An Image Restoration Method for Optical Synthetic Aperture Imaging (OSAI)

Abstract

The present invention proposes an image restoration method for optical synthetic aperture imaging (OSAI), and the implementation steps are as follows: (1) Filter the optical synthetic aperture imaging (OSAI) image based on the improved Wiener filtering algorithm; (2) Obtain the optimal cut-off frequency D best ; (3) Divide the spectrum diagram of the preliminarily restored image into high and low frequencies; (4) Perform high-frequency emphasis and bilateral filtering on the high and low frequency spectrum diagrams of the preliminarily restored image; (5) Obtain the restoration result of the OSAI image. By calculating the comprehensive evaluation index Q of the preliminarily restored image after processing the degraded image and selecting the optimal solution of the parameter K to replace the parameter K in the Wiener filtering algorithm, the present invention can more accurately select the appropriate K value of the Wiener filtering algorithm, so that the preliminary restoration effect reaches the optimal effect of the Wiener filtering algorithm. Then, high-frequency emphasis and bilateral filtering are performed on the high and low frequency spectrum diagrams of the preliminarily restored image, effectively improving the resolution of the restored image and meeting the needs of space observation.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and particularly relates to an image restoration method for optical synthetic aperture imaging (OSAI), which can be used in fields such as space observation. Background Art

[0002] The continuous development of deep space exploration has put extremely high requirements on the observation capabilities of optical imaging systems. A single-aperture imaging system with a resolution of 0.01 arcseconds requires an aperture on the order of 100 meters, and current processing technologies are difficult to meet this demand. An optical synthetic aperture imaging system forms an image on the imaging plane through a sub-aperture array that meets the co-phase requirements, thereby effectively improving the imaging resolution of the system. However, due to the discrete distribution of its sub-apertures, the actual light-transmitting area of the entire imaging system is relatively small compared to that of an equivalent single-aperture imaging system, resulting in the expansion of the actual point spread function (PSF) of the system, the decline or even absence of the intermediate-frequency response of the modulation transfer function (MTF), and thus serious degradation of the system imaging and reduction of resolution. Image restoration technology needs to be combined to obtain high-resolution images.

[0003] Currently, the restoration methods for optical synthetic aperture images mainly include restoration methods based on Wiener filtering algorithms, constrained least squares filtering algorithms, Richardson-Lucy algorithms, etc. The main idea is to use the prior knowledge of the image degradation process to establish a degradation model and reverse deduce to restore the image.

[0004] The Wiener filtering algorithm is a classic basic image restoration algorithm, also known as the minimum mean square error filtering algorithm. It has the advantages of good restoration effect, small algorithm computation, and less affected by noise, and is widely used in image restoration. Its basic idea is to find an estimated ideal image of the reference image f(x,y) under certain constraint conditions such that the mean square error between the reference image and the estimated ideal image is minimized. A major difficulty of the Wiener filtering algorithm is the determination of the parameter K value. The selection of the K value determines the quality of the restored image by the Wiener filtering algorithm. By selecting an appropriate parameter K value, a restored image with a better restoration effect can be obtained. However, the traditional Wiener filtering algorithm requires manual estimation of the K value, and it is difficult to select the optimal K value to obtain the best restored image.

[0005] Based on the advantages and disadvantages of the Wiener filtering algorithm, researchers have developed an improved image restoration method for optical synthetic aperture imaging (OSAI). For example, Cao Yang disclosed an image restoration method for optical synthetic aperture imaging that improves Wiener filtering based on the weighted maximum value method in his master's thesis in 2020. First, the ideal point spread function (PSF) is calculated through the characteristics of the array configuration, and its calculation formula is: where PSF sub (x,y) represents the point spread function of a single aperture, λ represents the wavelength, f represents the focal length, (x,y) represents the two variable coordinates of the rectangular coordinate system corresponding to the PSF in the image plane, t represents the t-th group of sub-apertures in the initial array, ζ,η represent the coordinate directions of the image plane to be observed, N represents the number of sub-apertures in the initial array, M represents the total number of scans, m represents the m-th scan, n represents the n-th sub-aperture after each scan, (Δζ mt , Δη mt ) represents the center distance of the t-th group of sub-apertures in the m-th scan of the initial array, and cos|·| represents the cosine operation; then, by comparing the correlation coefficient C and the peak signal-to-noise ratio PSNR of the restored image with the prior as the ratio of the parameter noise power spectrum to the ideal power spectrum (Noise-to-Signal Ratio, NSR) changes, the weighted average of the parameters NSR1 and NSR2 corresponding to the maximum values of the correlation coefficient C and the peak signal-to-noise ratio PSNR is determined, and (NSR1 + NSR2) / 2 is substituted into the Wiener filtering algorithm to replace the K value in the original algorithm to solve the restored image. This method effectively improves the overall clarity of the image, but the resolution of the restored image is still relatively low. Summary of the Invention

[0006] The object of the present invention is to propose an image restoration method for optical synthetic aperture imaging (OSAI) in view of the above-mentioned deficiencies of the prior art, so as to solve the technical problem of the relatively low resolution of the restored image existing in the prior art.

[0007] To achieve the above object, the technical solution adopted by the present invention includes the following steps:

[0008] (1) Filter the optical synthetic aperture imaging (OSAI) image based on the improved Wiener filtering algorithm:

[0009] Calculate the comprehensive evaluation index Q of the preliminary restored image based on the peak signal-to-noise ratio, correlation coefficient, and standard deviation of the preliminary restored images obtained by the Wiener filtering algorithm with different K values. Then, use the K value corresponding to the maximum value of Q as the optimal solution K_opt to replace the K value in the Wiener filtering algorithm, realizing the optimization of the K value of the Wiener filtering algorithm. Next, perform Wiener filtering on the Fourier transform result G(u,v) of the optical synthetic aperture imaging (OSAI) image g(x,y) of size M×N based on the improved Wiener filtering algorithm with K value K_opt to obtain the frequency spectrum diagram F(u,v) of the preliminary restored image. Among them, the expression of the improved Wiener filtering algorithm is:

[0010]

[0011] Among them, M≥128, N≥128, x and y represent spatial coordinates, u and v represent frequency domain coordinates, H(u,v) is the degradation function, and |·| 2 represents the complex conjugate operation, and K_opt represents the optimal solution of the K value;

[0012] (2) Obtain the optimal cut-off frequency D best :

[0013] (3) Divide the frequency spectrum diagram of the preliminary restored image into high and low frequencies:

[0014] Form the high-frequency spectrum diagram F1 of the preliminary restored image with the pixel points in the frequency spectrum diagram F(u,v) of the preliminary restored image whose frequencies are greater than the optimal cut-off frequency D best and form the low-frequency spectrum diagram F2 of the preliminary restored image with the remaining pixel points;

[0015] (4) Perform high-frequency emphasis and bilateral filtering on the high and low frequency spectrum diagrams of the preliminary restored image:

[0016] Perform high-frequency emphasis filtering on the high-frequency spectrum diagram F1 of the preliminary restored image to obtain the high-frequency spectrum diagram after emphasis filtering At the same time, perform bilateral filtering on the low-frequency spectrum diagram F2 of the preliminary restored image to obtain the low-frequency spectrum diagram after bilateral filtering

[0017] (5) Obtain the restored result of the OSAI image:

[0018] For the high-frequency spectrum diagram and the low-frequency spectrum diagram perform splicing, and perform inverse Fourier transform on the spliced frequency spectrum diagram to obtain the restored image of the OSAI image

[0019] Compared with the prior art, the present invention has the following advantages:

[0020] The present invention updates the parameter K value through the comprehensive evaluation index Q calculated from the peak signal-to-noise ratio, correlation coefficient, and standard deviation of the preliminary restored image after degraded image processing, obtains the optimal solution of the parameter K value corresponding to the maximum Q, and replaces the parameter K in the Wiener filtering algorithm with the K value of the optimal solution. Then, the preliminary restored image obtained by processing based on the improved Wiener filtering algorithm is further processed by high-frequency emphasis and bilateral filtering for the high-frequency and low-frequency parts in the frequency domain respectively, avoiding the influence of inaccurate optimal solution of the K value on the resolution of the restored image caused by incomplete evaluation index when replacing the K value in the Wiener filtering algorithm with the weighted average value of the parameter corresponding to the maximum of the correlation coefficient and peak signal-to-noise ratio in the prior art, and being able to obtain a high-resolution restored image with smooth background and clear details. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flowchart for the implementation of the present invention.

[0022] Figure 2 It is an optical synthetic aperture imaging (OSAI) image to be restored according to the present invention.

[0023] Figure 3 It is a comparison diagram of the restoration image effects between the present invention and the prior art. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following further describes the present invention in detail with reference to the drawings and specific embodiments.

[0025] Referring to Figure 1 , the present invention includes the following steps:

[0026] (1) Filter the optical synthetic aperture imaging (OSAI) image based on the improved Wiener filtering algorithm:

[0027] Calculate the comprehensive evaluation index Q of the preliminary restored image through the peak signal-to-noise ratio, correlation coefficient, and standard deviation of the preliminary restored image obtained by the Wiener filtering algorithm with different K values, and use the K value corresponding to the maximum Q as the optimal solution K_opt to replace the K value in the Wiener filtering algorithm, realizing the optimization of the K value of the Wiener filtering algorithm. Then, perform Wiener filtering on the Fourier transform result G(u, v) of the optical synthetic aperture imaging (OSAI) image g(x, y) with size M×N based on the improved Wiener filtering algorithm with K value of K_opt to obtain the frequency spectrum diagram F(u, v) of the preliminary restored image. Among them, the expression of the improved Wiener filtering algorithm is:

[0028]

[0029] Among them, M≥128, N≥128, x, y represent spatial coordinates, u, v represent frequency domain coordinates; H(u, v) is the degradation function, |·| 2 represents the complex conjugate operation;

[0030] The optical synthetic aperture imaging (OSAI) image is as Figure 2 shown, Figure 2 which is a standard resolution version image with degradation and noise addition, and its imaging quality is very poor, and basically no details in the image can be seen. The standard resolution plate is composed of multiple groups of black and white stripes with different spatial frequencies and has the characteristics of most resolution cards, so it is used to evaluate the imaging quality of the optical system.

[0031] Among them, the specific steps for optimizing the parameter K of the Wiener filtering algorithm are as follows:

[0032] (1a) Initialize the number of iterations as t, the maximum number of iterations as T, T ≥ 1000, the iteration step size as ΔK, and the parameter of the current Wiener filtering algorithm as K t , and let t = 0, K0 = 0;

[0033] (1b) Perform Fourier transform on the optical synthetic aperture imaging (OSAI) image g(x, y) of size M × N, and perform Wiener filtering on the Fourier-transformed frequency spectrum diagram G(u, v) with K t as the parameter of the Wiener filtering algorithm to obtain the frequency spectrum diagram F t (u, v);

[0034]

[0035]

[0036] H(u, v) = OTF(u, v)

[0037] where OTF(u, v) is the optical transfer function of the optical synthetic aperture;

[0038] (1c) Perform inverse Fourier transform on the Wiener-filtered frequency spectrum diagram F t (u, v), and calculate the comprehensive evaluation index Q t of the preliminary restored image f t (x, y) according to the peak signal-to-noise ratio PSNR t , correlation coefficient C t and standard deviation STD t of f t ;

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] Q t = w1 * PSNR t / 100 + w2 * C t + w3 * STD t

[0045] where, e [·] represents the exponential operation, lg[·] represents the logarithm operation with base 10, MAX I represents the maximum gray level of the image, ∑[·] represents the summation operation, m, n represent the summation variables, f′(x, y) represents the ideal reference image, E[·] represents the mathematical expectation, u represents the average value of image pixels, and w1, w2, w3 respectively represent the weight coefficients of PSNR t , C t , STD t , and w1 + w2 + w3 = 1;

[0046] (1e) Judge whether t = T holds. If so, obtain the optimal comprehensive evaluation index Q tmax , and take the K value Kopt corresponding to Q tmax as the optimal solution of the K value of the Wiener filtering algorithm parameter. Otherwise, let t = t + 1, let K t = K t-1 + ΔK, and execute step (1b).

[0047] In this example, it is set that w1, w2, w3 are 0.4, 0.4, 0.2 respectively, the initial K value K0 = 10 -9 , the iteration step size ΔK = 10 -7 , and the total number of iterations T = 1000. When the number of iterations is 643 times, the comprehensive evaluation index Q value reaches the maximum value Q = 0.7010. At this time, the corresponding K value is the optimal solution. To reduce the iteration error caused by a relatively large iteration step size, the optimal solution of the K value is calculated by fitting three points around the peak point through interpolation operation. After calculation, the obtained K_opt = 6.4692×10 -5 .

[0048] (2) Obtain the optimal cut-off frequency D best :

[0049] (2a) Initialize the number of iterations as s, the maximum number of iterations as S, S ≥ 100, the iteration step size as ΔD, and the current high-low frequency division cut-off frequency as D s , and let s = 0, D0 = 0;

[0050] (2b) Perform an inverse Fourier transform on the spectrum diagram F(u, v) of the preliminary restored image, perform bilateral filtering on the preliminary restored image f(x, y) after the inverse Fourier transform, and then perform a Fourier transform on the restored image l(x, y) after bilateral filtering to obtain the spectrum diagram L(u, v) of the restored image after bilateral filtering:

[0051]

[0052]

[0053]

[0054]

[0055] where w s represents the spatial domain weight, w r represents the gray domain weight, Ω represents the neighborhood range at the pixel (x, y), w p represents the normalization parameter, σ s represents the spatial variance, σ r represents the gray variance, and i, j represent the summation variables.

[0056] (2c) The pixel points in the spectrum diagram F(u, v) of the preliminary restored image with frequencies greater than D s form the high-frequency spectrum diagram F s1 of the preliminary restored image. Perform high-frequency emphasis filtering on F s1 to obtain the high-frequency spectrum diagram after emphasis filtering. And the remaining pixel points form the low-frequency spectrum diagram F s2 of the preliminary restored image. Replace the low-frequency spectrum diagram of the spectrum diagram L(u, v) of the restored image after bilateral filtering that is less than D s to obtain the low-frequency spectrum diagram after bilateral filtering. Superimpose the high-frequency spectrum diagram and the low-frequency spectrum diagram , and perform an inverse Fourier transform on the superimposed spectrum diagram to obtain the restored image of the OSAI image Calculate the comprehensive evaluation index Q s of s according to the peak signal-to-noise ratio PSNR s , the correlation coefficient C and the standard deviation STD s ;

[0057]

[0058] Wherein, a represents the offset, b is a constant greater than 1, and D(u, v) represents the distance from the point (u, v) in the frequency domain to the center of the filter;

[0059] (2d) Determine whether s = S holds. If so, obtain the maximum value Q of the comprehensive evaluation index smax and use Q tmax corresponding frequency D best as the optimal cut-off frequency. Otherwise, set s = s + 1, D s = D s-1 + ΔD, and execute step (2b).

[0060] In this example, the iteration step size ΔD = 10 -7 and the calculated D 0best = 333.4566.

[0061] (3) Divide the spectrum diagram of the preliminary restored image into high and low frequencies:

[0062] Form the high-frequency spectrum diagram F1 of the preliminary restored image from the pixel points in the spectrum diagram F(u, v) of the preliminary restored image whose frequencies are greater than the optimal cut-off frequency D best , and form the low-frequency spectrum diagram F2 of the preliminary restored image from the remaining pixel points;

[0063] The low-frequency components in the image refer to the areas where the image color and gray level change slowly, which is equivalent to the area of large color blocks; the high-frequency components in the image refer to the areas where the image color and gray level change rapidly, that is, the gray level difference between adjacent regions is large, which is what we often call the edge contour. Therefore, the high-frequency components of the image can refer to the edge and detail parts of the image, and the low-frequency components can refer to the background part of the image.

[0064] (4) Perform high-frequency emphasis and bilateral filtering on the high and low frequency spectrum diagrams of the preliminary restored image:

[0065] Perform high-frequency emphasis filtering on the high-frequency spectrum diagram F1 of the preliminary restored image to obtain the high-frequency spectrum diagram after the emphasized filtering At the same time, perform bilateral filtering on the low-frequency spectrum diagram F2 of the preliminary restored image to obtain the low-frequency spectrum diagram after the bilateral filtering

[0066] The high-frequency spectrum diagram F1 of the preliminary restored image uses high-frequency emphasis filtering to highlight details; the low-frequency spectrum diagram F2 of the preliminary restored image uses the bilateral filtering algorithm to achieve smooth denoising while protecting the edges.

[0067] In this example, the parameter spatial variance σ s in the bilateral filtering algorithm is 3, the gray level variance σ r is 0.1, the filtering radius is 5; the parameter offset a in the high-frequency emphasis filtering is 0.5, and the multiplier b is 3.

[0068] (5) Obtain the restored result of the OSAI image:

[0069] For the high-frequency spectrogram and the low-frequency spectrogram perform splicing, and perform inverse Fourier transform on the spliced spectrogram to obtain the restored image of the OSAI image

[0070] For the high-frequency spectrogram and the low-frequency spectrogram The specific method of splicing is: splice the high-frequency spectrogram after the low-frequency spectrogram and there is no overlap between .

[0071] The technical effects of the present invention will be described below in combination with simulation experiments:

[0072] 1. Simulation conditions and content

[0073] The simulation hardware is: a microcomputer, and the computer parameters are CPU: Intel(R) Core(TM) i5-8265U CPU@1.60GHz~1.80GHz; RAM: 8.00GB; operating system: Windows10. The simulation software is: computer software MATLAB2020b.

[0074] A comparative simulation is performed on the restored image effects of the present invention and the existing image restoration method of optical synthetic aperture imaging based on the weighted maximum value method improved Wiener filtering, and the results are as Figure 3 shown.

[0075] 2. Analysis of simulation results:

[0076] Referring to Figure 3 , Figure 3 (a), Figure 3 (b) are respectively the restored images of the prior art and the present invention. It can be seen that Figure 3 the background of (a) still has impurities and the details are not clear enough, Figure 3 the background of the image in (b) is smoother, basically free of impurities, and the details such as numbers are also clearer and more complete. Calculate the comprehensive evaluation index Q of the restored images of the present invention and the prior art. The comprehensive evaluation indexes of the prior art and the present invention are 0.7010 and 0.7435 respectively. Then the comprehensive evaluation index Q of the restored image of the optical synthetic aperture imaging OSAI image obtained by the present invention is increased by 6.06% compared with the prior art. The present invention can improve the resolution of the restored image and meet the visual effect of the restored image.

[0077] The above description is only a specific example of the present invention and does not constitute any limitation to the present invention. Obviously, for professionals in this field, after understanding the content and principle of the present invention, various modifications and changes in any form and details may be made without departing from the principle and structure of the present invention. However, these corrections and changes based on the idea of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. An image restoration method for optical synthetic aperture imaging (OSAI), characterized in that, It includes the following steps: (1) Filter the optical synthetic aperture imaging (OSAI) image based on the improved Wiener filtering algorithm: Calculate the comprehensive evaluation index Q of the preliminary restored image through the peak signal-to-noise ratio, correlation coefficient, and standard deviation of the preliminary restored images obtained by the Wiener filtering algorithm with different K values, and use the K value corresponding to the maximum value of Q as the optimal solution Kopt to replace the K value in the Wiener filtering algorithm, so as to optimize the K value of the Wiener filtering algorithm. Then, perform Wiener filtering on the Fourier transform result G(u, v) of the optical synthetic aperture imaging (OSAI) image g(x, y) with size M×N based on the improved Wiener filtering algorithm with K value of K_opt to obtain the frequency spectrum diagram F(u, v) of the preliminary restored image. The expression of the improved Wiener filtering algorithm is: where M≥128, N≥128, x and y represent spatial coordinates, u and v represent frequency domain coordinates; H(u, v) is the degradation function, and |·| 2 represents the complex conjugate operation; (2) Obtain the optimal cut-off frequency D best : (3) Divide the frequency spectrum diagram of the preliminary restored image into high and low frequencies: Pixels in the preliminary restored image spectrogram F(u, v) with frequencies greater than the optimal cut-off frequency D best are used to form the high-frequency spectrogram F1 of the preliminary restored image, and the remaining pixels are used to form the low-frequency spectrogram F2 of the preliminary restored image; (4) Perform high-frequency emphasis and bilateral filtering on the high and low frequency spectrum diagrams of the preliminary restored image: Perform high-frequency emphasis filtering on the high-frequency spectrum diagram F1 of the preliminary restored image to obtain the high-frequency spectrum diagram after the emphasized filtering. At the same time, perform bilateral filtering on the low-frequency spectrum diagram F2 of the preliminary restored image to obtain the low-frequency spectrum diagram after the bilateral filtering. (5) Obtain the restored result of the OSAI image: For the high-frequency spectrogram and the low-frequency spectrogram are stitched together, and the stitched spectrogram is subjected to inverse Fourier transform to obtain the restored image of the OSAI image 2. The image restoration method of an optical synthetic aperture imaging (OSAI) according to claim 1, characterized in that, The implementation steps of the optimization of the K value of the Wiener filtering algorithm described in step (1) are: (1a) Initialize the number of iterations as t, the maximum number of iterations as T, where T ≥ 1000, the iteration step size as ΔK, and the parameter of the current Wiener filtering algorithm as K t , and set t = 0, K0 = 0; (1b) Perform a Fourier transform on the optical synthetic aperture imaging OSAI image g(x,y) of size M×N, and use K t as the parameter of the Wiener filtering algorithm to perform Wiener filtering on the spectrogram G(u,v) after Fourier transform, obtaining the spectrogram F t (u,v); H(u, v) = OTF(u, v) where OTF(u, v) is the optical transfer function of the optical synthetic aperture; (1c) Perform inverse Fourier transform on the spectrogram F t (u, v) after Wiener filtering, and calculate the peak signal-to-noise ratio PSNR t of the preliminary restored image f t (x, y), the correlation coefficient C t and the standard deviation STD t , and calculate the comprehensive evaluation index Q t of f t ; Q t = w1 * PSNR t / 100 + w2 * C t + w3 * STD t Among them, e [·] represents the exponential operation, lg[·] represents the logarithm operation with base 10, MAX I represents the maximum gray level of the image, ∑[·] represents the summation operation, m and n represent the summation variables, f′(x,y) represents the ideal reference image, E[·] represents the mathematical expectation, u represents the mean value of the image pixels, and w1, w2, and w3 respectively represent the weight coefficients of t PSNR t , C t , and STD , and w1 + w2 + w3 = 1; (1e) Determine whether t = T holds. If so, obtain the optimal comprehensive evaluation index Q tmax , and use the K value Kopt corresponding to Q tmax as the optimal solution of the K value of the Wiener filtering algorithm parameter. Otherwise, set t = t + 1, set K t = K t-1 +ΔK, and execute step (1b).

3. The image restoration method of an optical synthetic aperture imaging OSAI according to claim 1, wherein, The obtaining of the optimal cut-off frequency D described in step (2) best , and the implementation steps are as follows: (2a) Initialize the number of iterations as s, the maximum number of iterations as S, where S ≥ 100, the iteration step size as ΔD, and the current high-low frequency division cut-off frequency as D s , and set s = 0 and D0 = 0; (2b) Perform inverse Fourier transform on the frequency spectrum diagram F(u, v) of the preliminary restored image, perform bilateral filtering on the preliminary restored image f(x, y) after the inverse Fourier transform, and then perform Fourier transform on the restored image l(x, y) after the bilateral filtering to obtain the frequency spectrum diagram L(u, v) of the restored image after bilateral filtering: Among them, w s represents the spatial domain weight, w r represents the gray-scale domain weight, Ω represents the neighborhood range at the pixel (x, y), w p represents the normalization parameter, σ s represents the spatial variance, σ r represents the gray-scale variance, and i, j represent the summation variables; (2c) The pixel points in the preliminary restored image spectrogram F(u, v) with frequencies greater than D s form the high-frequency spectrogram F of the preliminary restored image s1 . Perform high-frequency emphasis filtering on F s1 to obtain the high-frequency spectrogram after emphasis filtering and the remaining pixel points form the low-frequency spectrogram F of the preliminary restored image s2 . Use bilateral filtering to replace the low-frequency spectrogram of the restored image spectrogram L(u, v) less than D s to obtain the low-frequency spectrogram after bilateral filtering For the high-frequency spectrogram and the low-frequency spectrogram are superimposed, and the superimposed spectrogram is subjected to inverse Fourier transform to obtain the restored image of the OSAI image According to 's peak signal-to-noise ratio PSNR s , correlation coefficient C s and standard deviation STD s Calculate 's comprehensive evaluation index Q s ; where a represents the offset, b is a constant greater than 1, and D(u, v) represents the distance from the point (u, v) in the frequency domain to the center of the filter; (2d) Determine whether s = S holds. If so, obtain the maximum value Q of the comprehensive evaluation index smax , and use Q tmax corresponding frequency D best as the optimal cut-off frequency. Otherwise, set s = s + 1, D s = D s-1 +ΔD, and execute step (2b).

4. The image restoration method of an optical synthetic aperture imaging OSAI according to claim 3, characterized in that, The high-frequency emphasis and bilateral filtering of the high and low frequency spectrum diagrams of the preliminary restored image described in step (4) are the same as the method in step (2c).

5. A method for image restoration of an optical synthetic aperture imaging OSAI according to claim 1, characterized in that, The high-frequency spectrogram described in step (5) and the low-frequency spectrogram are spliced, which means that the high-frequency spectrogram is spliced after the low-frequency spectrogram and there is no overlap between and ​

Images