Method for correcting moving dust on focal plane of super-large area array space camera

Abstract

The invention discloses a method for correcting moving dust on a focal plane of a super-large area array space camera, which comprises the following steps of: by taking irradiance distribution of an on-satellite calibration device of the super-large area array space camera as a reference, comparing an on-satellite calibration device image with a laboratory calibration device image to obtain an on-orbit variation of an area array space camera pixel; the method is applied to a laboratory relative correction coefficient to obtain an on-satellite relative correction coefficient. In addition, the invention also provides a method for detecting the dust position by utilizing a uniform image acquired by an on-satellite calibration device, so that the defects of the existing method in practical application are overcome. According to the on-orbit relative calibration method, the calibration precision is improved, the calibration cost can be reduced, and the precision of subsequent absolute radiation correction and inversion can be guaranteed.

Description

Technical Field

[0001] The invention relates to a correction method for dust movement on the focal plane of a super-large array space camera, and belongs to the technical field of optical remote sensing science. Background Art

[0002] The purpose of relative radiation correction of space optical cameras is to remove the sensor-level error caused by the difference in pixel response. It is an important way to improve the quality of satellite radiation. Its uncertainty can be transferred to absolute radiation correction, affecting the correction accuracy and inversion accuracy.

[0003] At present, the internationally used on-orbit space camera correction methods mainly include the dark current correction method, which is commonly used in foreign remote sensing satellite cameras, such as SPOT5, Landsat8, WorldView series and other satellite cameras; the relative gain correction method (mainly divided into on-board calibration device correction method, on-orbit uniform field correction method, and statistical correction method), which is widely used in remote sensing cameras at home and abroad.

[0004] The on-board calibration device generally uses a calibration lamp or a diffuse reflector as a uniform surface light source to perform on-orbit relative radiation calibration on the space camera. For example, the ALOS satellite uses an on-board calibration lamp for on-orbit relative radiation calibration, and the MODIS Aqua satellite uses an on-board diffuse reflector for on-orbit relative radiation calibration. However, these two methods are difficult to apply to the calibration of large-aperture and wide-field-of-view remote sensors.

[0005] The relative calibration method based on uniform field is also only applicable to remote sensors with narrow field of view. For large-array space cameras with wide field of view, it is impossible to find a uniform field on the ground that can fill the camera's field of view. The relative calibration based on statistical methods requires massive remote sensing satellite image data, which affects the timeliness of the relative calibration.

[0006] In a certain ultra-large array camera, due to the large detector area and the unsealed front lens barrel, technicians found that dust that entered during the camera assembly and launch process was adsorbed on the cover of the focal plane detector and moved and fell off, eventually resulting in black or white-gray dust shadows on the image. According to statistics, more than ten new or moving dust appeared on a 10K*10K focal plane detector. The dust size ranged from dozens to hundreds of pixels, appearing as gray-black or gray-white circular or flocculent patches. These patches first affected the relative radiation correction accuracy, resulting in a decrease in the local relative radiation correction accuracy of 0.1% to 0.5%. Secondly, these patches reduced the accuracy of ground object classification and target detection rate. Traditional correction methods mainly target problems such as inconsistency between pixels, inconsistency between slices, and inconsistency caused by slow changes in devices after entering orbit. They cannot solve the inconsistency caused by new and moving dust on the focal plane.

[0007] The laboratory calibration data currently mainly used on-orbit to correct on-orbit images will increase the non-uniformity of the image. In addition, this method needs to first calculate the irradiance distribution of the calibration component in the focal plane of the camera, and then use this distribution to calculate the on-board relative calibration coefficient. The calibration link is long and the final relative calibration accuracy will be affected. Therefore, the scope of application of this method is limited. Summary of the invention

[0008] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art, to provide a correction method for moving dust on the focal plane of an ultra-large array space camera, to solve the problem that the existing on-orbit relative correction method for space cameras cannot solve the correction of newly added or moving dust areas on the camera focal plane, and to obtain a stable relative correction coefficient that can be used for a long time.

[0009] The technical solution of the present invention is:

[0010] The present invention discloses a method for correcting dust on the focal plane of a super-large array space camera, comprising:

[0011] Using the on-board calibration device to obtain several frames of high-brightness uniform images and low-brightness uniform images of the large-array space camera under different imaging parameters;

[0012] Perform arithmetic averaging on several frames of high brightness uniform images and low brightness uniform images to obtain high brightness average image and low brightness average image respectively;

[0013] Perform edge extraction on the high brightness average image to obtain the dust area image;

[0014] According to the high brightness average image and the low brightness average image, the change rate of each pixel in the dust area image is calculated;

[0015] Calculate the correction coefficient of the dust area image according to the change rate;

[0016] The correction coefficient of the dust area image is used to obtain the corrected on-orbit image.

[0017] Furthermore, in the above method, the edge extraction is performed on the high brightness average image to obtain the dust area image, and the specific method is:

[0018] Calculate the gradient strength and direction of each pixel in the high-brightness average image;

[0019] According to the gradient intensity and direction of each pixel, the non-edge points in the high-brightness uniform image on track are eliminated to obtain the edge points;

[0020] Using double thresholds, the edge points are connected to obtain the dust area image.

[0021] Furthermore, in the above method, the step of calculating the gradient intensity and direction of each pixel in the high-brightness average image is as follows:

[0022] G(i,j) x =[MI orb_H (i+1,j-1)+2*MI orb_H (i+1,j)+MI orb_H (i+1,j+1)]

[0023] -[MI orb_H (i-1,j-1)+2*MI orb_H (i-1,j)]+MI orb_H (i-1,j+1)]

[0024] G(i,j) y =[MI orb_H (i-1,j-1)+2*MI orb_H (i,j-1)+MI orb_H (i+1,j-1)]

[0025] -[MI orb_H (i-1,j+1)+2*MI orb_H (i,j+1)+MI orb_H (i+1,j+1)]

[0026]

[0027] Among them, MI orb_H is the on-track highlight average image, G(x) is the lateral gradient intensity of each pixel in the on-track highlight average image, G(x) is the longitudinal gradient intensity of each pixel in the on-track highlight average image, G is the total gradient intensity of each pixel in the on-track highlight average image, and θ is the gradient direction of each pixel in the on-track highlight average image.

[0028] Furthermore, in the above method, according to the gradient strength and direction of each pixel, non-edge points in the high-brightness uniform image on track are eliminated to obtain edge points, specifically:

[0029]

[0030] Among them, I(i,j) orb_dust Mark the image for dust, with dust areas as 1 and non-dust areas as 0.

[0031] Furthermore, in the above method, the calculation of the change rate of each pixel in the dust area image is specifically:

[0032]

[0033] Among them, MIorb_H is the average value of high brightness image, MI orb_L is the low brightness average image, I lab_calib_H Calibrate the highlighted image for the laboratory, I lab_calib_L Calibrate low-light images for the laboratory, I lab_dark is the laboratory dark current image, K lab and B lab is the laboratory correction factor, R(i,j) orb_H is the highlight average image change rate, R(i,j) orb_L Low-brightness average image change rate, R(i,j) is the average change rate.

[0034] Furthermore, in the above method, the correction coefficient of the dust area image is calculated as follows:

[0035] K orb (i, j) = K lab (i, j)*R orb (i, j)

[0036] B orb (i, j) = B lab (i, j)*R orb (i, j)

[0037] Among them, K lab and B lab is the laboratory correction factor, R orb (i, j) is the average rate of change.

[0038] Furthermore, in the above method, the correction coefficient of the dust region image is used to obtain the corrected on-track image, specifically:

[0039] PI orb (i, j) = K orb (i, j)*PI orb_ori (i,j)+B orb (i, j)

[0040] Among them, PI orb (i, j) is the corrected on-orbit image, K orb (i, j) and B orb (i, j) is the correction coefficient of the dust area image, PI orb_ori For on-orbit images.

[0041] The beneficial effects of the present invention and the prior art are:

[0042] (1) The present invention adopts a method combining Candy edge detection technology and an on-board calibration device to determine the position of the dust moving on the camera focal plane, and uses the on-board calibration device to calculate the brightness change rate of the dust moving on the focal plane area. On this basis, the laboratory calibration coefficient of the area is corrected, which can achieve the technical effect of correcting the decrease in the accuracy of local relative radiation correction, and solve the problem that the current laboratory calibration data cannot adapt to the changes of the dust moving on the camera focal plane and increase the non-uniformity of the image.

[0043] (2) The present invention uses a multi-frame (10-frame) averaging method to adjust the noise of the dust edge, effectively improving the edge detection accuracy.

[0044] (3) The focal plane dust correction method for a large-array space camera based on an on-board calibration device of the present invention can effectively solve the problem of the correction result accuracy becoming poor due to the addition or movement of dust on-orbit for large-array space cameras, eliminate the image noise caused by the irrelevant pixels in the dust area, and improve the correction accuracy and image quality.

[0045] (4) The focal plane dust correction method of the present invention can effectively remove dust pollution on remote sensing satellite images of the earth, making up for the defects of existing methods in practical applications, and also providing a new idea for image correction of other types of space cameras.

[0046] (5) The present invention only corrects the laboratory calibration data of the dust area in the camera focal plane, and can improve the calibration and application accuracy through data processing methods without adding hardware equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The figure is a flow chart of the testing method of the present invention. DETAILED DESCRIPTION

[0048] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0049] The present invention discloses a method for correcting dust on the focal plane of a super-large array space camera, comprising:

[0050] Using the on-board calibration device to obtain several frames of high-brightness uniform images and low-brightness uniform images of the large-array space camera under different imaging parameters;

[0051] Perform arithmetic averaging on several frames of high brightness uniform images and low brightness uniform images to obtain high brightness average image and low brightness average image respectively;

[0052] Perform edge extraction on the high brightness average image to obtain the dust area image;

[0053] According to the high brightness average image and the low brightness average image, the change rate of each pixel in the dust area image is calculated;

[0054] Calculate the correction coefficient of the dust area image according to the change rate;

[0055] The correction coefficient of the dust area image is used to obtain the corrected on-orbit image.

[0056] Preferably, edge extraction is performed on the high brightness average image to obtain a dust region image, and the specific method is as follows:

[0057] Calculate the gradient strength and direction of each pixel in the high-brightness average image;

[0058] According to the gradient intensity and direction of each pixel, the non-edge points in the high-brightness uniform image on track are eliminated to obtain the edge points;

[0059] Using double thresholds, the edge points are connected to obtain the dust area image.

[0060] Preferably, the gradient strength and direction of each pixel in the high-brightness average image are calculated as follows:

[0061] G(i,j) x =[MI orb_H (i+1,j-1)+2*MI orb_H (i+1,j)+MI orb_H (i+1,j+1)]

[0062] -[MI orb_H (i-1,j-1)+2*MI orb_H (i-1,j)]+MI orb_H (i-1,j+1)]

[0063] G(i,j) y =[MI orb_H (i-1,j-1)+2*MI orb_H (i,j-1)+MI orb_H (i+1,j-1)]

[0064] -[MI orb_H (i-1,j+1)+2*MI orb_H (i,j+1)+MI orb_H (i+1,j+1)]

[0065]

[0066] Among them, MI orb_His the on-track highlight average image, G(x) is the lateral gradient intensity of each pixel in the on-track highlight average image, G(x) is the longitudinal gradient intensity of each pixel in the on-track highlight average image, G is the total gradient intensity of each pixel in the on-track highlight average image, and θ is the gradient direction of each pixel in the on-track highlight average image.

[0067] Preferably, according to the gradient strength and direction of each pixel, non-edge points in the high-brightness uniform image on track are removed to obtain edge points, specifically:

[0068]

[0069] Among them, I(i,j) orb_dust Mark the image for dust, with dust areas as 1 and non-dust areas as 0.

[0070] Preferably, the change rate of each pixel in the dust region image is calculated, specifically:

[0071]

[0072] Among them, MI orb_H is the average value of high brightness image, MI orb_L is the low brightness average image, I lab_calib_H Calibrate the highlighted image for the laboratory, I lab_calib_L Calibrate low-light images for the laboratory, I lab_dark is the laboratory dark current image, K lab and B lab is the laboratory correction factor, R(i,j) orb_H is the highlight average image change rate, R(i,j) orb_L Low-brightness average image change rate, R(i,j) is the average change rate.

[0073] Preferably, the correction coefficient of the dust area image is calculated as follows:

[0074] B orb (i, j) = K lab (i, j)*R orb (i, j)

[0075] B orb (i, j) = B lab (i, j)*R orb (i, j)

[0076] Among them, K lab and B lab is the laboratory correction factor, R orb (i, j) is the average rate of change.

[0077] Preferably, the correction coefficient of the dust region image is used to obtain the corrected on-track image, specifically:

[0078] PI orb (i, j) = K orb (i, j)*PI orb_ori (i,j)+B orb (i, j)

[0079] Among them, PI orb (i, j) is the corrected on-orbit image, K orb (i, j) and B orb (i, j) is the correction coefficient of the dust area image, PI orb_ori For on-orbit images.

[0080] Example

[0081] Relative radiation calibration is to remove the errors caused by the response differences of the sensor detectors themselves, so that different pixels have consistent output responses under the same radiation input. It is a key step to unify the radiation benchmarks of each sensor detector, and its accuracy will be transferred to the absolute radiation calibration, increasing the uncertainty of absolute radiation calibration.

[0082] There is inevitably non-uniformity between different pixels of the focal plane array detector, resulting in different pixels not having completely consistent output responses under the same radiation input. Relative radiation calibration is an important step to make its imaging benchmark return consistent. The accuracy of non-uniformity correction will affect the accuracy of absolute radiation calibration of subsequent images, bringing additional uncertainty to absolute radiation calibration. The reasons for the different responsivity and bias between pixels include: manufacturing process level, device non-uniformity, optical system processing and adjustment errors, etc. For ultra-large array space cameras, due to the large detector area and the unsealed front lens barrel, dust will enter during the camera adjustment and launch process and be adsorbed on the cover of the focal plane detector. During the launch and on-orbit period, the dust will fall off and shift, resulting in the relative calibration coefficient of the original dust position obtained in the laboratory no longer being applicable in this area. Finally, black or white-gray dust shadows appear on the image, seriously affecting the accuracy of relative radiation correction.

[0083] At present, the on-orbit non-uniformity correction of large-array space cameras mainly includes dark current correction method, laboratory integrating sphere correction method, on-board calibration device correction method, on-orbit uniform field correction method, and statistical correction method. The above correction methods have their own limitations in the on-orbit non-uniformity correction of large-array space cameras: the dark current correction method can only effectively correct the response non-uniformity when there is no light entering the pupil; the laboratory integrating sphere correction method has good stability in the visible light range, but cannot correct large changes in local areas after launching into orbit; the on-orbit uniform field correction method can only correct small-field-of-view space cameras, and it is difficult to find a uniform field of a single type of ground object for large-array and large-field-of-view cameras; the statistical correction method is only applicable to the non-uniformity correction of small-array or linear-array push-broom cameras. The above methods cannot effectively solve the problem of moving dust on the focal plane during the installation and launch process of ultra-large-array space cameras.

[0084] Therefore, this embodiment combines the laboratory integrating sphere correction data and the ultra-large array space camera moving focal plane dust correction method of the on-orbit calibration field, utilizes the on-board calibration component during on-orbit operation to produce a stable irradiance distribution in the camera focal plane, and establishes a change mapping in combination with the laboratory integrating sphere correction data to obtain the non-uniformity correction data of the local area.

[0085] like Figure 1 As shown, the method for correcting dust on the focal plane of a super-large array space camera provided in this embodiment includes the following steps:

[0086] (1) Turn off the on-board calibration light and obtain the on-orbit dark current image I orb_dark 1 frame,

[0087] (2) Turn on the calibration light on the satellite, adjust the parameters to 60% of the full well under a certain parameter, sample 10 frames, obtain a high-brightness uniform image and perform arithmetic averaging to obtain the image MI orb_H ;

[0088] (3) Adjust the parameters to 30% full well under certain parameters, sample 10 frames, obtain low brightness uniform images and perform arithmetic averaging to obtain image MI orb_L ;

[0089] (4) Laboratory calibration image I under certain known parameters lab_calib_H (average is 60% full well), I lab_calib_L (average is 30% of full well), dark current image I lab_dark , Laboratory correction factor K lab and B lab ;

[0090] (5) Use the Candy algorithm to perform image MI orb_H Perform edge extraction:

[0091] (5.1) Using the Sobel operator, we can obtain the gradients in the x-axis and y-axis directions and calculate the gradient intensity and direction of each pixel.

[0092] G(i,j) x =[MI orb_H (i+1,j-1)+2*MI orb_H (i+1,j)+MI orb_H (i+1,j+1)]

[0093] -[MI orb_H (i-1,j-1)+2*MI orb_H (i-1,j)]+MI orb_H (i-1,j+1)]

[0094] G(i,j) y =[MI orb_H (i-1,j-1)+2*MI orb_H (i,j-1)+MI orb_H (i+1,j-1)]

[0095] -[MI orb_H (i-1,j+1)+2*MI orb_H (i,j+1)+MI orb_H (i+1,j+1)]

[0096]

[0097] Among them, MI orb_H is the on-track highlight average image, G(x) is the lateral gradient intensity of each pixel in the on-track highlight average image, G(x) is the longitudinal gradient intensity of each pixel in the on-track highlight average image, G is the total gradient intensity of each pixel in the on-track highlight average image, and θ is the gradient direction of each pixel in the on-track highlight average image.

[0098] (5.2) Use double threshold detection to remove non-edge points and perform edge connection processing to obtain the dust area image I orb_dust , and I orb_dust The pixel value of the dust area is marked as 1, and the pixel values ​​of the rest are marked as 0;

[0099]

[0100] I(i,j) orb_dust Mark the image for dust, with dust areas as 1 and non-dust areas as 0.

[0101] (6) Calculate the change rate of the dust area point by point:

[0102]

[0103] R orb (i, j) = (R orb_H (i,j)+R orb_L (i, j)) / 2

[0104] Among them, MI orb_H is the highlight average image, MI orb_L is the low-brightness average image, I lab_calib_H Calibrate the highlighted image for the laboratory, I lab_calib_L Calibrate low-light images for the laboratory, I lab_dark is the laboratory dark current image, K lab and B lab is the laboratory correction factor, R(i,j) orb_H is the highlight average image change rate, R(i,j) orb_L Low-brightness average image change rate, R(i,j) is the average change rate.

[0105] (7) Calculate the correction coefficient of the dust area point by point:

[0106] K orb (i, j) = K lab (i, j)*R orb (i, j)

[0107] B orb (i, j) = B lab (i, j)*R orb (i, j)

[0108] Among them, K lab and B lab is the laboratory correction factor, R orb (i, j) is the average rate of change.

[0109] (8) Update the laboratory correction coefficient at the dust position and use the new correction coefficient to correct the on-orbit image PI orb_ori :

[0110] PI orb (i, j) = K orb (i, j)*PI orb_ori (i,j)+B orb (i, j)

[0111] Among them, PI orb (i, j) is the corrected on-orbit image, K orb (i, j) and B orb (i, j) is the correction coefficient of the dust area image, PI orb_ori For on-orbit images.

[0112] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

[0113] The contents not described in detail in the specification of the present invention belong to the common knowledge of the professionals in this field.

Claims

1. A method for correcting dust on the focal plane of a very large array space camera, characterized in that: include: Using the on-board calibration device to obtain several frames of high-brightness uniform images and low-brightness uniform images of the large-array space camera under different imaging parameters; Perform arithmetic averaging on several frames of high brightness uniform images and low brightness uniform images to obtain high brightness average image and low brightness average image respectively; Perform edge extraction on the high brightness average image to obtain the dust area image; According to the high brightness average image and the low brightness average image, the change rate of each pixel in the dust area image is calculated; Calculate the correction coefficient of the dust area image according to the change rate; The correction coefficient of the dust area image is used to obtain the corrected on-orbit image.

2. The method for correcting dust on the focal plane of a very large area array space camera according to claim 1, characterized in that: The edge extraction of the high brightness average image is performed to obtain the dust area image, and the specific method is as follows: Calculate the gradient strength and direction of each pixel in the high-brightness average image; According to the gradient intensity and direction of each pixel, the non-edge points in the high-brightness uniform image on track are eliminated to obtain the edge points; Using double thresholds, the edge points are connected to obtain the dust area image.

3. The method for correcting dust on the focal plane of a very large area array space camera according to claim 2, characterized in that: The calculation of the gradient intensity and direction of each pixel in the high brightness average image is specifically as follows: G(i,j) x =[MI orb_H (i+1,j-1)+2*MI orb_H (i+1,j)+MI orb_H (i+1,j+1)] -[MI orb_H (i-1,j-1)+2*MI orb_H (i-1,j)]+MI orb_H (i-1,j+1)] G(i,j) y =[MI orb_H (i-1,j-1)+2*MI orb_H (i,j-1)+MI orb_H (i+1,j-1)] -[MI orb_H (i-1,j+1)+2*MI orb_H (i,j+1)+MI orb_H (i+1,j+1)] Among them, MI orb_H is the on-track highlight average image, G(x) is the lateral gradient intensity of each pixel in the on-track highlight average image, G(x) is the longitudinal gradient intensity of each pixel in the on-track highlight average image, G is the total gradient intensity of each pixel in the on-track highlight average image, and θ is the gradient direction of each pixel in the on-track highlight average image.

4. The method for correcting dust on the focal plane of a very large area array space camera according to claim 2, characterized in that: According to the gradient strength and direction of each pixel, non-edge points in the high-brightness uniform image on track are eliminated to obtain edge points, specifically: Among them, I(i,j) orb_dust Mark the image for dust, with dust areas as 1 and non-dust areas as 0.

5. The method for correcting dust on the focal plane of a very large area array space camera according to claim 1, characterized in that: The calculation of the change rate of each pixel in the dust region image is specifically as follows: R orb (i,j)=(R orb_H (i,j)+R orb_L (i,j)) / 2 Among them, MI orb_H is the average value of high brightness image, MI orb_L is the low brightness average image, I lab_calib_H Calibrate the highlighted image for the laboratory, I lab_calib_L Calibrate low-light images for the laboratory, I lab_dark is the laboratory dark current image, K lab and B lab is the laboratory correction factor, R(i,j) orb_H is the highlight average image change rate, R(i,j) orb_L Low-brightness average image change rate, R(i,j) is the average change rate.

6. The method for correcting dust on the focal plane of a very large area array space camera according to claim 1, characterized in that: The calculation of the correction coefficient of the dust area image is specifically: K orb (i,j)=K lab (i,j)*R orb (i,j) B orb (i,j)=B lab (i,j)*R orb (i,j) Among them, K lab and B lab is the laboratory correction factor, R orb (i, j) is the average rate of change.

7. The method for correcting dust on the focal plane of a very large area array space camera according to claim 1, characterized in that: The correction coefficient of the dust region image is used to obtain the corrected on-track image, specifically: PI orb (i,j)=K orb (i,j)*PI obr_ori (i,j)+B orb (i,j) Among them, PI orb (i, j) is the corrected on-orbit image, K orb (i, j) and B orb (i, j) is the correction coefficient of the dust area image, PI orb_ori For on-orbit images.

Images