refer to the construction process of the hyperspectral imaging simulation model, and the two correspond to each other.

[0133] For example, operation S2 is a detailed operation of emitting randomly distributed light rays from the origin to the three-dimensional scene objects through the virtual imaging plane based on the global coordinate system, which corresponds to the detailed operation of projecting light rays from the remote sensor in step (2) during the model construction process.

[0134] Operation S3 determines the details of the intersection of each ray and the three-dimensional scene object, which corresponds to the details of the calculation of the intersection of the ray and the scene in step (3) during the model construction process.

[0135] The detailed operation of operation S4, which calculates the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and the atmospheric radiation parameters, is consistent with the detailed operation of calculating the radiation data at the entrance pupil of the remote sensor in step (4) during the model construction process. Therefore, it will not be repeated here.

[0136] On the basis of the above embodiment, the hyperspectral imaging simulation model is implemented based on a graphics processor, wherein the rendering pipeline corresponding to the graphics application programming interface of the graphics processor may specifically include: a path tracing renderer and a radiosity renderer.

[0137] The path tracing renderer is used to calculate the light transmission path result based on the input three-dimensional scene data using the ray tracing model, and output the light transmission path result through the pixel shader and encode it into the texture buffer for storage.

[0138] The radiosity renderer and the path tracing renderer are set up in series to calculate the radiation data at the entrance pupil of the remote sensor based on the light transmission path results and the spectral radiation parameters.

[0139] In this embodiment, in order to accelerate the hyperspectral imaging simulation model, the rendering pipeline corresponding to the GPU graphics application programming interface is further improved. Specifically:

[0140] The Open Graphics Library (OGL) is a graphics application programming interface for rendering 2D and 3D vector graphics. Its core architecture is the OGL rendering pipeline, which defines the entire processing flow from input data to output image.

[0141] The rendering pipeline of the traditional graphics application programming interface is a rasterization-based graphics rendering architecture that efficiently converts three-dimensional geometry into two-dimensional pixel images through a series of fixed processing stages. However, when it is necessary to implement physically based ray tracing hyperspectral simulation, the original pipeline structure must be modified and expanded.

[0142] The present invention chooses to implement ray tracing by deeply modifying the two key shading stages, vertex shader and pixel shader, through the open graphics library shader language, and also adjusts the input and output methods of data in the pipeline. The preliminary improved graphics application programming interface ray tracing RGB rendering pipeline is as follows Figure 7 shown.

[0143] Figure 7 The flowchart of the preliminary improved graphics application programming interface ray tracing RGB rendering pipeline according to an embodiment of the present invention is schematically shown.

[0144] like Figure 7 As shown in the figure, after the initial transformation of the graphics application programming interface rendering pipeline, a core architecture of ray tracing based on the open graphics library was constructed. However, when applying it to hyperspectral remote sensing imaging simulation, there are still some challenges, such as: first, the transmission of large-scale scenes and hyperspectral data; second, the quantitative expression of radiation; third, the expansion from RGB simulation to hyperspectral simulation.

[0145] In this embodiment, specific solutions will be further provided for these three challenges.

[0146] First, regarding the transmission of large-scale scenes and hyperspectral data.

[0147] The Open Graphics Library primarily uses vertex array objects (VAOs) and vertex buffer objects (VBOs) to transfer 3D scene data. This data transmission method has significant shortcomings when performing ray tracing. For example, when VAOs and VBOs are used as input sources for the vertex shader, the pixel and pixel shaders at the end of the rendering pipeline cannot directly read their configuration information, effectively preventing them from obtaining scene and spectral data.

[0148] Taking these two points into consideration, this embodiment uses Texture Buffer Objects (TBOs) to transmit scene and hyperspectral data. TBOs encode the 3D scene's geometric data, hyperspectral material data, and hyperspectral radiation parameter data into a texture map as a one-dimensional array in a custom data format and bind it to a texture buffer. Texture samplers (texelFetch) can then directly retrieve the data from the texture buffer.

[0149] Second, regarding the quantitative expression of radiation.

[0150] The output channel of the Open Graphics Library (OGL) is a 4×2 byte array in the pixel shader—the fargColor array (RGBA, where the A channel is typically used to represent transparency). This results in compression loss in the output radiance data, making it difficult to meet the quantitative requirements of hyperspectral remote sensing simulation. To overcome this limitation, this embodiment uses a method that encodes a single floating-point value into four channels, thereby addressing the OGL's insufficient quantitative accuracy in hyperspectral remote sensing simulation.

[0151] Third, about the expansion from RGB simulation to hyperspectral simulation.

[0152] The above improvements result in the Open Graphics Library rendering pipeline only being able to calculate the radiance of a single band for each rendering calculation. To achieve the simulation of hundreds of spectral bands, a complete ray-scene intersection operation must be performed for each band, which greatly affects the efficiency of large-scale hyperspectral simulation. To overcome this limitation, considering that the ray paths of all bands in hyperspectral simulation are completely consistent and the computational complexity of radiance calculation is much lower than that of ray-scene intersection, this embodiment performs the following on the Open Graphics Library rendering pipeline: Figure 8 Improvements shown.

[0153] Figure 8 The flowchart of the secondary improved graphics application programming interface ray tracing hyperspectral rendering pipeline according to an embodiment of the present invention is schematically shown.

[0154] like Figure 8 As shown in Figure 1, the improved rendering pipeline splits the original ray tracing renderer into two serially executed renderers.

[0155] The first is the path tracing renderer, which only calculates the light transmission path, so its input only includes all geometry-related parameters. The path tracing renderer will output the calculation results (normal of each intersection point between the light and the scene, material number, direction of the incident light and direction of the outgoing light) through the fargColor array in the pixel shader and encode them into texture storage. The output result will be used for subsequent radiation calculations. Since the information of one intersection point between a light ray and the scene needs to store four data, a fargColor array can only output the information of one intersection point. The multiple render targets (MRT) technology of the open graphics library supports the output of shader operation results to up to 8 texture maps at the same time. This technology is implemented through framebuffer objects (FBO). By creating up to 8 fargColor arrays and binding them to 8 FBOs respectively, an output channel of up to 64 bytes can be obtained, which is sufficient to record all the intersection information in a radiation transmission path.

[0156] The second is the radiometric renderer, which performs only radiometric calculations, so its input consists solely of spectral-related parameters. The radiometric renderer simulates a single spectral band for each frame, taking the radiation path information calculated by the path tracing renderer as input. Combined with its own input spectral radiometric information, it calculates the spectral radiance at the remote sensor's entrance pupil, ultimately outputting and storing a hyperspectral remote sensing simulation image.

[0157] The hyperspectral remote sensing image generation method based on 3D scene radiometric simulation, provided by an embodiment of the present invention, utilizes a GPU to accelerate the hyperspectral imaging simulation model. Specifically, by improving the rendering pipeline corresponding to the graphics application programming interface (API) of the graphics processor, a path tracing renderer and a radiometric renderer are introduced to implement hyperspectral rendering. This solves the data input and output issues, quantification issues, and multi-spectral rendering issues of the original rendering pipeline. Because the improved rendering pipeline splits the original ray tracing renderer into two serially executed renderers, it significantly reduces the computational complexity of radiometric calculations and improves the efficiency of large-scale hyperspectral simulations.

[0158] Based on the above embodiments, in order to further reduce resource consumption and running time, the path tracing renderer is further accelerated.

[0159] Even after adapting the graphics application programming interface rendering pipeline, ray tracing algorithms still require significant computational resources. The long runtime and resource consumption hinder the practical application of hyperspectral imaging simulation models. To address this issue, this paper proposes an acceleration method while maintaining the required accuracy standards. While meeting the accuracy requirements, the following acceleration optimization operations are implemented.

[0160] According to an embodiment of the present invention, a path tracing renderer is configured to accelerate the processing of a ray tracing model through Monte Carlo random sampling, wherein the path tracing renderer includes: a hierarchical bounding box model, an importance sampling module, a low-discrepancy sequence sampling module, and a simulation accuracy and simulation efficiency balancing module.

[0161] The hierarchical bounding box model is used to accelerate the detection of intersection features between light and three-dimensional scene objects, wherein the hierarchical bounding box model is configured to search for intersection features between light and the bounding boxes of each node in sequence, starting from the root node, according to a depth-first search strategy.

[0162] The bounding volume hierarchy (BVH) is an efficient data structure based on the spatial partitioning of objects. It leverages the spatial clustering characteristics of geometric elements in a 3D scene. By constructing a hierarchical bounding volume structure, it forms a progressive screening mechanism from the global to the local, significantly reducing unnecessary geometric primitive intersection tests.

[0163] The typical BVH construction process uses a recursive algorithm. Starting from the root node containing all primitives, the algorithm then iterates through the following steps: calculating the bounding box of the current node, selecting the optimal split axis, determining the split position, and generating the left and right child nodes. This recursive process continues until a predefined termination condition is met (e.g., the number of primitives contained in the node reaches a lower limit or the maximum recursion depth is reached).

[0164] In this embodiment, when ray tracing is applied, the BVH traversal adopts a depth-first search strategy, that is, the algorithm starts from the root node and checks the intersection of the ray with the bounding box of each node in turn. For nodes that intersect with the ray, all triangles contained in it are further checked; while non-intersecting nodes and their child nodes are directly skipped. When the ray passes through multiple child nodes at the same time, the algorithm uses data structures such as priority queues to prioritize the node closest to the starting point of the ray. This "nearest first" processing strategy can quickly locate the nearest intersection, significantly improving the efficiency of the algorithm.

[0165] The importance sampling module is used to prioritize the directions or targets that have the greatest impact on the results during the Monte Carlo random sampling process.

[0166] In this embodiment, the Monte Carlo method is an algorithm that uses random sampling to estimate the probability of a random event. If the Monte Carlo method is used to calculate the function integral, let X1, X2...X n are independent samples and obey the same distribution. The probability density function is expressed as pdf(x). Then the integral of the function can be expressed as:

[0167]

[0168] The variance of this estimation method decreases as the number of samples increases, that is, it becomes more accurate. The importance sampling method is a method that increases accuracy and reduces variance while keeping the number of samples constant. Its principle is: let the standard deviation of the Monte Carlo estimator be σ, we can know that:

[0169]

[0170] To minimize the variance of the Monte Carlo integral, we need to set Minimum if and only if , that is Therefore, importance sampling is to perform more sampling in the area where the original function contributes more to the integral in order to obtain more accurate data. At this time, the sampling is not uniformly distributed, so each sampling value needs to be weighted. This weight is called the importance weight.

[0171] The original Monte Carlo ray tracing uses a sampling probability density function uniformly distributed over the hemisphere, namely:

[0172]

[0173] This sampling distribution does not satisfy Since there is a fixed cosine factor in the Monte Carlo integral formula of hemispherical radiation , so it is used as the importance weight and the sampling distribution is adjusted to the cosine distribution, that is:

[0174]

[0175] The sampling distribution at this time satisfies , which has the highest accuracy under the same number of samples.

[0176] The low-discrepancy sequence sampling module is used to generate a uniformly distributed point set in the sample space through mathematical construction to improve the convergence speed of Monte Carlo integration.

[0177] In Monte Carlo integration for ray tracing, a key factor affecting the convergence rate of the integration is the quality of the random number sequence. However, in some cases, random number sequences can exhibit uneven distribution, leading to slow convergence. Low-discrepancy sequences, as a random number generation scheme, use mathematical construction to produce a more evenly distributed set of points in the sample space, significantly improving the convergence rate of Monte Carlo integration.

[0178] In this embodiment, in actual ray tracing simulation calculations, low-discrepancy sequences are often used in combination with importance sampling. Importance sampling is achieved by adjusting the sampling density function This makes it similar in shape to the integrand, while the low-discrepancy sequence ensures that the sample points remain well distributed in the adjusted space.

[0179] The simulation accuracy and simulation efficiency balance module is used to determine the optimal number of rays based on simulation accuracy and simulation efficiency.

[0180] Although the increase in the number of rays determines the calculation accuracy, its marginal benefit will gradually decrease as the number of rays increases. The complexity T of ray tracing is linearly positively correlated with the number of rays n, as shown in the following formula:

[0181]

[0182] Therefore, this embodiment explores the relationship between the number of rays and computational accuracy from the perspective of balancing computational efficiency and computational accuracy, and attempts to find an optimal number of rays. The simulation results of 10,000 rays are used as the benchmark image, and the differences in simulation results under different numbers of rays are statistically analyzed. The comparison parameters include the mean and variance of the relative differences in pixel-by-pixel radiance. Ultimately, the model selects 2,000 rays as the optimal balance between efficiency and computational accuracy. Compared with 10,000 rays, this optimization achieves a 70.4% improvement in computational efficiency, with only an acceptable accuracy loss of 0.16%.

[0183] The hyperspectral remote sensing image generation method based on three-dimensional scene radiation simulation provided by the embodiment of the present invention adopts hierarchical bounding boxes to accelerate the intersection of light and scene, uses importance sampling and low-discrepancy sequences to accelerate the convergence of Monte Carlo integration, and explores the relationship between the number of light rays and simulation accuracy. The optimal number of light rays is selected for simulation experiments, thereby achieving a significant improvement in simulation efficiency at a very low cost in accuracy.

[0184] On the basis of the above embodiment, in order to better put the hyperspectral imaging simulation model into actual hyperspectral imaging simulation application, this embodiment is based on the improved open graphics library ray tracing rendering pipeline and the scene editing module of the BS (Browser / Serve) architecture as the front end, and further develops hyperspectral remote sensing imaging simulation software. The simulation process and main functional modules of the software are as follows Figure 9 shown.

[0185] Figure 9 The flowchart and functional module block diagram of the hyperspectral remote sensing imaging simulation software according to an embodiment of the present invention are schematically shown.

[0186] like Figure 9As shown, the hyperspectral remote sensing imaging simulation software process of the embodiment of the present invention may include:

[0187] (1) Open the software, start the simulation system, and enter the main interface of the software.

[0188] (2) Create a new scene. The user starts to build a new simulation task scene, which usually includes specified objects, terrain, buildings, sensors, etc.

[0189] (3) Scene editing: import or edit 3D models and material information: including background, target objects, etc.; finally output a complete scene combination package (geometry + material).

[0190] (4) Create a new calculation task. Create a new simulation calculation task for the scene. The task may include: simulation type (hyperspectral / multispectral), sensor parameters, solar altitude angle, band settings, etc.

[0191] (5) Parameter configuration: the user inputs or loads an existing parameter configuration file. The configuration items include: ray tracing settings (accuracy, sampling rate, number of rays, etc.); material reflectivity, BRDF model; band range, spectral resolution; environmental conditions (atmospheric model, sun angle); and loading of material reflectivity files (for hyperspectral reflectance calculations).

[0192] (6) Simulation calculation, start the ray tracing engine for rendering and hyperspectral inversion simulation; apply the acceleration module: hierarchical bounding box acceleration intersection; importance sampling optimizes the light path direction; low-discrepancy sequence improves convergence efficiency; precision control reduces the number of rays; output is a hyperspectral simulation data cube or image.

[0193] (7) Output simulation results, including hyperspectral images, band graphs, spectral curves, etc.; they can be used for subsequent imaging analysis, sensor design verification, or target recognition simulation.

[0194] An embodiment of the present invention further provides a hyperspectral remote sensing image generation device based on three-dimensional scene radiation simulation, which includes: a first determination module, an input module, a transmission module, a second determination module, a calculation module and a generation module.

[0195] The first determination module is used to determine the three-dimensional scene data according to user needs.

[0196] The input module is used to input 3D scene data into a pre-built hyperspectral imaging simulation model, wherein the hyperspectral imaging simulation model is configured with a global coordinate system and a 3D scene object for matching the 3D scene data, and executes the following modules:

[0197] The transmitting module is used to emit randomly distributed light from the origin to the three-dimensional scene object through the virtual imaging plane based on the global coordinate system. The virtual imaging plane is used to simulate the imaging surface of the remote sensor, and the three-dimensional scene object is used to simulate the optical response of the real ground structure after the light is incident.

[0198] The second determining module is used to determine the intersection point of each ray and the three-dimensional scene object.

[0199] The calculation module is used to calculate the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and the atmospheric radiation parameters.

[0200] The generating module is used to generate hyperspectral remote sensing image data according to the radiation data at the entrance pupil.

[0201] Figure 10 The structure block diagram of an electronic device suitable for implementing a method for generating a hyperspectral remote sensing image based on three-dimensional scene radiation simulation according to an embodiment of the present invention is schematically shown. Figure 10 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0202] like Figure 10 As shown, an electronic device 1000 according to an embodiment of the present invention includes a processor 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage unit 1008 into a random access memory (RAM) 1003. The processor 1001 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1001 may also include onboard memory for caching purposes. The processor 1001 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.

[0203] The storage unit 1008 stores various programs and data required for the operation of the electronic device 1000. The processor 1001, ROM 1002, and storage unit 1008 are interconnected via a bus 1004. The processor 1001 executes the programs stored in the ROM 1002 and / or storage unit 1008 to perform the various operations of the method flow according to the embodiment of the present invention. It should be noted that the programs may also be stored in one or more memories other than the ROM 1002 and storage unit 1008. The processor 1001 may also execute the programs stored in the one or more memories to perform the various operations of the method flow according to the embodiment of the present invention.

[0204] According to an embodiment of the present invention, electronic device 1000 may further include an input / output (I / O) interface 1005, which is also connected to bus 1004. Electronic device 1000 may also include one or more of the following components connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 1008 including a hard disk; and a communication section 1009 including a network interface card such as a LAN card or modem. Communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. Removable media 1011, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 1010 as needed, so that computer programs read from the removable media can be installed into storage section 1008 as needed.

[0205] 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 1009, and / or installed from the removable medium 1011. When the computer program is executed by the processor 1001, the above-mentioned functions defined in the system of the embodiment of the present invention are performed. According to an embodiment of the present invention, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0206] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0207] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium. Examples include, but are not limited to, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0208] For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 1002 and / or the storage portion 1008 described above and / or one or more memories other than the ROM 1002 and the storage portion 1008 .

[0209] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, 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, program segment, or portion of code, which 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 boxes may occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession may actually be executed substantially in parallel, or they may 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, as well as the combination of boxes in the block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or may be implemented using a combination of dedicated hardware and computer instructions. It will be understood by those skilled in the art that the features described in the various embodiments of the present invention may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention may be combined and / or coupled in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

[0210] The above describes embodiments of the present invention. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be advantageously used in combination. 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 method for generating hyperspectral remote sensing images based on three-dimensional scene radiation simulation, characterized in that: The method comprises: Determine 3D scene data based on user needs; Input the three-dimensional scene data into a pre-built hyperspectral imaging simulation model, wherein the hyperspectral imaging simulation model is configured with a global coordinate system and a three-dimensional scene object for matching the three-dimensional scene data, and perform the following operations: Based on the global coordinate system, starting from the origin, randomly distributed light rays are emitted through a virtual imaging plane toward the three-dimensional scene object, wherein the virtual imaging plane is used to simulate the imaging surface of the remote sensor, and the three-dimensional scene object is used to simulate the optical response of a real ground structure after being incident with light rays; Determine the intersection point of each ray with the three-dimensional scene object; Calculating the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and combining atmospheric radiation parameters; Hyperspectral remote sensing image data is generated based on the radiation data at the entrance pupil.

2. The method according to claim 1, characterized in that The hyperspectral imaging simulation model includes a plurality of triangular primitives, wherein determining the intersection point of each ray with the three-dimensional scene object includes: Determining whether the ray intersects the triangle element face; In response to the ray intersecting the triangle face, determining whether the intersection point is within the boundary of the triangle face; In response to the intersection point being within the boundary of the triangle element face, geometric information and material information of the intersection point are obtained.

3. The method according to claim 1, characterized in that The step of calculating the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and in combination with atmospheric radiation parameters includes: Calculate a first surface reflected radiation parameter from direct solar radiation at the intersection point; Calculate a second surface reflected radiation parameter from solar scattered radiation at the intersection point; Calculating a third surface reflected radiation parameter of the intersection point from background reflected radiation; Based on the first surface reflected radiation parameter, the second surface reflected radiation parameter and the third surface reflected radiation parameter, total radiation data at the entrance pupil of the remote sensor is calculated.

4. The method according to claim 3, characterized in that The calculating, based on the first surface reflected radiation parameter, the second surface reflected radiation parameter, and the third surface reflected radiation parameter, to obtain total radiation data at the entrance pupil of the remote sensor comprises: Calculating the sum of the first surface reflected radiation parameter and the second surface reflected radiation parameter, or calculating the sum of the first surface reflected radiation parameter and the third surface reflected radiation parameter, to obtain a surface zero-line-of-sight reflected radiance parameter of each light ray; Superimpose the zero-view-distance reflected radiance parameters of the surface of all light rays and calculate the average to obtain the zero-view-distance reflected radiance parameter of the surface of each pixel; Based on the zero-viewing distance reflection radiance parameter of each pixel surface, the fusion path radiation parameter and the cross radiation, the total radiation data at the entrance pupil of the remote sensor is obtained.

5. The method according to claim 1, wherein Pre-building the hyperspectral imaging simulation model includes: Establishing a local coordinate system with the optical center of the remote sensor as the origin, and transforming the position and orientation of the remote sensor from the local coordinate system to a global coordinate system based on specified remote sensor observation geometry information; Based on the global coordinate system, starting from the origin, emitting randomly distributed light rays through the virtual imaging plane toward the three-dimensional scene object; Determine the intersection point of each ray with the three-dimensional scene object; Calculating the radiation data at the entrance pupil of the remote sensor based on the geometric information and material information corresponding to the intersection point and combining atmospheric radiation parameters, wherein the radiation data at the entrance pupil represents the radiation brightness actually received by each pixel; The hyperspectral imaging simulation model is constructed based on the generation path of the radiation data at the entrance pupil.

6. The method according to claim 1, wherein The hyperspectral imaging simulation model is implemented based on a graphics processor, wherein the rendering pipeline corresponding to the graphics application programming interface of the graphics processor includes: A path tracing renderer is used to calculate the light transmission path result using the ray tracing model based on the input 3D scene data, and output the light transmission path result through the pixel shader and encode it into the texture buffer for storage; A radiosity renderer is provided in series with the path tracing renderer, and is used to calculate the radiation data at the entrance pupil of the remote sensor based on the light transmission path result and the spectral radiation parameters.

7. The method according to claim 6, characterized in that The path tracing renderer is configured to accelerate the ray tracing model by using Monte Carlo random sampling, wherein the path tracing renderer includes: A hierarchical bounding box model for accelerating the detection of intersection features between light and the three-dimensional scene objects, wherein the hierarchical bounding box model is configured to search for intersection features between light and the bounding boxes of each node in sequence, starting from the root node, according to a depth-first search strategy; Importance sampling module, used to prioritize the direction or target that has the greatest impact on the result during the Monte Carlo random sampling process; Low-discrepancy sequence sampling module, which is used to generate a uniformly distributed set of points in the sample space through mathematical construction to improve the convergence speed of Monte Carlo integration; The simulation accuracy and simulation efficiency balance module is used to determine the optimal number of rays based on simulation accuracy and simulation efficiency.

8. A hyperspectral remote sensing image generation device based on three-dimensional scene radiation simulation, characterized in that: The device comprises: A first determination module is used to determine three-dimensional scene data according to user needs; An input module is configured to input the three-dimensional scene data into a pre-built hyperspectral imaging simulation model, wherein the hyperspectral imaging simulation model is configured with a global coordinate system and a three-dimensional scene object for matching the three-dimensional scene data, and execute the following modules: a transmitting module, configured to transmit randomly distributed light rays from an origin through a virtual imaging plane toward the three-dimensional scene object based on the global coordinate system, wherein the virtual imaging plane is used to simulate an imaging plane of a remote sensor, and the three-dimensional scene object is used to simulate an optical response of a real ground structure after being incident with light rays; A second determining module, configured to determine an intersection point between each ray and the three-dimensional scene object; A calculation module, configured to calculate radiation data at the entrance pupil of the remote sensor based on geometric information and material information corresponding to the intersection point and atmospheric radiation parameters; A generation module is used to generate hyperspectral remote sensing image data based on the radiation data at the entrance pupil.

9. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having executable instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor executes the method according to any one of claims 1 to 7.

Images