Target characteristic inversion method based on satellite data

Abstract

The invention discloses a target characteristic inversion method based on satellite data. The method comprises the steps of data acquisition and preprocessing; coordinate conversion and radiation transmission parameter preparation; forward calculation based on a full-link model; and carrying out target characteristic inversion solving. According to the invention, the influence of the cloud background, the atmospheric path radiation, the atmospheric attenuation, the detector response and the like on the optical characteristics of the target is comprehensively considered, and a full-link optical target characteristic inversion model of the imaging transmission medium-optical system-imaging detector is established. The method has important practical significance and application value in inversion of imaging characteristics of a target under different detection conditions (detection spectrum segments and detection backgrounds).

Description

TECHNICAL FIELD

[0001] The present application relates to a target characteristic inversion method based on satellite data, belonging to the technical field of remote sensing imaging and image processing. BACKGROUND

[0002] Satellite remote sensing technology has been widely used in many fields, such as resource exploration, environmental monitoring and other aspects, and plays a key role. Satellite remote sensing is profoundly changing the way people obtain earth information. Satellites can obtain massive target data by carrying various sensors, which contain rich characteristic information of the target. However, how to accurately invert the characteristics of the target from these complex data has become an important problem to be solved.

[0003] In the prior art, although some progress has been made in the research of target characteristic inversion, there are still many deficiencies. For example, some inversion methods are strongly dependent on specific environmental conditions, and when the environment changes, the inversion accuracy will decrease significantly. In complex terrain or climate change areas, traditional methods are difficult to effectively cope with the influence of atmospheric interference, terrain shielding and other factors on satellite data, resulting in large deviation of the inversion result. In addition, the existing methods often lack effective feature extraction means when dealing with multi-source and multi-modal satellite data, and cannot fully tap the potential information in the data, so that the inverted target characteristics are not comprehensive and accurate. At the same time, the calculation complexity of some inversion models is high, and the requirement for hardware equipment is harsh, which limits its wide application in practical scenarios. SUMMARY

[0004] The technical problem to be solved by the present application is to overcome the deficiencies of the prior art and provide a target characteristic inversion method based on satellite data, which realizes accurate inversion of target characteristics by establishing a full-link mathematical model from the target point to the detector output.

[0005] The technical solution of the present application is:

[0006] A target characteristic inversion method based on satellite data, comprising the following steps:

[0007] (1) collecting original image data of a target area by a high-resolution optical satellite;

[0008] (2) converting image coordinates and detector coordinates;

[0009] (3) determining the image DN value and the spectral radiance of the object point at wavelength λ;

[0010] (4) establishing a radiation transfer model from the object side to the image side to calculate the spectral response of the object point radiation characteristics on the detector;

[0011] (5) In the camera spectral range [λ1, λ2] spectrum integration is carried out to obtain the illumination of the object point on the detector;

[0012] (6) According to the obtained illumination on the detector, the detector light power is calculated in combination with the detector pixel area;

[0013] (7) The point spread function is introduced to perform spatial convolution on the detector light power to obtain the light power after adding the spatial response;

[0014] (8) According to the light power after adding the spatial response, the detector light response energy is obtained by integrating in the time dimension;

[0015] (9) According to the obtained detector light response energy, the detector theoretical DN value is obtained after photoelectric conversion and adding noise;

[0016] (10) The target characteristic inversion is solved according to the obtained detector theoretical DN value to realize the inversion of the target characteristic.

[0017] Further, the conversion of the image coordinates and the detector coordinates is specifically: the image coordinates (x i ,y i ) are converted into the detector coordinates (x d (s1,s2),y d (s1,s2)), and the conversion relationship is as follows:

[0018]

[0019] In the formula, T x and T y are the conversion functions of the image coordinates to the detector x and y direction coordinates, s1 and s2 are inherent parameters of the detector, and are the pixel spacing and array arrangement parameters in turn.

[0020] Further, the conversion function is determined by the following method:

[0021] T x and T y Use a polynomial correction model, assuming that the detector coordinates are (x d ,y d ) and the image coordinates are (x i ,y i ), then the x direction conversion function T x is expressed as:

[0022]

[0023] Similarly, the y direction conversion function T y is expressed as:

[0024]

[0025] where a ij , b ij are polynomial coefficients obtained by least square fitting of points with known ground accurate detector coordinates and corresponding image coordinates.

[0026] Further, the image DN value and the spectral radiance of the object point at the wavelength λ are determined, specifically:

[0027] The image DN value corresponding to the image coordinates (x i, y i ) is determined by the digital signal output by the detector, and the image DN value corresponding to the image coordinates (x i y i ) is as follows

[0028] DN(x i y i ) = DN(x d (s1, s2), y d (s1, s2))

[0029] Radiation transmission from the target to the detector, according to the position coordinates (x g (s1, s2, t), y g (s1, s2, t), z g (s1, s2, t)) and the target characteristics at the coordinates, the spectral radiance R g (x g (s1, s2, t), y g (s1, s2, t), z g (s1, s2, t), λ) of the object point at the wavelength λ is determined as the initial value of the radiation transmission model.

[0030] Further, the radiation transmission model from the object side to the image side is F(K a (λ), K o (λ), K d (λ), θ, R g ), wherein θ is the solar elevation angle on the ground, K a (λ) is the atmospheric radiation transmission, K o (λ) is the optical system transmittance, and K d (λ) is the detector response.

[0031] The spectral response E of the object point radiation characteristics on the detector (x d (s1, s2), y d (s1, s2)) is calculated as follows:

[0032]

[0033] Further, spectrum integration is performed to obtain the illumination of the object point on the detector The specific implementation is as follows:

[0034]

[0035] Wherein, λ1~λ2 are the spectral range of the camera;

[0036] According to the obtained illumination on the detector Integrating in the spatial dimension, the detector (x d (s1,s2),y d The light power P corresponding to the point pair (s1,s2),t) is obtained d (x d (s1,s2),y d (s1,s2),t), as follows:

[0037]

[0038] In the formula, A is the area of the detector pixel.

[0039] Further, according to the light power P d (x d (s1,s2),y d (s1,s2),t) corresponding to the detector pixel point t, convolution processing is performed, and the spatial response is added in the spatial domain to obtain the light power P blur (x d (s1,s2),y d (s1,s2),t) corresponding to each pixel of the detector after adding the spatial response, and the specific implementation is as follows:

[0040] P blur (x d (s1,s2),y d (s1,s2),t) = P d (x d (s1,s2),y d (s1,s2),t) * h

[0041] In the formula, h is the point spread function caused by atmospheric disturbance, optical diffraction, and aberration, that is, the spatial response.

[0042] Further, according to the obtained P blur (x d (s1,s2),y d (s1,s2),t), integration is performed in the time dimension to obtain the energy Q(x d (s1,s2),y d (s1,s2)) of the corresponding point of the detector, and the specific implementation is as follows:

[0043] For linear array and area array camera, the calculation is as follows:

[0044]

[0045] In the formula, T is integral time, t is exposure time;

[0046] For TDICCD camera, the calculation is as follows:

[0047]

[0048] In the formula, N is integral grade, T is single integral time, and t is exposure time.

[0049] Further, according to the obtained Q (x d (s1,s2),y d (s1,s2)) is photoelectrically converted, noise is added, AD conversion is carried out, and the digital image obtained by the detector (x d (s1,s2),y d (s1,s2)) from t to t+T is obtained:

[0050]

[0051] In the formula, S a is the saturation electron number of the detector, h is the Planck constant, c is the speed of light, is the average wavelength, is (λ1+λ2) / 2, Noise is the device noise, and q is the quantization bit number; for TDICCD camera, the above formula represents the digital image obtained from t to t+NT.

[0052] Further, the target characteristic inversion is solved according to the obtained detector theoretical DN value, the target characteristic inversion is realized, and specifically, the calculated detector theoretical DN value is compared with the image DN value, the initial spectral radiance of the target is adjusted through an iterative optimization algorithm

[0053] R g (x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ), until the error between the theoretical DN value and the actual DN value is less than a preset threshold, and the spectral radiance at this time is the inverted target spectral radiance characteristic.

[0054] The beneficial effects of the present application compared with the prior art are:

[0055] (1) The application constructs a target characteristic inversion method based on satellite data, realizes accurate inversion of target characteristics by establishing a full-link mathematical model from target points to detector output

[0056] (2) The method comprehensively considers the influences of image coordinate conversion, radiation transmission, spatial response, time integration, photoelectric conversion and other links, describes the physical processes and parameter correlations of each link through mathematical modeling, and finally obtains the original characteristics of the target from the digital image output of the detector. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a schematic diagram of the target characteristic inversion model; DETAILED DESCRIPTION

[0058] The specific embodiments of the application will be further described in detail below with reference to the accompanying drawings.

[0059] In the prior art, there are still many deficiencies in the research of target characteristic inversion. The application aims to construct a target characteristic inversion method based on satellite data, realize accurate inversion of target characteristics by establishing a full-link mathematical model from target points to detector output, as shown in Figure 1 The method comprehensively considers the influences of image coordinate conversion, radiation transmission, spatial response, time integration, photoelectric conversion and other links, describes the physical processes and parameter correlations of each link through mathematical modeling, and finally obtains the original characteristics of the target from the digital image output of the detector.

[0060] The following are specific mathematical models and implementation steps:

[0061] Step 1, collect the original image data of the target area by a high-resolution optical satellite;

[0062] Step 2, conversion model of image coordinates and detector coordinates

[0063] Convert the image coordinates (x i ,y i ) into detector coordinates (x d (s1,s2),y d (s1,s2)), and the conversion relationship is as follows:

[0064]

[0065] In the formula, T x and T yrespectively, are the conversion functions from image coordinates to detector x, y direction coordinates, and s1, s2 are intrinsic parameters of the detector (such as pixel pitch, array arrangement parameters, etc.). The conversion model ensures the accurate correspondence between image coordinates and detector physical positions by calibrating the geometric distortion of the satellite imaging system (such as lens distortion, satellite attitude error, etc.), and provides a spatial coordinate basis for subsequent radiation transfer calculation.

[0066] The conversion function is determined by the following method:

[0067] T x and T y Using a polynomial correction model, assuming that the detector coordinates are (x d ,y d ), and the image coordinates are (x i ,y i ), the x-direction conversion function T x is expressed as:

[0068]

[0069] Similarly, the y-direction conversion function T y is expressed as:

[0070]

[0071] where a ij and b ij are polynomial coefficients obtained by least squares fitting of points with known accurate detector coordinates and corresponding image coordinates on the ground.

[0072] Step 3, determining the image DN value and the spectral radiance of the object point at wavelength λ

[0073] The image DN value corresponding to the image coordinates (x i, y i ) is determined by the digital signal output by the detector, which is essentially the final manifestation of the object point radiance after multiple stages of conversion. To establish the correlation between DN value and object point radiance, the modeling needs to start from the source of radiation transfer - the spectral radiance of the object point. The image DN value corresponding to the image coordinates (x i ,y i ) is as follows

[0074] DN(x i ,y i ) = DN(x d (s1, s2), y d (s1, s2)) (2)

[0075] From the target to the detector, the radiation transfer is according to the position coordinates (x g (s1, s2, t), y g(s1,s2,t),z g Based on the target characteristics (s1,s2,t) and their coordinates, determine the spectral radiance R of the object point at wavelength λ. g (x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ), are used as the initial values ​​for the radiative transfer model.

[0076] The radiative transfer model is used to characterize the radiative transfer process from the target object to the detector. It comprehensively considers factors such as solar elevation angle, atmospheric radiative transfer, optical system transmittance, and detector response to construct the radiative transfer relationship from the object side (target point) to the image side (detector), obtains the response of the object point's radiative characteristics on the detector, and correlates the radiative characteristics of the target object with the signal received by the detector, providing a basis for subsequent inversion of target characteristics based on the detector output.

[0077] Step 4: Establish a radiation transfer model from the object side to the image side, and calculate the spectral response of the object point radiation characteristics on the detector.

[0078] Establish a radiative transfer model F(K) from the object side to the image side. a (λ),K o (λ),K d (λ),θ,R g The radiative transfer model comprehensively considers the ground solar altitude angle θ and atmospheric radiative transfer K. a (λ), optical system transmittance K o (λ), detector response K d The influence of (λ), etc.

[0079] The point radiation characteristics were obtained in the detector (x) d (s1,s2),y d The response E on (s1,s2) is as follows:

[0080]

[0081] Step 5: Perform spectral dimension integration: Calculate detector illuminance.

[0082] Integrating along the spectral dimension yields the illuminance of the object point on the detector. Specifically as follows:

[0083]

[0084] Where λ1~λ2 is the spectral range of the camera, and the atmospheric radiative transfer rate K a (λ) is calculated using MODTRAN software, representing the transmittance K of the optical system. o(λ) is given by the test results of the optical system, the spectral response K of the detector d (λ) is given by the test results of the detector.

[0085] Step 6, spatial dimension integration: detector light power calculation

[0086] According to the obtained illumination on the detector Integrating in the spatial dimension, the detector (x d (s1,s2),y d The light power corresponding to the point (s1,s2),t) is as follows:

[0087]

[0088] In the formula, A is the area of the detector pixel.

[0089] Step 7, spatial response convolution: adding spatial blur effect

[0090] According to the obtained light power P d (x d (s1,s2),y d (s1,s2),t) of the detector pixel point t, convolution processing is performed, spatial response is added in the spatial domain, and the light power P blur (x d (s1,s2),y d (s1,s2),t) corresponding to each pixel point of the detector after adding the spatial response is obtained, such as atmospheric disturbance, optical diffraction, aberration, etc.

[0091] P blur (x d (s1,s2),y d (s1,s2),t) = P d (x d (s1,s2),y d (s1,s2),t) * h (6)

[0092] In the formula, h is the point spread function caused by atmospheric disturbance, optical diffraction, aberration, etc., that is, the spatial response.

[0093] Step 8, time dimension integration: detector energy calculation

[0094] According to the obtained P blur (x d (s1,s2),y d (s1,s2),t), integrating in the time dimension, the energy Q(x d (s1,s2),y d (s1,s2)) of the corresponding point of the detector is obtained.

[0095] According to the dynamic imaging process of the detector, a dynamic integral model in time dimension is established to obtain the dynamic integral model of the detector (x d (s1,s2),y d The light response energy of the (s1, s2) point is shown in the following model for linear array and plane array cameras:

[0096]

[0097] In the formula, T is the integral time, and t is the exposure time.

[0098] For TDICCD cameras, the model is shown in the following formula

[0099]

[0100] In the formula, N is the integral grade, T is the single integral time, and t is the exposure time.

[0101] Step 9, photoelectric conversion and digital image output

[0102] According to the obtained Q(x d (s1,s2),y d (s1,s2), the photoelectric conversion is performed, the noise is added, the AD conversion is performed, and the digital image of the detector (x d (s1,s2),y d (s1,s2) point obtained from t to t+T (for TDICCD, t+NT) is obtained.

[0103]

[0104] In the formula, S a is the saturation electron number of the detector, h is the Planck constant, c is the light speed, is the average wavelength, is (λ1+λ2) / 2, Noise is the device noise, and q is the quantization bit number.

[0105] Solving:

[0106] R g (x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ),

[0107] The target characteristic inversion is realized.

[0108] Step 9, target characteristic inversion solving

[0109] According to the above model, the digital image DN value is related to the target original radiation characteristic R g(x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ) exist clear mathematical association. Based on the association, the inverse solution of the model can be obtained by inversion algorithm (such as least square method, iterative optimization algorithm, etc.), and the radiation characteristic parameters of the target are obtained by backstepping from the known DN value, so as to realize the inversion of the target characteristics.

[0110] The present application comprehensively considers the influence of cloud background, atmospheric path radiation, atmospheric attenuation, and detector response on the optical characteristics of the target, establishes a full-link optical target characteristic inversion model of imaging transmission medium-optical system-imaging detector, and has important practical significance and application value for the inversion of the imaging characteristics of the target under different detection conditions (detection spectral range, detection background).

[0111] Embodiment:

[0112] The present embodiment takes the characteristic inversion of the typical target (such as large-scale vehicle, ship, etc.) on the ground by the space-based high-resolution optical satellite as an example, and the implementation process of the present application is described in detail, and the specific steps are as follows:

[0113] (1) Data acquisition and pretreatment

[0114] The original image data of the target area is collected by the high-resolution optical satellite (such as the satellite carrying the area CCD or TDICCD sensor), and the key parameters during the imaging of the satellite are recorded, including the imaging time, the sun elevation angle, the satellite orbit parameters, the optical system parameters (such as focal length, transmittance test data), the detector parameters (such as pixel size, quantum efficiency, saturated electron number, quantization bit number) and the like.

[0115] The original image data is pretreated: the bad point noise in the image is removed, the image geometric distortion is corrected through the satellite attitude data and the orbit parameters, and the basis is provided for the subsequent coordinate conversion and radiation transmission calculation.

[0116] (2) Coordinate conversion and radiation transmission parameter preparation

[0117] The image coordinates (x i ,y i ) are converted into the detector coordinates (x d (s1,s2),y d (s1,s2)) through formula (1), wherein the conversion function T is determined according to the interior orientation elements (such as the principal point coordinates, the distortion coefficient) and the exterior orientation elements (such as the attitude angle, the position) of the satellite imaging system.

[0118] The atmospheric radiation transfer rate is calculated by inputting the imaging time, target region latitude and longitude, and atmospheric profile data (such as aerosol concentration and water vapor content) into the MODTRAN software; the optical system transmittance test data and the detector spectral response data are called as input parameters of the radiation transfer model.

[0119] Atmospheric radiation transfer rate K a (λ) is calculated by the MODTRAN software, the optical system transmittance K o (λ) is given by the test results of the optical system, and the spectral response of the detector K d (λ) is given according to the test results of the detector.

[0120] (3) Forward calculation based on the full link model

[0121] For each pixel in the image, the following calculations are performed:

[0122] a. Take the spectral radiance L(λ) of the target point as the initial value (the initial value can be assumed or set through prior data), substitute it into the radiation transfer model formula, and calculate the spectral response E(λ) on the detector;

[0123] b. Perform spectral integration in the camera spectral range [λ1, λ2] to obtain the illumination

[0124] c. Combine the detector pixel area A to calculate the optical power P;

[0125] d. Introduce the point spread function h(xd, yd) (determined according to the atmospheric disturbance intensity and optical system aberration parameters) to perform spatial convolution on P to obtain the optical power P' after adding the spatial response;

[0126] e. According to the type of detector carried by the satellite (such as formula (7) for area array cameras and formula (8) for TDICCD cameras), integrate in the time dimension to obtain the light response energy Q;

[0127] f. After photoelectric conversion and adding noise (device noise and AD conversion following Gaussian distribution), the theoretical DN value is obtained.

[0128] (4) Target characteristic inversion solution

[0129] Compare the theoretical DN value calculated in step (3) with the preprocessed actual image DN value, and adjust the initial spectral radiance L(λ) of the target through an iterative optimization algorithm (such as the least squares method) until the error between the theoretical DN value and the actual DN value is less than a preset threshold (such as the root mean square error RMSE < 5). At this time, L(λ) is the inverted target spectral radiation characteristic.

[0130] (5) Result verification

[0131] The ground measured data (such as typical target reflectivity measured by a ground spectrometer) in the target area is selected and compared with the inversion result, and if the error is within the allowable range (such as the relative error is less than 10%), it is indicated that the inversion is effective.

[0132] The part of the present application not described in detail is common knowledge to those skilled in the art.

Claims

1. A target characteristic inversion method based on satellite data, characterized in that, include: Raw image data of the target area is acquired using high-resolution optical satellites; Perform the conversion between image coordinates and detector coordinates; Determine the image DN value and the spectral radiance of the object point at wavelength λ; Establish a radiative transfer model from the object side to the image side, and calculate the spectral response of the object point's radiative characteristics on the detector; Spectral integration is performed within the camera spectral range [λ1,λ2] to obtain the illuminance of the object point on the detector; The detector optical power is calculated based on the illuminance obtained on the detector and the detector pixel area. By introducing a point spread function and spatially convolving the detector optical power, the optical power after adding the spatial response is obtained. Based on the optical power after adding the spatial response, the detector optical response energy is obtained by integrating over the time dimension. Based on the obtained detector photoresponse energy, the theoretical DN value of the detector is obtained through photoelectric conversion and noise addition; The target characteristics are inverted by obtaining the theoretical DN value of the detector.

2. The target characteristic inversion method based on satellite data according to claim 1, characterized in that: The conversion between image coordinates and detector coordinates specifically involves: converting the image coordinates (x, y, z) to detector coordinates. i ,y i Convert to detector coordinates (x) d (s1,s2),y d (s1,s2)), the transformation relationship is as follows: In the formula, T x and T y These are the transformation functions from image coordinates to detector x and y coordinates, respectively. s1 and s2 are the inherent parameters of the detector, namely the pixel spacing and array arrangement parameters, respectively.

3. The target characteristic inversion method based on satellite data according to claim 2, characterized in that: The transformation function is determined as follows: T x and T y Using a polynomial correction model, assuming the detector coordinates are (x... d ,y d The image coordinates are (x...). i ,y i If the x-direction transformation function T is given, then... x Represented as: Similarly, the y-direction transformation function T y Represented as: Where a ij b ij The polynomial coefficients are obtained by least-squares fitting of points with known accurate detector coordinates and corresponding image coordinates on the ground.

4. The target characteristic inversion method based on satellite data according to claim 2, characterized in that: The determination of the image DN value and the spectral radiance of the object point at wavelength λ is specifically as follows: Image coordinates (x) i, y i The corresponding image DN value is determined by the digital signal output by the detector, and the image coordinates (x) i ,y i The corresponding image DN values ​​are as follows: DN(x i ,y i )=DN(x d (s1,s2),y d (s1,s2)) Radiation transfer from the target to the detector, based on position coordinates (x g (s1,s2,t),y g (s1,s2,t),z g Based on the target characteristics (s1,s2,t) and their coordinates, determine the spectral radiance R of the object point at wavelength λ. g (x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ), are used as the initial values ​​for the radiative transfer model.

5. The target characteristic inversion method based on satellite data according to claim 4, characterized in that: The radiative transfer model from the object side to the image side is F(K a (λ),K o (λ),K d (λ),θ,R g ), where θ is the ground solar altitude angle, K a (λ) represents atmospheric radiative transfer, K o (λ) represents the transmittance of the optical system, K d (λ) represents the detector response; Calculate the point radiation characteristics in the detector (x) d (s1,s2),y d The spectral response E on (s1,s2) is as follows:

6. The target characteristic inversion method based on satellite data according to claim 5, characterized in that: Perform spectral integration to obtain the illuminance of the object point on the detector. Specifically as follows: Where λ1~λ2 is the spectral range of the camera; Based on the illuminance obtained from the detector Integrating in the spatial dimension yields the detector (x) d (s1,s2),y d The optical power P corresponding to point (s1,s2),t) is P d (x d (s1,s2),y d (s1,s2),t), as follows: In the formula, A is the area of ​​the detector pixel.

7. The target characteristic inversion method based on satellite data according to claim 6, characterized in that: Based on the optical power P corresponding to the detector pixel at time t d (x d (s1,s2),y d The array (s1, s2), t) is convolved to add a spatial response in the spatial domain, thus obtaining the optical power P corresponding to each pixel of the detector after adding the spatial response. blur (x d (s1,s2),y d (s1,s2),t), specifically: P blur (x d (s1,s2),y d (s1,s2),t)=P d (x d (s1,s2),y d (s1,s2),t)*h In the formula, h is the point spread function caused by atmospheric disturbance, optical diffraction, and aberrations, i.e., the spatial response.

8. The target characteristic inversion method based on satellite data according to claim 7, characterized in that: According to the obtained P blur (x d (s1,s2),y d Integrating over the time dimension (s1,s2),t), we obtain the energy Q(x) at the corresponding point of the detector. d (s1,s2),y d (s1,s2)), specifically: For linear and area scan cameras, the calculations are as follows: In the formula, T is the integration time and t is the exposure time; For TDICCD cameras, the calculation is as follows: In the formula, N is the integral series, T is the single-stage integration time, and t is the exposure time.

9. The target characteristic inversion method based on satellite data according to claim 8, characterized in that: Based on the obtained Q(x) d (s1,s2),y d (s1,s2) is used for photoelectric conversion, noise is added, and then AD conversion is performed to obtain the detector (x). d (s1,s2),y d The digital image obtained from point (s1,s2) from t to t+T: In the formula, S a where is the number of saturated electrons in the detector, h is Planck's constant, and c is the speed of light. The average wavelength is (λ1+λ2) / 2, noise is the device noise, and q is the number of quantization bits. For a TDICCD camera, the above formula represents the digital image obtained from t to t+NT.

10. The target characteristic inversion method based on satellite data according to claim 9, characterized in that: The process of inverting the target characteristics based on the obtained theoretical DN value of the detector to achieve the inversion of target characteristics is as follows: The calculated theoretical DN value of the detector is compared with the image DN value, and the initial spectral radiance R of the target is adjusted through an iterative optimization algorithm. g (x g (s1,s2,t),y g (s1,s2,t),z g (s1,s2,t),λ), until the error between the theoretical DN value and the actual DN value is less than the preset threshold, at which point the spectral radiance is the target spectral radiance characteristic obtained by inversion.

Images