Fine detection method of focal plane jitter of spaceborne optical camera

Through high-precision dense matching and tremor error modeling of surface array CMOS sequence images, the impact of satellite platform tremor on the imaging focal plane is solved, accurate measurement and compensation of periodic tremor errors are achieved, and the quality of remote sensing images is improved.

CN114170192BActive Publication Date: 2024-11-15WUHAN INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111502706.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-10
Publication Date
2024-11-15
Estimated Expiration
2041-12-10

AI Technical Summary

Technical Problem

High-resolution optical satellites cause imaging focal plane jitter due to tremor in satellite platforms, which affects image quality. It is difficult for the prior art to accurately measure and compensate for high-frequency tremor errors.

Method used

The surface array CMOS sequence image is used for high-precision and dense matching, generate a parallax map, analyze the relative tremor error curve, fit it through Fourier transform and sinusoidal function model, establish an absolute tremor error model, and accurately measure the periodic tremor error of the imaging focal plane .

Benefits of technology

It realizes accurate measurement of the focal surface tremor error of the star-borne optical camera, provides a reliable data basis for compensation of remote sensing image tremor errors, and improves the geometric quality of the image.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114170192B_ABST
    Figure CN114170192B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for fine detection of focal plane tremor of a satellite-borne optical camera, including: S1, high-precision dense matching: determining the image overlap area between adjacent frames of a planar array CMOS sequence image, and then performing pixel-by-pixel matching to obtain the same-name points of all pixel points in the overlap area, and generating a disparity map; S2, tremor relative error analysis: combining multiple single disparity map tremor relative error curves with time to form a tremor relative error curve of a sequence image, and filling the gaps; performing Fourier transform analysis to obtain the frequency and amplitude of the tremor relative error curve, and obtain the frequency, amplitude and initial phase of the tremor relative error; S3, tremor absolute error modeling: performing tremor absolute error modeling, and then combining the frequency, amplitude and initial phase of the tremor relative error to obtain the frequency, amplitude and initial phase of the tremor absolute error. The present invention realizes the precise measurement of the periodic tremor error of the imaging focal plane, and can provide a reliable data basis for the tremor error compensation of remote sensing images.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of satellite-borne optical camera imaging, and in particular relates to a method for finely detecting focal plane vibration of a satellite-borne optical camera. Background Art

[0002] Satellite platform tremor refers to a vibration phenomenon with a small amplitude and high frequency caused by the influence of internal and external factors on the satellite. High-resolution optical satellites are affected by platform tremor, which causes the focal plane of the imaging system carried on the satellite platform to shake at the same time, directly affecting the quality of satellite images. According to the imaging geometry principle, for a satellite with an orbital altitude of 500km, a platform tremor of 1 arc second will cause a distortion error of 2 to 3 meters in the push-scan image. With the increase of orbital altitude and spatial resolution, the impact of platform tremor will become more and more significant. In order to achieve high-precision processing and application of high-resolution optical satellites, platform tremor is a key issue that needs to be solved urgently.

[0003] At present, the research on the tremor processing of high-resolution optical satellite imaging platforms is mainly divided into two aspects: platform tremor detection and compensation based on attitude sensors and remote sensing images. Platform tremor measurement based on attitude sensors mainly uses sensors such as angular displacement, angular accelerometers, and angular vibration to obtain information such as high-frequency angular displacement, high-frequency angular acceleration, and high-frequency angular vibration of the platform, and fuses the information with the absolute attitude obtained by the star sensor to obtain high-precision and high-frequency attitude measurement results. For example, the reconstruction of high-resolution satellite tremor distortion images based on the linear accelerometer satellite platform tremor measurement method can effectively eliminate the distortion of high-resolution satellite image imaging of a certain model under platform tremor. The satellite platform tremor error detection and compensation based on remote sensing images mainly uses the optical satellite time-sharing imaging characteristics or reference data to obtain the platform tremor law, and compensates for the tremor error, while improving the geometric quality of remote sensing images.

[0004] In recent years, with the continuous improvement of the area array CMOS (Complementary Metal Oxide Semiconductor) process, area array CMOS image sensors have the advantages of high imaging quality, low cost, low power consumption, high speed, and strong radiation resistance. They have gradually occupied a certain position in the aerospace field and played an important role. In addition, area array CMOS also has flexible imaging modes, and can set multiple imaging modes such as global, rolling shutter, and open window. Among them, the rolling shutter adopts a line-by-line exposure method, which has a super strong platform stability perception ability and can become a new platform vibration detection method. Summary of the invention

[0005] The object of the present invention is to provide a method for finely detecting focal plane vibration of a satellite-borne optical camera, so as to achieve accurate measurement of periodic vibration error of the imaging focal plane.

[0006] To achieve the above object, the present invention provides a method for fine detection of focal plane jitter of a spaceborne optical camera, comprising the following steps:

[0007] S1. High-precision dense matching

[0008] Determine the image overlap area between adjacent frames of the area array CMOS sequence image, then perform pixel-by-pixel matching to obtain the same-name points of all pixels in the overlap area, and then generate a disparity map;

[0009] S2. Analysis of relative error of tremor

[0010] Based on a single disparity map, a single disparity map tremor relative error curve is generated;

[0011] Through time association, multiple single-view disparity image tremor relative error curves are combined to form a tremor relative error curve of the sequence image;

[0012] There are gaps between the tremor relative error curves of adjacent images, and the gaps are filled by fitting interpolation method;

[0013] Perform Fourier transform analysis on the tremor relative error curve of the filled sequence images to obtain the frequency and amplitude of the tremor relative error curve;

[0014] The obtained frequency and amplitude are used as initial values, and the sine function model is used to perform the overall least square fitting on the tremor relative error curve of the sequence image to obtain the frequency, amplitude and initial phase of the tremor relative error;

[0015] S3. Tremor Absolute Error Modeling

[0016] The absolute error of the tremor is modeled, and then the frequency, amplitude and initial phase of the relative error of the tremor are combined to obtain the frequency, amplitude and initial phase of the absolute error of the tremor.

[0017] In some optional implementations, determining the image overlap area between adjacent frames of the area array CMOS sequence image includes:

[0018] Using SIFT algorithm, at least one point of the same name is matched between two adjacent frames of the area array CMOS sequence image;

[0019] Count the coordinate differences of the same-name points between two adjacent frames of images to obtain the offset;

[0020] According to the offset, the initial position of each pixel of the previous sequence image on the subsequent sequence image is calculated; if the initial position of a pixel exceeds the area of ​​the subsequent sequence image, the pixel is not in the overlapping area.

[0021] In some optional implementations, correlation coefficient matching and least square matching algorithms are used to perform pixel-by-pixel matching.

[0022] In some optional implementations, generating a disparity map includes calculating the coordinate difference of all pixels with the same name in the overlapping area of ​​two adjacent frames of images in both row and column directions, using the coordinate difference as a DN value, and generating a coordinate difference map in both row and column directions.

[0023] In some optional implementations, generating a single disparity map tremor relative error curve based on a single disparity map includes:

[0024] The average value of each row of the single disparity image is calculated as the relative error of the row, and then the tremor relative error curve along the scanning line is obtained; the scanning line is converted into time to obtain the tremor relative error curve.

[0025] In some optional embodiments, the fitting interpolation method includes straight-line function interpolation fitting and spline function interpolation fitting. Further, when the absolute value of the slope of the samples at both ends of the gap is greater than 300, straight-line function interpolation fitting or spline function interpolation fitting is adopted; when the absolute value of the slope of the samples at both ends of the gap is between 100 and 300, straight-line function interpolation fitting is adopted; when the absolute value of the slope of the samples at both ends of the gap is less than 100, spline function interpolation fitting is adopted.

[0026] In some optional implementations, the sine function model is as follows:

[0027]

[0028] Among them, f k ,Ar k , are the frequency, amplitude and initial phase of the kth sinusoidal component of the tremor relative error respectively.

[0029] In some optional implementation schemes, the absolute tremor error modeling is performed, and then the frequency, amplitude and initial phase of the relative tremor error are combined to obtain the frequency, amplitude and initial phase of the absolute tremor error, including:

[0030] The frequency of the absolute tremor error is the same as the frequency of the relative tremor error;

[0031] When there is only one tremor frequency, the absolute tremor amplitude is calculated as follows:

[0032] Ad=Ar / (2sin(πfΔt))

[0033] Among them, Ad is the amplitude of the absolute tremor error, Ar is the amplitude of the relative tremor error, f is the tremor frequency, and Δt is the time interval between imaging the same-name points of adjacent images;

[0034] The initial phase calculation formula of the absolute error of the tremor is as follows:

[0035]

[0036] in, represents the initial phase of the absolute error of the tremor, It represents the initial phase of the relative error of the tremor, and mod(fΔt / 2,2) represents the remainder of fΔt / 2 divided by 2.

[0037] In some optional embodiments, the absolute error of the tremor is expressed as follows:

[0038]

[0039] Where D(t) is the function of the absolute error of the tremor changing with time.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] The present invention realizes accurate measurement of periodic chatter error of imaging focal plane through three main steps: dense matching of area array CMOS sequence images, relative chatter error analysis and absolute chatter error modeling, and can provide a reliable data basis for remote sensing image chatter error compensation. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The present invention is a flow chart of a method for finely detecting focal plane vibration of a spaceborne optical camera.

[0043] Figure 2 It is a single disparity map at 100 Hz of the present invention, a tremor relative error curve and an overall fitting map, and a Fourier analysis result map of the tremor error curve.

[0044] Figure 3 It is the tremor relative error curve of the serial disparity map at 10Hz of the present invention, the local straight line interpolation fitting map, the local spline function interpolation fitting map, the error maps of the two interpolation methods, the tremor curve Fourier analysis result map and the overall tremor fitting curve map. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0046] In view of the limitations of traditional tremor detection methods based on attitude sensors and remote sensing imaging platforms, the present invention uses a rolling shutter area array CMOS detector to perform precise detection of focal plane tremor of a satellite-borne optical camera. The rolling shutter area array CMOS sensor and the linear array CCD (Charge Coupled Device) are installed on the same imaging focal plane, and the tremor error at the imaging focal plane can be obtained using the RS effect. Through the three main steps of dense matching of rolling shutter area array CMOS sequence images, relative tremor error analysis, and absolute tremor error modeling, accurate measurement of periodic tremor errors in the imaging focal plane is achieved, which can provide a reliable data basis for remote sensing image tremor error compensation.

[0047] The present invention provides a method for finely detecting focal plane vibration of a spaceborne optical camera based on a rolling shutter area array CMOS, comprising the following steps:

[0048] Step 1: High-precision dense matching

[0049] According to the imaging time, the array CMOS sequence images are densely matched, including determining the image overlap between adjacent frames, determining the range that can be used for tremor detection, and then performing pixel-by-pixel matching to obtain the same-name points of all image points in the overlapping area and generate the corresponding disparity map.

[0050] Specifically, the tremor detection area is first determined, and the SIFT algorithm is used to match a small number of same-name points between two frames of images. By counting the coordinate differences of the same-name points, the initial position of each pixel point in the previous sequence image on the subsequent sequence image can be calculated. The correlation coefficient matching and least squares matching algorithms are used to perform pixel-by-pixel matching to obtain the same-name points of all image points, and the coordinate differences of the same-name points of all pixels in the overlapping area obtained by pixel-by-pixel matching in both row and column directions are calculated. The coordinate differences are used as DN values ​​to generate coordinate difference maps in two directions, which are called disparity maps.

[0051] Step 2: Relative tremor error analysis

[0052] 1) Generate a single disparity map tremor error curve. The tremor relative error curve along the scan line is obtained from the average value of each row of the single disparity map.

[0053] 2) Generate a tremor error curve of the sequence disparity map. When the tremor frequency is low, a single set of data is difficult to show the change trend of the tremor, so the sequence image analysis is performed. Multiple sets of single disparity map tremor error curves of the same frequency are combined by time to obtain the tremor relative error curve of the sequence image.

[0054] 3) Fitting interpolation. Due to the size of the matching window, there are "gaps" between the detection results of adjacent images. Therefore, before analyzing the tremor detection curve of the sequence image, the gaps must be filled, that is, interpolation. The present invention uses two different methods for interpolation, one is linear interpolation fitting, and the other is spline function interpolation fitting.

[0055] 4) Fourier transform analysis: Perform Fourier transform analysis on the local interpolated relative error curve to obtain the frequency and amplitude of the error curve.

[0056] 5) Overall least squares fitting. The frequency and amplitude obtained in the previous step are used as initial values, and the sine function model is used to perform overall least squares fitting on the tremor relative error curve to accurately estimate the frequency, amplitude and initial phase of the relative error.

[0057] Step 3: Absolute tremor error modeling

[0058] The absolute vibration error model is established, and the frequency, amplitude and initial phase of the absolute vibration error are obtained according to the corresponding formula. The amplitude of the absolute vibration is the same as the relative vibration error frequency. According to the amplitude and phase of the relative vibration error, the amplitude and phase of the absolute vibration can be calculated according to the formula.

[0059] A specific embodiment is given below:

[0060] Figure 1 The figure shows the process flow of fine detection of focal plane vibration of a spaceborne optical camera based on rolling shutter area array CMOS, which mainly includes the following steps:

[0061] Step 1: High-precision dense matching: Dense matching is performed on the array CMOS sequence images according to the imaging time.

[0062] Since the overlap of adjacent frames of the area array CMOS image sequence is closely related to the flight status and is not a fixed value, for each set of sequence effects, it is necessary to first determine the overlap of the images between adjacent frames, that is, first determine the area that can be used for tremor detection. The present invention uses the SIFT algorithm to match a small number of points of the same name between two frames of images. By counting the coordinate differences of the points of the same name, the offset can be calculated, that is, the initial position of each pixel of the previous sequence image on the subsequent sequence image can be calculated. Image points that exceed the coordinate area of ​​the subsequent sequence image indicate that they are not in the overlapping area.

[0063] Then, correlation coefficient matching and least squares matching algorithms are used to perform pixel-by-pixel matching. The coordinate differences of the same-name points of all pixels in the overlapping area obtained by pixel-by-pixel matching in both row and column directions are calculated. The coordinate differences are used as DN values ​​to generate coordinate difference maps in two directions, which are the so-called disparity maps. Figure 2 (a) is the disparity map at 100 Hz.

[0064] Step 2: Analysis of relative vibration error: Obtain accurate estimates of the frequency, amplitude and initial phase of the relative vibration error.

[0065] 1) Generate a single disparity image error curve. Calculate the average value of each row of the single disparity image as the relative error of the row, and then obtain the relative error curve of the tremor along the scan line. Convert the scan line to time to obtain a set of relative error curves of tremor, see Figure 2 The dark grey solid dotted curve in (b) Figure 2 The vibration frequency is 100HZ), where Figure 2 The black solid line curve in (b) is the overall fitting curve, and the light gray circle curve is the fitting error. Fourier transform analysis is performed on this group of error curves to obtain Figure 2 (c) Fourier analysis results.

[0066] If there is platform tremor, the time-varying error in the disparity map is the relative error between the two frames of images caused by the platform tremor. The relative error of platform tremor can be expressed by the formula:

[0067] g(t)=f(t+Δt)-f(t) (1)

[0068] Among them, g(t) represents the function of the relative error of the platform shaking changing with time, f(t) represents the function of the absolute error of the shaking changing with time, and Δt is the imaging time interval of the same ground object in two frames of images.

[0069] 2) Generate a sequence disparity map error curve. When the tremor frequency is low, a single set of data is difficult to show the change trend of the tremor, so a sequence image analysis is performed. Through time association, the tremor relative error curves of multiple single disparity maps are combined to form a sequence image tremor relative error curve. Due to the size of the matching window, there are "gaps" between the detection results of adjacent images, see Figure 3 (a)( Figure 3 The medium vibration frequency is 10HZ).

[0070] 3) Fitting interpolation. Due to the size of the matching window, "gaps" appear between the detection results of adjacent images. Therefore, before analyzing the tremor detection curve of the sequence image, the data of the missing segments must be filled, that is, interpolation. The present invention uses two different methods for local fitting interpolation, one is local straight line fitting interpolation, and the other is local spline function fitting interpolation. If the data of the two end points of the gap segment between adjacent images are known, the line segment function between the two points can be calculated through the two point data, and the function is generated to fill the gap segment. Then, the corresponding function value is calculated from the known independent variable data and inserted into the gap, which is the so-called straight line interpolation method. The spline function is composed of some polynomials, and each polynomial is determined by two adjacent points. Therefore, a spline function is constructed according to the existing data of the function, and then fitting interpolation is performed.

[0071] The linear interpolation function is:

[0072] y=ax+b (2)

[0073] The slope in formula (2) is a=(vx(k)-vx(k-1)) / (t(k)-t(k-1)), b=vx(k)-a*t(k), where vx(k), vx(k-1), t(k), t(k-1) are known function values ​​and independent variables.

[0074] The cubic spline interpolation function is:

[0075] S i (x) = a i +b i (xx i )+c i (xx i ) 2 +d i (xx i ) 3 ,i=0,1,...,n (3)

[0076] where a i ,b i ,c i ,d i Represents an unknown number. According to the continuity of the second-order derivative of S(x) on [a,b], at the node x i The continuity condition and interpolation condition are satisfied at , and the function S(x) can be calculated.

[0077] in Figure 3 (b)( Figure 3 Medium vibration frequency is 10HZ) and Figure 3 (c) are the effect diagrams of local straight line interpolation fitting and local spline function interpolation fitting. Figure 3 (b) and Figure 3The two small figures in (c) are partial enlarged images of two selected representative fitting results. The first partial enlarged image is the first gap (with a larger absolute slope value). Figure 3 (b) and Figure 3 (c), we can find that the straight line fitting effect is better, and the corresponding Figure 3 The first error in (d) is the larger error. The second partial enlargement is the third gap (the absolute value of the slope is small). Figure 3 (b) and Figure 3 (c), we can find that the two interpolation methods have similar effects and correspond to Figure 3 The third error in (d), that is, the error between the two is small.

[0078] After the method proposed in the present invention is experimented, a suitable interpolation method can be selected according to the slope of the samples at both ends of the gap. When the absolute value of the slope is greater than 300, the effects of the two interpolation methods are not much different, and both can perform local fitting of the disconnected curve well; when the absolute value of the slope is between 100-300, compared with the fitting effect of the spline function, the result of the linear function fitting is closer to the original tremor curve; when the absolute value of the slope is less than 100, the spline function has a better fitting effect.

[0079] 4) Fourier transform analysis. Perform Fourier transform analysis on the local fitted relative error curve to obtain the frequency and amplitude of the relative error curve, as shown in the attached figure. Figure 3 (e).

[0080] 5) Overall least squares fitting. The frequency and amplitude of the obtained vibration relative error curve are used as the initial values. Since the platform vibration can be considered as a harmonic motion of one or more sine functions in a short period of time, the sine function model is used to perform overall least squares fitting on the vibration relative error curve to accurately estimate the frequency, amplitude and initial phase of the relative error. The implementation method is as shown in formula (4):

[0081]

[0082] Among them, f k ,Ar k , are the frequency, amplitude and initial phase of the kth sinusoidal component of the relative error of the vibration. Usually, in a short period of time, the satellite only has a single main frequency platform vibration, so in this invention, only the case where k is 1 is discussed.

[0083] Step 3: Absolute tremor error modeling. Perform tremor absolute error modeling, and the amplitude and initial phase of the absolute tremor can be obtained according to the corresponding calculation formula.

[0084] According to the principle of motion synthesis, the frequency of the absolute tremor error should be the same as the relative tremor error frequency. Combining formulas (1) and (4), when there is only one tremor frequency, the amplitude of the absolute tremor can be calculated according to formula (5):

[0085] Ad=Ar / (2sin(πfΔt)) (5)

[0086] Among them, Ad is the amplitude of the absolute tremor error, Ar is the amplitude of the relative tremor error, f is the tremor frequency, and Δt is the time interval between the imaging of the same-name points in adjacent images.

[0087] The initial phase of the absolute error of the tremor is calculated by formula (6):

[0088]

[0089] Here, mod(fΔt / 2,2) represents the remainder when fΔt / 2 is divided by 2.

[0090] The absolute error of tremor can be expressed as formula (7):

[0091]

[0092] In summary, the present invention provides a tremor detection method for high-precision recovery of platform tremor frequency, amplitude and initial phase, in order to address the problem that there are slight differences in tremor parameters between the posture sensor installation point and the camera imaging focal plane, which makes it difficult to accurately compensate for the tremor error of ultra-high resolution images.

[0093] The present invention provides a method for fine detection of focal plane vibration of a satellite optical camera based on a rolling shutter area array CMOS, which includes three steps: high-precision dense matching, vibration relative error curve analysis, and vibration absolute error modeling. Among them, the high-precision dense matching includes first using the SIFT algorithm to match the same-name points between two frames of images, and then using the correlation coefficient matching and the least squares matching algorithm to perform pixel-by-pixel matching to obtain the same-name points of all image points in the overlapping area, and obtain the disparity map of the image. The relative error analysis of the vibration includes the error curve analysis of the single disparity map and the error curve analysis of the disparity map of the sequence image, and before analyzing the vibration error curve of the sequence image, the gaps of the sequence image curve need to be filled, and two interpolation methods, linear interpolation fitting and spline function interpolation fitting, are used to perform local fitting of the curve. Then, the vibration error curve after local fitting is subjected to Fourier transform analysis to obtain the frequency and amplitude of the vibration relative error curve. The obtained frequency and amplitude are used as initial values, and the error curve is subjected to overall least squares fitting using a sine function model to obtain the frequency, amplitude, and initial phase of the relative error of the vibration. The absolute error modeling of the tremor means that the amplitude and phase of the absolute tremor can be calculated according to the formula based on the amplitude and phase of the relative error of the tremor.

[0094] At the same time, the present invention carries out simulation experiments based on real remote sensing data to achieve accurate measurement of periodic vibration errors in the imaging focal plane, which can directly provide a reliable data basis for remote sensing image vibration error compensation.

[0095] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0096] It will be easily understood by those skilled in the art that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for fine detection of focal plane jitter of a spaceborne optical camera, characterized in that: The following steps are involved: S1. High-precision dense matching Determine the image overlap area between adjacent frames of the area array CMOS sequence image, then perform pixel-by-pixel matching to obtain the same-name points of all pixels in the overlap area, and then generate a disparity map; S2. Analysis of relative error of tremor Based on a single disparity map, a single disparity map tremor relative error curve is generated; Through time association, multiple single-view disparity image tremor relative error curves are combined to form a tremor relative error curve of the sequence image; There are gaps between the tremor relative error curves of adjacent images, and the gaps are filled by fitting interpolation method; Perform Fourier transform analysis on the tremor relative error curve of the filled sequence images to obtain the frequency and amplitude of the tremor relative error curve; The obtained frequency and amplitude are used as initial values, and the sine function model is used to perform the overall least square fitting on the tremor relative error curve of the sequence image to obtain the frequency, amplitude and initial phase of the tremor relative error; S3. Tremor Absolute Error Modeling The absolute error of the tremor is modeled, and then the frequency, amplitude and initial phase of the relative error of the tremor are combined to obtain the frequency, amplitude and initial phase of the absolute error of the tremor, including: The frequency of the absolute tremor error is the same as the frequency of the relative tremor error; When there is only one tremor frequency, the absolute tremor amplitude is calculated as follows: Ad=Ar / (2sin(πfΔt)) Among them, Ad is the amplitude of the absolute tremor error, Ar is the amplitude of the relative tremor error, f is the tremor frequency, and Δt is the time interval between imaging the same-name points of adjacent images; The initial phase calculation formula of the absolute error of the tremor is as follows: in, represents the initial phase of the absolute error of the tremor, represents the initial phase of the relative error of the tremor, mod(fΔt / 2,2) represents the remainder of fΔt / 2 divided by 2; The absolute error of tremor is expressed as follows: Where D (t) A function representing the absolute error of tremor as a function of time.

2. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: Determining the image overlap area between adjacent frames of the area array CMOS sequence image includes: Using SIFT algorithm, at least one point of the same name is matched between two adjacent frames of the area array CMOS sequence image; Count the coordinate differences of the same-name points between two adjacent frames of images to obtain the offset; According to the offset, the initial position of each pixel of the previous sequence image on the subsequent sequence image is calculated; if the initial position of a pixel exceeds the area of ​​the subsequent sequence image, the pixel is not in the overlapping area.

3. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: Correlation coefficient matching and least squares matching algorithms are used for pixel-by-pixel matching.

4. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: Generating a disparity map includes: calculating the coordinate difference of all pixels with the same name in the overlapping area of ​​two adjacent frames of images in both row and column directions, using the coordinate difference as a DN value, and generating a coordinate difference map in both row and column directions.

5. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: Based on a single disparity map, generating a single disparity map tremor relative error curve includes: The average value of each row of the single disparity image is calculated as the relative error of the row, and then the tremor relative error curve along the scanning line is obtained; the scanning line is converted into time to obtain the tremor relative error curve.

6. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: The fitting interpolation methods include straight line function interpolation fitting and spline function interpolation fitting.

7. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 6, characterized in that: When the absolute value of the slope of the samples at both ends of the gap is greater than 300, straight line function interpolation fitting or spline function interpolation fitting is adopted; when the absolute value of the slope of the samples at both ends of the gap is between 100 and 300, straight line function interpolation fitting is adopted; when the absolute value of the slope of the samples at both ends of the gap is less than 100, spline function interpolation fitting is adopted.

8. The method for fine detection of focal plane jitter of a spaceborne optical camera according to claim 1, characterized in that: The sine function model is as follows: Among them, f k ,Ar k , are the frequency, amplitude and initial phase of the kth sinusoidal component of the tremor relative error respectively.