A method for correcting radiation error noise of a large-array space infrared camera system

Abstract

The present application relates to a kind of big face array space infrared camera system radiation error noise correction method, belong to optical remote sensing science and technology field.The present application is with big face array space infrared camera laboratory calibration image, including inner blackbody calibration image and outer blackbody calibration image as reference to calculate correction function, using radial gradient statistical method to judge whether there is system inherent inner radiation circular noise, and obtain the relevant statistics of big face array space camera system inner radiation noise, it is applied to the circular inherent background noise formed by the heat radiation of camera internal element (such as lens, shell, circuit), the fixed circular noise in image is compensated.The infrared camera system inherent system inner radiation noise correction method of the present application improves calibration accuracy, reduces calibration cost, and can guarantee the accuracy of subsequent absolute radiation calibration and inversion.

Description

Technical Field

[0001] This invention belongs to the field of optical remote sensing science and technology, and relates to a method for correcting radiation error noise in a large-area array space infrared camera system. Background Technology

[0002] Large-area space infrared camera (its pixel count is generally ≥2K) (2K) Inherent system radiometric error is the radiometric signal generated by its own components rather than the target scene. It is superimposed on the target signal, leading to a decrease in image signal-to-noise ratio and radiometric accuracy, and in severe cases, it can even directly affect target detection. Theoretically, this inherent system radiometric error cannot be completely eliminated, but can only be reduced through technical means.

[0003] After linear or locally linear correction (single-point correction, two-point correction, multi-point correction) of images from a large-area spatial infrared camera based on an internal blackbody, limitations in the optical system design mean that light rays interact with lens elements and barrel components as they propagate within the lens. Some light is scattered, and this internal radiation can reduce image sharpness and may even lead to inherent residual errors that cannot be eliminated, resulting in a decrease in relative correction accuracy. From the perspective of the corrected blackbody image, residual errors manifest as fixed-pattern noise with inherent shapes, such as circles. These residual errors cannot be resolved by reference-based non-uniform correction, affecting absolute accuracy and application performance.

[0004] Reference-based blackbody radiation source calibration methods involve acquiring uniform blackbody images at different temperatures, leading to single-point calibration of uniform images at a single temperature point, two-point calibration of uniform images at two temperature points, and multi-point calibration of uniform images at multiple temperature points. These methods assume that the response of the infrared focal plane is linear or locally linear, or can be fitted using a curve. However, the aforementioned reference-based calibration techniques are largely ineffective against fixed-pattern noise caused by internal radiation.

[0005] Scene-based large-area non-uniformity correction methods include statistical model-based methods and constant statistics-based methods. However, statistical model-based methods, while filtering out fixed-pattern noise, also remove static features in the scene, leading to severe ghosting. Furthermore, determining the cutoff frequencies of the spatial and temporal filters is difficult. Constant statistics-based methods require high diversity in scene radiation received by each detector pixel, making them highly dependent on sufficient scene motion. Additionally, these methods require accumulating a large number of image frames to ensure grayscale values ​​at each pixel location, which is nearly impossible for current large-area infrared cameras (2K*2K and above). In summary, statistical model-based non-uniformity correction methods primarily rely on the degree of conformity of the infrared image sequence to statistical assumptions, but in practice, it is difficult to perfectly conform to these assumptions. Summary of the Invention

[0006] The technical problem solved by this invention is to overcome the shortcomings of the prior art and provide a method for correcting radiation error noise in a large-area array space infrared camera system.

[0007] The solution of this invention is: a method for correcting radiation error noise in a large-area array space infrared camera system, comprising: (1) Using internal blackbody imaging, the large-area array space infrared camera was used to obtain images under different imaging parameters and brightness conditions. L Intraframe blackbody uniform image ，L ≥10; (2) For different imaging parameters and different brightness levels L The arithmetic mean of the uniform blackbody images within each frame is then performed to obtain the mean blackbody image. N t ( i,j ) and the spatial mean of the inner blackbody average image , t This represents different brightness levels, with values ​​for level 1, level 2, level 3, level 4, and level 5. i, j These represent the horizontal and vertical coordinates of the image pixels, respectively. (3) Using external blackbody imaging, according to the imaging parameters and brightness settings described in (1), obtain uniform external blackbody images of each U frame of the large-area array space infrared camera, U≥100; (4) Calculate the quadratic calibration coefficients of the inner blackbody average image in (2) using the least squares method, and use the quadratic calibration coefficients to correct the outer blackbody uniform image in (3) to obtain corrected images with different brightness under different parameters. (5) Perform radial gradient statistics on the corrected images with different brightness under different parameters to determine whether there is inherent internal radiation circular noise in the corrected images; if there is, proceed to step (6), otherwise end the correction. (6) The blackbody images of each U-frame with different brightness under each parameter are calibrated, the compensation function is statistically analyzed, and the compensation function is used to correct the calibrated blackbody correction image to obtain the final correction image.

[0008] Furthermore, the method of correcting the uniform image of the outer blackbody in (3) using quadratic scaling coefficients includes: firstly, calculating the quadratic scaling coefficients for five sets of known data points. , , , , Perform least squares fitting, assuming the fitting function is The quadratic scaling coefficients are obtained. , , ); in,( i,j () represents the horizontal and vertical components of a pixel. N 1 ( i,j ), N 2 ( i,j ), N 3 ( i,j ), N 4 ( i,j ), N 5 ( i,j The numbers ) represent the average blackbody images within the first, second, third, fourth, and fifth brightness levels, respectively. i,j The value at the position; the first brightness level is 0.1 full well, the fifth brightness level is 0.9 full well, and the second, third, and fourth brightness levels take different values ​​within the range of 0.1 to 0.9 full well; , , , , These represent the spatial mean values ​​of the blackbody average image within the first, second, third, fourth, and fifth brightness levels, respectively. The uniform image of the external blackbody is corrected using quadratic scaling coefficients to obtain corrected images with different brightness levels under different parameters. ; in, Uniform images of an outer blackbody under different imaging parameters and brightness.

[0009] Furthermore, step (5) involves performing radial gradient statistics on the corrected images with different brightness levels under different parameters to determine whether the corrected images contain inherent internal radiation circular noise, including: Determine the center point of the corrected image; Calculate the total radial gradient intensity and radial gradient direction for each pixel in the corrected image; Remove non-edge points from the corrected image to obtain edge points; Calculate the radial gradient exponent and determine whether circular noise exists in the corrected image.

[0010] Furthermore, the calculation of the total radial gradient intensity and radial gradient direction of each pixel in the corrected image specifically involves: assuming the center of the corrected image is... , x, y To correct the number of pixels in the image in both the horizontal and vertical directions, the correction is performed on any point in the image. The Sobel operator is used to perform convolution calculations on the calibrated image to obtain the total radial gradient intensity and radial gradient direction for each pixel: ; ; ; ; in, To correct the lateral radial gradient intensity of each pixel in the image, To correct the longitudinal and radial gradient intensity of each pixel in the image, To correct the total radial gradient intensity of each pixel in the image, To correct the radial gradient direction of each pixel in the image.

[0011] Furthermore, based on the total radial gradient intensity and radial gradient direction of each pixel in the corrected image, non-edge points in the image are removed to obtain edge points, specifically: in, These are radial gradient edge points, with a value of 1 for edge points and 0 for non-edge points.

[0012] Furthermore, based on the total radial gradient intensity of each pixel in the corrected image and the radial gradient direction, the radial gradient exponent is calculated. RGI Specifically: in, M It is the set of radial gradient edge points. r p ( i,j (with ) as the center point To the edge point The radial vector, for The total radial gradient intensity of the cell when =1; the threshold is set to 0.9, if RGI If the value is greater than 0.9, then the calibrated image under this parameter is considered to still have inherent circular noise. RGI If the value is ≤0.9, it is considered that the calibrated image under this parameter does not have inherent circular noise.

[0013] Furthermore, the statistical compensation function, and the use of the compensation function to correct the calibrated outer blackbody correction image, includes: Calculate the spatiotemporal mean and standard deviation of the U-frame out-of-frame blackbody corrected images under different brightness levels, and statistically analyze their deviation. The deviation at each point under different brightness levels is fitted using a polynomial function to obtain the deviation compensation function; The calibrated image is corrected using a deviation compensation function to obtain a corrected image that can effectively correct the inherent annular radiation error.

[0014] Furthermore, the calculation of the spatiotemporal mean and standard deviation of the U-frame out-of-frame blackbody corrected image under different brightness levels, and the statistical analysis of its deviation, includes: Calculate the blackbody-corrected image outside the U-frame under different brightness levels. Arithmetic mean: Calculate the spatiotemporal mean of the U-frame out-of-frame blackbody corrected image under different brightness levels. Standard deviation : ; ; Statistical analysis of the spatiotemporal mean of U-frame out-of-frame blackbody corrected images under different brightness levels Standard deviation Deviation: t Represents different brightness levels ，s Represents the number of frames. This represents the spatiotemporal mean of the external blackbody corrected image under different brightness levels. Standard deviation The degree of deviation; x, y To correct the number of pixels in the image in both the horizontal and vertical directions; To correct any point in the image, Represents the horizontal and vertical components of any pixel in the corrected image; This represents the arithmetic mean of the blackbody-corrected image outside the U-frame.

[0015] Furthermore, the deviation compensation function is obtained by performing a least-squares fit on the spatiotemporal mean and standard deviation of each pixel. The fitted deviation compensation function is as follows: R ( i,j )= A ( i,j ) I t s ( i,j ) 2 + B ( i,j ) I t s ( i,j )+ C ( i,j ), to obtain the quadratic fitting coefficients for each point ( A ( i,j ), B ( i,j ), C (i,j )).

[0016] Furthermore, the calibrated image is corrected using a deviation compensation function to obtain a corrected image that can effectively correct the inherent annular radiation error. for: I n ( i,j )= A ( i,j ) I t s ( i,j ) 2 + B ( i,j ) I t s ( i,j )+ C ( i,j ) To correct any point in the image.

[0017] The beneficial effects of this invention compared to the prior art are: This invention uses laboratory calibration images of a large-area array space infrared camera, including inner blackbody calibration images and outer blackbody calibration images, as a reference to calculate a correction function. It uses the radial gradient statistical method to determine whether there is inherent internal radiation circular noise in the system and obtains the relevant statistics of the internal radiation noise of the large-area array space camera system. These statistics are then applied to the inherent circular background noise formed by thermal radiation of the camera's internal components to compensate for the fixed circular noise in the image. This improves calibration accuracy, reduces calibration cost, and ensures the accuracy of subsequent absolute radiometric correction and inversion. Attached Figure Description

[0018] Figure 1 This is a flowchart of the testing method of the present invention. Detailed Implementation

[0019] The present invention will be further explained and described below with reference to the accompanying drawings and specific embodiments.

[0020] A method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system includes: Using on-board blackbody imaging, ≥10 frames of uniformly bright images with different brightness from a large-area space infrared camera under different imaging parameters are obtained. Typically, the brightness is selected as 0.1 full well, 0.9 full well, and three more levels within the range of 0.1 to 0.9 full well. In this embodiment, the brightness is selected as 0.1 full well, 0.3 full well, 0.5 full well, 0.7 full well, and 0.9 full well. Using external blackbody imaging, we obtained uniform images of different brightness from a large-area space infrared camera under different imaging parameters (the parameters correspond one-to-one with the on-board internal blackbody imaging), with each parameter having ≥100 frames. The arithmetic mean of 10 frames of blackbody images with different brightness levels was calculated to obtain the average blackbody images with different brightness levels. N t ( i,j ) and the spatial mean of the average image ; t These represent different brightness levels, divided into five levels: Level 1, Level 2, Level 3, Level 4, and Level 5. The least squares method is used to calculate the quadratic calibration coefficients of the mean image of the inner blackbody under different parameters, and the outer blackbody image is corrected to obtain the corrected image. Radial gradient statistics are performed on the corrected images with different parameters to determine whether the images contain inherent internal radiation circular noise of the system. If inherent noise exists, calibrate 100 frames of external blackbody images with four brightness levels under the same parameters. Typically, five different brightness levels are selected within the full-well range of 0.1 to 0.9. In this embodiment, the brightness levels are 0.1 full-well, 0.3 full-well, 0.5 full-well, 0.7 full-well, and 0.9 full-well. Then, the compensation function is calculated statistically and used to correct the external blackbody image to obtain the final corrected image.

[0021] Preferably, least squares fitting is performed on the average brightness images under each parameter to obtain the quadratic scaling coefficients. Specifically, the quadratic scaling coefficients are calculated based on known data points (…). , , , , We perform least-squares fitting, assuming the fitting function is... The quadratic scaling coefficients are obtained. ;in,( i,j () represents the horizontal and vertical components of a pixel. N 1 ( i,j ) represents the average image of the blackbody within the full well at 0.1. i,j The value at the position, N 2 ( i,j ) represents the average image of a blackbody within a 0.3 full-well. i,j The value at the position, N 3 ( i,j ) represents the average image of the blackbody within a 0.5 full-well. i,j The value at the position, N 4 ( i,j ) represents the average image of a blackbody within a 0.7 full-well. i,j The value at the position,N 5 ( i,j ) represents the average image of a blackbody within a 0.9 full-well. i,j The value at position D ( i,j ( ) is the calibrated image.

[0022] Then, least squares fitting is performed to obtain the quadratic scaling coefficients. .

[0023] Preferably, the blackbody image outside 100 frames is corrected using a quadratic scaling factor, specifically as follows: ,in, Uniform images of an outer blackbody under different imaging parameters and brightness; in, These are the second-order scaling coefficients.

[0024] Preferably, radial gradient statistics are performed on the corrected image to determine whether the image contains inherent internal radiation circular noise of the system. The specific method is as follows: Determine the center point of the image.

[0025] Calculate the radial gradient magnitude and direction; Determine the edge points; Calculate the radial gradient exponent to determine if circular noise exists.

[0026] Preferably, the radial gradient magnitude and direction are calculated as follows: Assuming the corrected image center is , x , y To correct the number of pixels in the image in both the horizontal and vertical directions, the correction is performed on any point in the image. The gradient magnitude and direction are obtained by performing convolution on the calibrated image using the Sobel operator: in, To correct the lateral radial gradient intensity of each pixel in the image, To correct the longitudinal and radial gradient intensity of each pixel in the image, To correct the total radial gradient intensity of each pixel in the image, To correct the radial gradient direction of each pixel in the image.

[0027] Preferably, based on the gradient intensity and direction of each pixel, non-edge points in the image are removed to obtain edge points, specifically as follows: in, These are radial gradient edge points, with a value of 1 for edge points and 0 for non-edge points.

[0028] Preferably, the radial gradient exponent is calculated based on the radial vector gradient strength and direction, specifically as follows: Where M is the set of radial gradient edge points, r p Center point To the edge point ( i,j The radial vector of ) for The total radial gradient intensity of pixels when =1; the threshold is set to 0.9. If the RGI is greater than 0.9, the calibrated image under this parameter is considered to still have inherent circular noise; if it is less than or equal to 0.9, the calibrated image under this parameter is considered to not have inherent circular noise.

[0029] Preferably, the compensation function is calculated statistically, specifically as follows: Calculate the spatiotemporal mean standard deviation of 100 frames of external blackbody corrected images; The deviation of the statistically corrected external blackbody image from the standard deviation of the spatiotemporal mean; The deviation at each point under different brightness levels is fitted using a polynomial function to obtain the deviation compensation function; The compensation function corrects the calibrated image, resulting in an image that effectively corrects the inherent circular radiation error.

[0030] Preferably, the arithmetic mean of 100 frames of external blackbody-corrected images under different brightness levels is calculated, specifically as follows: t =1, 2, 3, 4, 5 represent different brightness levels, which correspond to 0.1 full-well, 0.3 full-well, 0.5 full-well, 0.7 full-well, and 0.9 full-well, respectively. Represents the arithmetic mean of the blackbody-corrected image outside the U-frame; Then calculate the spatiotemporal mean. W t mean and standard deviation W t σ .

[0031] Preferably, the deviation of the statistically corrected outer blackbody image from the standard deviation of the spatiotemporal mean is as follows: Preferably, the spatiotemporal deviation of each pixel is fitted using least squares, assuming the fitting function is... The quadratic fitting coefficients for each point are obtained. .

[0032] Preferably, a compensation function is used to correct the calibrated image to obtain the final corrected image. The specific method is as follows: I n ( i,j )= A ( i,j ) I t s ( i,j ) 2 + B ( i,j ) I t s ( i,j )+ C ( i,j ).

[0033] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A method for correcting radiation error noise in a large-area array space infrared camera system, characterized in that, Includes the following steps: (1) Using internal blackbody imaging, the large-area array space infrared camera was used to obtain images under different imaging parameters and brightness conditions. L Intraframe blackbody uniform image ，L ≥10; (2) For different imaging parameters and different brightness levels L The arithmetic mean of the uniform blackbody images within each frame is then performed to obtain the mean blackbody image. N t ( i,j ) and the spatial mean of the inner blackbody average image , t These represent different brightness levels, categorized as Level 1, Level 2, Level 3, Level 4, and Level 5. i, j These represent the horizontal and vertical coordinates of the image pixels, respectively. (3) Using external blackbody imaging, according to the imaging parameters and brightness settings described in (1), obtain uniform external blackbody images of each U frame of the large-area array space infrared camera, U≥100; (4) Calculate the quadratic calibration coefficients of the inner blackbody average image in (2) using the least squares method, and use the quadratic calibration coefficients to correct the outer blackbody uniform image in (3) to obtain corrected images with different brightness under different parameters. (5) Perform radial gradient statistics on the corrected images with different brightness under different parameters to determine whether there is inherent internal radiation circular noise in the corrected images; if there is, proceed to step (6), otherwise end the correction. (6) The blackbody images of each U-frame with different brightness under each parameter are calibrated, the compensation function is statistically analyzed, and the compensation function is used to correct the calibrated blackbody correction image to obtain the final correction image.

2. The method for correcting radiation error noise in a large-area array space infrared camera system according to claim 1, characterized in that, The method of correcting the uniform image of the outer blackbody in (3) using quadratic scaling coefficients includes: First, calculate the second-order scaling coefficients for the five sets of known data points. , , , , Perform least squares fitting, assuming the fitting function is The quadratic scaling coefficients are obtained. , , ); in,( i,j () represents the horizontal and vertical components of a pixel. N 1 ( i,j ), N 2 ( i,j ), N 3 ( i,j ), N 4 ( i,j ), N 5 ( i,j The numbers ) represent the average blackbody images within the first, second, third, fourth, and fifth brightness levels, respectively. i,j The value at the position; the first brightness level is 0.1 full well, the fifth brightness level is 0.9 full well, and the second, third, and fourth brightness levels take different values ​​within the range of 0.1 to 0.9 full well; , , , , These represent the spatial mean values ​​of the blackbody average image within the first, second, third, fourth, and fifth brightness levels, respectively. The uniform image of the external blackbody is corrected using quadratic scaling coefficients to obtain corrected images with different brightness levels under different parameters. ; in, Uniform images of an outer blackbody under different imaging parameters and brightness.

3. The method for correcting radiation error noise in a large-area array space infrared camera system according to claim 1, characterized in that, The step (5) involves performing radial gradient statistics on the corrected images with different brightness levels under different parameters to determine whether the corrected images contain inherent internal radiation circular noise, including: Determine the center point of the corrected image; Calculate the total radial gradient intensity and radial gradient direction for each pixel in the corrected image; Remove non-edge points from the corrected image to obtain edge points; Calculate the radial gradient exponent and determine whether circular noise exists in the corrected image.

4. The method for correcting radiation error noise in a large-area array space infrared camera system according to claim 3, characterized in that, The calculation of the total radial gradient intensity and radial gradient direction of each pixel in the corrected image is specifically as follows: Assuming the corrected image center is , x, y To correct the number of pixels in the image in both the horizontal and vertical directions, the correction is performed on any point in the image. The Sobel operator is used to perform convolution calculations on the calibrated image to obtain the total radial gradient intensity and radial gradient direction for each pixel: ； ； ； ； in, To correct the lateral radial gradient intensity of each pixel in the image, To correct the longitudinal and radial gradient intensity of each pixel in the image, To correct the total radial gradient intensity of each pixel in the image, To correct the radial gradient direction of each pixel in the image.

5. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 4, characterized in that, Based on the total radial gradient intensity and radial gradient direction of each pixel in the corrected image, non-edge points in the image are removed to obtain edge points, specifically as follows: in, These are radial gradient edge points, with a value of 1 for edge points and 0 for non-edge points.

6. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 5, characterized in that, The radial gradient exponent is calculated based on the total radial gradient intensity of each pixel in the corrected image and the radial gradient direction. RGI Specifically: in, M It is the set of radial gradient edge points. r p ( i,j (with ) as the center point To the edge point The radial vector, for The total radial gradient intensity of the cell when =1; the threshold is set to 0.9, if RGI If the value is greater than 0.9, then the calibrated image under this parameter is considered to still have inherent circular noise. RGI If the value is ≤0.9, it is considered that the calibrated image under this parameter does not have inherent circular noise.

7. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 1, characterized in that, The statistical compensation function, and the use of the compensation function to correct the calibrated outer blackbody correction image, includes: Calculate the spatiotemporal mean and standard deviation of the U-frame out-of-frame blackbody corrected images under different brightness levels, and statistically analyze their deviation. The deviation at each point under different brightness levels is fitted using a polynomial function to obtain the deviation compensation function; The calibrated image is corrected using a deviation compensation function to obtain a corrected image that can effectively correct the inherent annular radiation error.

8. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 7, characterized in that, The calculation of the spatiotemporal mean and standard deviation of the U-frame out-of-frame blackbody corrected image under different brightness levels, and the statistical analysis of its deviation, includes: Calculate the arithmetic mean of the U-frame out-of-frame blackbody corrected images under different brightness levels: Calculate the spatiotemporal mean of the U-frame out-of-frame blackbody corrected image under different brightness levels. Standard deviation : ； ； Statistical analysis of the spatiotemporal mean of U-frame out-of-frame blackbody corrected images under different brightness levels Standard deviation Deviation: t Represents different brightness levels ，s Represents the number of frames. This represents the spatiotemporal mean of the external blackbody corrected image under different brightness levels. Standard deviation The degree of deviation; x, y To correct the number of pixels in the image in both the horizontal and vertical directions; To correct any point in the image, Represents the horizontal and vertical components of any pixel in the corrected image; This represents the arithmetic mean of the blackbody-corrected image outside the U-frame.

9. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 8, characterized in that, The aforementioned deviation compensation function is obtained by performing a least-squares fit on the spatiotemporal mean and standard deviation deviation of each pixel. The fitted deviation compensation function is as follows: R ( i,j )= A ( i,j ) I t s ( i,j ) 2 + B ( i,j ) I t s ( i,j )+ C ( i,j ), to obtain the quadratic fitting coefficients for each point ( A ( i,j ), B ( i,j ), C ( i,j )).

10. The method for correcting the inherent internal radiation circular noise of a large-area array space infrared camera system according to claim 9, characterized in that, The calibrated image is corrected using a deviation compensation function to obtain a corrected image that can effectively correct the inherent annular radiation error. for: I n （ i,j ）= A （ i,j ） I t s （ i,j ） 2 + B （ i,j ） I t s （ i,j ）+ C （ i,j ） in, I t s ( i,j () is used to correct any point in the image.

Images