An imaging system and super-resolution method based on variable aperture coding

Abstract

The application discloses an imaging system based on variable aperture coding and a super-resolution method. The system is composed of an imaging lens group, a variable aperture, an image acquisition device and an image processing module. The variable aperture is composed of an aperture blade part and an aperture fixing base. A series of low-resolution images are obtained by adjusting the aperture blade part to modulate a system point spread function, exposure compensation is performed on the captured images, and iterative back projection is performed under double constraints in a space domain and a frequency domain, so that a cost function is minimized, a super-resolution image is obtained, and the imaging limitation caused by insufficient space sampling of a detector is broken through. Compared with an existing coding imaging system, the imaging system does not need to introduce an additional mechanical scanning device, coding imaging can be realized by adjusting the aperture, the operation is simple, the cost is low, the signal-to-noise ratio is improved, and better image quality is obtained. The imaging system and the super-resolution algorithm thereof can successfully realize super-resolution imaging in a visible light wave band, and the imaging resolution can be improved by nearly two times.

Description

Technical Field

[0001] This invention belongs to computational optical imaging technology, specifically an imaging system and super-resolution method based on variable aperture coding. Background Technology

[0002] Optoelectronic technology, as a cutting-edge technology, plays a vital role in various industries. High-resolution imaging is crucial in the automotive, environmental monitoring, and military imaging fields. Obtaining clear images with high resolution, high sensitivity, and high dynamic range has always been a perpetual goal in imaging. However, for traditional "what you see is what you get" imaging systems, image resolution is heavily dependent on the sensor's pixel size. Achieving even a slight improvement in imaging performance using traditional optical imaging system design approaches usually means a sharp increase in hardware costs. Conversely, simply shrinking the sensor pixel size can lead to low sensitivity, photoelectric crosstalk, and other problems that negatively impact image quality and increase data storage complexity.

[0003] In order to overcome the imaging limitations caused by the pixel size of the detector, researchers have proposed a variety of super-resolution imaging techniques ([1] Zhang Haitao, Zhao Dazun. Mathematical principle and implementation of micro-scanning to reduce spectral aliasing in photoelectric imaging systems [J]. Acta Optica Sinica, 1999(09):1263-1268.). Micro-scanning is one of the common solutions to reduce aliasing: acquiring multiple frames of sub-pixel level displacement images and reconstructing a single high-resolution image. A common implementation method is mechanical translation (Jean Fortin, Paul Chevrere. Realization of a fast microscanning device for infrared focal plane arrays [A]. SPIE [C]. 1996, 2743:185~196.). It uses piezoelectric drive to directly drive the lens to achieve lens translation. The whole system is simple to control and the displacement is precisely adjustable. However, the piezoelectric drive circuit is relatively complex, and the optical design is limited by the micro-displacement mechanism, resulting in poor versatility. The mirror-scanning method (Jean Fortin, Paul Chevrere. Realization of a fast microscanning device for infrared focal plane arrays[A]. SPIE[C].1996,2743:185~196.) can achieve multi-frame sub-pixel displacement image acquisition in the same scene, thereby optimizing the final imaging quality. However, this method requires additional moving parts or mirrors, making the system very complex; and when the image quality is affected by different displacements, more microscans are needed, sacrificing temporal resolution. As a result, the experimental conditions are very demanding, and the experimental operability is poor, so reconstructing high-resolution images is a very difficult process. In 2005, Solomon J et al. proposed placing a mask in the Fourier plane of the imaging system, which encodes the object spectrum and then decodes the image spectrum (Solomon J, Zalevsky Z, Mendlovic D. Geometric superresolution by code division multiplexing[J].Applied optics,2005,44(1):32-40.). While masking can overcome the spectral aliasing problem caused by the distance between two adjacent pixel centers of a CCD, this method ignores the size of CCD pixels, treating them as ideal points, and does not solve the low-pass effect caused by the size and shape of each CCD pixel. Therefore, achieving high-resolution imaging beyond the resolution limitations of imaging detectors without using any mechanical scanning devices has become a crucial technical challenge. Summary of the Invention

[0004] The purpose of this invention is to provide an imaging system based on variable aperture coding, which does not require the addition of additional mechanical components or complex coding patterns, thereby reducing the exposure time required by the image acquisition device, increasing the image acquisition speed of the system, improving the signal-to-noise ratio, and improving the image reconstruction quality.

[0005] The technical solution for achieving the objective of this invention is as follows: an imaging system based on variable aperture coding, comprising: an imaging lens group, a variable aperture, an image acquisition device, and an image processing module; wherein, the variable aperture includes an aperture blade portion and an aperture mounting base, the aperture blade portion of the variable aperture is coaxial with the imaging lens group, and the aperture blade portion is fixed to the imaging lens group by the aperture mounting base, and the image acquisition device is located on the back focal plane of the imaging lens group; the imaging lens group and the variable aperture are relatively fixedly installed, and the variable aperture and the image acquisition device are mounted and fixed on an optical platform; the incident light entering the imaging lens group is controlled by adjusting the effective aperture size of the aperture blade portion of the variable aperture, the image acquisition device acquires coded images at different aperture sizes and real-world low-resolution images corresponding to different coded patterns, and the image processing module is used to process the acquired images to obtain high-resolution images.

[0006] This invention also proposes a super-resolution method based on variable aperture coding, the specific steps of which are as follows:

[0007] (1) Based on the wavelength λ of the current imaging system, the pixel size of the detector, and the required number of codes N, calculate the cutoff aperture number F1 and the number of equally spaced reduction apertures ΔF;

[0008] (2) Adjust the aperture blades of the variable aperture, reduce the number of aperture blades with equal spacing to obtain N different coding patterns. k Let the k-th coded pattern be denoted as Pattern. k and its corresponding optical transfer function (OTF) k , where k = 1...N;

[0009] (3) Use an image acquisition device to capture N different coded patterns. k Corresponding real-world low-resolution images All real-shot low-resolution images Compared with the selected reference image standard Exposure compensation is applied. Here, the last coded pattern is selected. N Corresponding real-world low-resolution images Using the baseline image, the gain coefficient matrix (gain) is obtained. k After obtaining exposure compensation, low-resolution images were captured.

[0010] (4) The aperture value after exposure compensation is F N Real-world low-resolution images Upsampling is performed to obtain a high-resolution object. ini As the initial value for iteration, a Fourier transform is performed on it to obtain the initial high-resolution object spectrum. ini ;

[0011] (5) In the iterth iteration, select the aperture number F. k The corresponding optical transfer function (OTF) k To capture the spectrum of a high-resolution object iter,k Information is used to obtain an estimated high-resolution image spectrum. After performing an inverse Fourier transform on the spectrum of the high-resolution image and downsampling, an estimated low-resolution image is obtained.

[0012] (6) Estimated low-resolution image Low-resolution images captured after exposure compensation Divide to obtain a low-resolution intensity coefficient matrix. For low-resolution intensity coefficient matrix Upsampling is performed to obtain a high-resolution intensity coefficient matrix. k The light intensity distribution used to update the estimated high-resolution image Get updated high-resolution images

[0013] (7) Update the high-resolution image spectrum With the estimated high-resolution image spectrum Subtraction yields the frequency domain increment ΔImage iter,k The increment is then subjected to Wiener filtering to obtain the updated high-resolution object spectrum.

[0014] (8) Calculate the updated high-resolution object light intensity distribution The cost function ε of the low-resolution image after exposure compensation k ;

[0015] (9) Let k = k + 1, and repeat steps (5)-(8) until N exposure compensations are applied to capture a low-resolution image. Iterate through all of them once, and set the N cost functions ε k The average cost function ε is obtained by summing and averaging.

[0016] (10) Let iter = iter + 1, enter the next iteration, and repeat steps (5)-(9) until the average cost function ε is less than the error threshold T.

[0017] Compared with the prior art, the present invention has the following significant advantages: (1) The present invention controls the incident light entering the imaging lens group by adjusting the effective aperture size of the aperture blades of the variable aperture, thereby acquiring coded images under different aperture sizes. No additional mechanical scanning device is required, the structure is compact, the encoding method is simple, and the operation is easy. (2) The coded imaging super-resolution algorithm used in the present invention effectively overcomes the image pixelation problem caused by excessively large pixel size, improves the signal-to-noise ratio, obtains better image quality, and can improve the imaging resolution by nearly two times.

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0019] Figure 1 This is a structural diagram of the variable aperture coded imaging system of the present invention.

[0020] Figure 2 This is a structural diagram of the aperture blade section of the variable aperture used in this invention.

[0021] Figure 3 This is the aperture holder with variable aperture used in this invention.

[0022] Figure 4 The variable aperture used in this invention refers to the aperture number F corresponding to the aperture number from the cutoff aperture F1 to the maximum effective aperture. N The modulated coded pattern.

[0023] Figure 5 The aperture number ranges from the cutoff aperture F1 to the aperture number F corresponding to the maximum effective aperture. N The modulated coded pattern corresponds to the generated optical transfer function.

[0024] Figure 6 The variable aperture used in this invention ranges from the number of apertures corresponding to the smallest effective aperture to the number of apertures corresponding to the largest effective aperture. N The modulated coded pattern.

[0025] Figure 7 The aperture number ranges from the smallest effective aperture to the largest effective aperture (F). N The modulated coded pattern corresponds to the generated optical transfer function.

[0026] Figure 8 This is a schematic diagram of the super-resolution reconstruction process of the present invention.

[0027] Figure 9 The images show the low-resolution raw images and their spectra captured by an image acquisition device with a pixel size of 1.85μm × 1.85μm before super-resolution reconstruction. Figure 10 The variable aperture used in this invention refers to the aperture number F corresponding to the aperture number from the cutoff aperture F1 to the maximum effective aperture. N The super-resolution results of the modulation method on the USAF resolution plate imaging experiment. Figure 11 The variable aperture used ranges from the f-number corresponding to the smallest effective aperture to the f-number corresponding to the largest effective aperture (F). N The super-resolution results of the modulation method on the USAF resolution plate imaging experiment. Detailed Implementation

[0028] like Figure 1 As shown, this invention is based on a variable aperture coded imaging system, which includes an imaging lens group 1, a variable aperture 2, an image acquisition device 3, and an image processing module 6. Figure 2 As shown in Figure 3, the variable aperture consists of aperture blades 4 and an aperture mounting base 5. In this optical path structure, the aperture blades 4 of the variable aperture 2 are coaxial with the imaging lens group 1, and the variable aperture is fixed tightly against the imaging lens group 1. The image acquisition device 3 is located on the back focal plane of the imaging lens group 1. The imaging lens group 1 and the variable aperture 2 are relatively fixedly installed, and the variable aperture 2 and the image acquisition device 3 are mounted and fixed on the optical platform. When adjusting the variable aperture 2, the effective aperture size of the aperture blades 4 is adjusted, thereby controlling the incident light entering the imaging lens group 1 and achieving the effect of adjusting the optical transfer function. By adjusting the effective aperture size of the aperture blades 4 of the variable aperture 2, the incident light entering the imaging lens group 1 is controlled. The image acquisition device 3 acquires coded images with different aperture sizes and real-world low-resolution images corresponding to different coded patterns. The image processing module 6 is used to process the acquired images to obtain high-resolution images.

[0029] In a specific embodiment, the imaging lens group 1 has a maximum light transmission diameter of 37.5mm and an aperture number of F=2. The variable aperture 2 is an electrically operated variable aperture with an effective light transmission diameter range of 1.5mm-40mm. The image acquisition device is an Imaging Precision industrial camera DMK33UX226 with a pixel size of 1.85μm×1.85μm.

[0030] Figure 4 The variable aperture used in this invention refers to the aperture number F corresponding to the aperture number from the cutoff aperture F1 to the maximum effective aperture. N The modulated coded pattern, Figure 5 The aperture number ranges from the cutoff aperture F1 to the aperture number F corresponding to the maximum effective aperture. NThe modulated coded pattern corresponds to the generated optical transfer function. Figure 6 The variable aperture used in this invention ranges from the number of apertures corresponding to the smallest effective aperture to the number of apertures corresponding to the largest effective aperture. N The modulated coded pattern, Figure 7 The aperture number ranges from the smallest effective aperture to the largest effective aperture (F). N The modulated coded pattern corresponds to the generated optical transfer function. This invention achieves different coded patterns by adjusting the aperture blades of the variable aperture, thereby adjusting the aperture value.

[0031] In a specific embodiment, the cutoff aperture F1 is calculated when the optical resolution of the current imaging system matches the detector pixel size. At this time, the system's aperture F1 = 6.7; the aperture number when the effective aperture is at its maximum is F... N =2. Adjust the aperture number of the variable aperture, and record the aperture number F1 = 6.7 as the first coded pattern. Decrease the aperture number to F1 at equal aperture intervals ΔF = 0.313. N =2, with a total of 16 coded patterns. Compared to using... Figure 6 The encoding method shown adopts Figure 4 The encoding method shown contains more sub-pixel-level frequency domain aliasing information, avoiding the impact of diffraction-limited modulated sub-images on high-frequency detail recovery during the update process. Furthermore, the iteration process requires changing the optical transfer function (OTF1) corresponding to the cutoff aperture number F1 to the aperture number F at which the effective aperture is maximized. N The corresponding optical transfer function (OTF) N The inversion iteration is performed sequentially to avoid the transfer function at large aperture numbers truncating the high-frequency information of the image after the previous iteration at small aperture numbers, thereby reducing the quality of the reconstructed image.

[0032] Combination Figure 8 A super-resolution method based on variable aperture coding, the specific steps of which are as follows:

[0033] 1. Based on the wavelength λ of the current imaging system, the pixel size of the detector, and the required number of encoding steps N, calculate the cutoff aperture number F1 and the number of equally spaced reduction apertures ΔF.

[0034]

[0035]

[0036] 2. By adjusting the variable aperture blades, the aperture number is reduced from F1 = 6.7 to the aperture number F at equal intervals of ΔF = 0.313. N=2, record the coded patterns under N different effective light-transmitting apertures. k and the corresponding optical transfer function (OTF) k k = 1...N, N = 16

[0037]

[0038]

[0039] Among them, operators This indicates a two-dimensional convolution operation on matrices a and b. The operator conj{·} indicates a conjugate operation on matrix {·}. The operator max{·} indicates the maximum value among the elements of matrix {·}.

[0040] 3. Use an image acquisition device to capture N different coded patterns. k Corresponding real-world low-resolution images All real-shot low-resolution images Compared with the selected reference image standard Exposure compensation is applied. Specifically, the low-resolution real-shot image corresponding to the last coded pattern is selected. Using the baseline image, the gain coefficient matrix (gain) is obtained. k After obtaining exposure compensation, take low-resolution images.

[0041]

[0042]

[0043] In this context, the superscript cap represents the actual image captured by the image acquisition device, and the superscript cap,ec represents the image after exposure compensation.

[0044] 4. The aperture value after exposure compensation is F. N Real-world low-resolution images Upsampling is performed to obtain a high-resolution object. ini As the initial value for iteration, a Fourier transform is performed on it to obtain the initial high-resolution object spectrum. ini ;

[0045]

[0046] Object ini =Fourier{object ini}

[0047] Where Upsample{a,b,c} represents the upsampling operation, a represents the matrix to be upsampled, b represents the upsampling factor, and c represents the upsampling method, where the upsampling factor is Multiple and the upsampling method is Nearest interpolation; Fourier{·} represents the Fourier transform operation on the matrix {·}.

[0048] 5. In the iterth iteration, select the aperture number F. k The corresponding optical transfer function (OTF) k To capture the spectrum of a high-resolution object iter,k Information is used to obtain an estimated high-resolution image spectrum. After performing an inverse Fourier transform on the spectrum of the high-resolution image and downsampling, an estimated low-resolution image is obtained.

[0049]

[0050]

[0051]

[0052] Where Downsample{a,b,c} represents the downsampling operation, a represents the matrix to be downsampled, b represents the downsampling factor, and c represents the downsampling method, where the downsampling factor is... The downsampling method uses nearest neighbor interpolation; InverseFourier{·} indicates that Fourier transform operation is performed on matrix {·}, the superscript estimate represents the estimated high-resolution image, and the superscript estimate,LR represents the estimated low-resolution image;

[0053] 6. Estimate the low-resolution image Low-resolution images captured after exposure compensation Divide to obtain a low-resolution intensity coefficient matrix. For low-resolution intensity coefficient matrix Upsampling is performed to obtain a high-resolution intensity coefficient matrix. k The light intensity distribution used to update the estimated high-resolution image Get updated high-resolution images

[0054] High-resolution intensity coefficient matrix k Specifically:

[0055]

[0056]

[0057]

[0058]

[0059]

[0060] Where Downsample{a,b,c} represents the downsampling operation, a represents the matrix to be downsampled, b represents the downsampling factor, and c represents the downsampling method, where the downsampling factor is... The downsampling method uses nearest neighbor interpolation; InverseFourier{·} denotes the Fourier transform operation on matrix {·}. Represents the estimated high-resolution image. This represents the estimated low-resolution image. This represents the intensity coefficient matrix at low resolution.

[0061] Updated high-resolution images Specifically:

[0062]

[0063] The superscript "update" indicates the updated image;

[0064] 7. Update the spectrum of high-resolution images With the estimated high-resolution image spectrum Subtraction yields the frequency domain increment ΔImage iter,k The increment is then subjected to Wiener filtering to obtain the updated high-resolution object spectrum.

[0065]

[0066]

[0067]

[0068]

[0069] Where β represents the update step size, which is typically 0.5, and τ represents the regularization factor;

[0070] 8. Calculate the updated high-resolution object light intensity distribution The cost function ε of the low-resolution image after exposure compensation k ;

[0071]

[0072] Where, |{a}| 2This represents taking the square of the absolute value of matrix {a}.

[0073] 9. Let k = k + 1, and repeat steps 5-8 until N exposure compensations are applied to capture a low-resolution image. Iterate through all of them once, and set the N cost functions ε k The average cost function ε is obtained by summing and averaging.

[0074]

[0075] 10. Let iter = iter + 1, and proceed to the next iteration. Repeat steps 5 to 9 until the average cost function ε is less than the error threshold T, which is usually 0.001. At this point, the reconstructed high-resolution object converges.

[0076] ε≤T

[0077] To test the effectiveness of the present invention based on a variable aperture coded imaging system and an iterative reconstruction super-resolution method, a set of experiments were selected for illustration.

[0078] Figure 9 The image shows a low-resolution raw image of a USAF resolution target taken with an image acquisition device with a pixel size of 1.85μm × 1.85μm, before super-resolution reconstruction, along with its spectrum. Figure 10 To utilize the variable aperture coded imaging system of this invention, a variable aperture is used from the cutoff aperture number F1 to the aperture number F corresponding to the maximum effective aperture. N The modulation method affects the super-resolution results of the USAF resolution plate imaging experiment. Figure 11 To utilize the variable aperture coded imaging system of this invention, a variable aperture is employed, ranging from the aperture number corresponding to the minimum effective aperture to the aperture number F corresponding to the maximum effective aperture. N The modulation method was used to obtain super-resolution results for the USAF resolution plate imaging experiment. It can be seen that the acquired images have obvious pixelation problems, the edges of the target object are blurred, and the details of the target object cannot be identified. Compared with the reconstruction results of the -1 group of line pairs, compared with the method used... Figure 6 The encoding method adopts Figure 4 The encoding method, that is, from the cutoff aperture number F1 to the aperture number F corresponding to the maximum effective aperture. N The reconstruction results using the encoding method can better recover high-frequency detail information of objects during reconstruction, and the line pair recovery is smoother. In the super-resolution reconstructed images based on the variable aperture encoding imaging system of this invention, it can be seen that the super-resolution effect is improved by nearly 2 times, the details of the target object are significantly enhanced, and the super-resolution reconstruction effect is remarkable.

Claims

1. An imaging system based on variable aperture coding, characterized in that, include: An imaging lens group (1), a variable aperture (2), an image acquisition device (3), and an image processing module (6) are provided. The variable aperture (2) includes an aperture blade section (4) and an aperture mounting base (5). The aperture blade section (4) of the variable aperture (2) is coaxial with the imaging lens group (1), and the aperture blade section (4) is fixed to the imaging lens group (1) by the aperture mounting base (5). The image acquisition device (3) is located on the back focal plane of the imaging lens group (1). The imaging lens group (1) and the variable aperture (2) are fixedly installed relative to each other. The variable aperture (2) and the image acquisition device (3) are installed and fixed on an optical platform. The incident light entering the imaging lens group (1) is controlled by adjusting the effective aperture size of the aperture blade section (4) of the variable aperture (2). The image acquisition device (3) acquires coded images with different aperture sizes and real low-resolution images corresponding to different coded patterns. The image processing module (6) is used to process the acquired images to obtain high-resolution images. Adjusting the aperture blades of the variable aperture allows the effective aperture diameter of the variable aperture to change from the cutoff aperture number... With equal-spaced aperture numbers Reduce to f / 100 ,Record Encoding patterns at different aperture numbers and its corresponding optical transfer function , ,in ; The image processing module processes the coded image to obtain a high-resolution image using the following specific method: A. All real-shot low-resolution images With the selected reference image Perform exposure compensation to obtain the gain coefficient matrix. After obtaining exposure compensation, take low-resolution images. B. The aperture value after exposure compensation Real-world low-resolution images Upsampling is performed to obtain high-resolution objects As the initial value for iteration, a Fourier transform is performed on the initial value to obtain the initial high-resolution object spectrum. ; C. In the In the next iteration, the aperture number is selected. Corresponding optical transfer function To extract the spectrum of high-resolution objects Information is used to obtain an estimated high-resolution image spectrum. After performing an inverse Fourier transform on the spectrum of the high-resolution image and downsampling, an estimated low-resolution image is obtained. ; D. Estimating the low-resolution image Low-resolution images captured after exposure compensation Divide to obtain a low-resolution intensity coefficient matrix. For low-resolution intensity coefficient matrices Upsampling is performed to obtain a high-resolution intensity coefficient matrix. The light intensity distribution used to update the estimated high-resolution image To obtain updated high-resolution images ; E. Update the high-resolution image spectrum With the estimated high-resolution image spectrum Subtraction yields the frequency domain increment. The increment is then subjected to Wiener filtering to obtain the updated high-resolution object spectrum. ; F. Calculate the updated high-resolution object light intensity distribution Cost function of the low-resolution image after exposure compensation ; G. Order And repeat step CF until Low-resolution images taken after exposure compensation Iterate through them all once, and... Cost function The average cost function is obtained by summing and averaging. ; H. Order Repeat step CG until the average cost function is obtained. Less than the error threshold .

2. The variable aperture coded imaging system according to claim 1, characterized in that, Optical transfer function The formula for generating it is: in, This indicates the pixel size of the detector. This indicates the wavelength of the incident light. This indicates the optical resolution of the current imaging system and the detector pixel size. The aperture number used for matching is denoted as the cutoff aperture number. This indicates the aperture number at which the current imaging system reaches its maximum effective aperture; operator Indicates the matrix sum matrix Performing two-dimensional convolution operations, operators Represents a matrix Perform conjugate operations, operators Indicates taking the matrix The maximum value among the elements.

3. The variable aperture coded imaging system according to claim 1, characterized in that, Low-resolution images captured after exposure compensation Specifically: 。 4. The variable aperture coded imaging system according to claim 1, characterized in that, High-resolution intensity coefficient matrix The specific formula is: in, This indicates a downsampling operation. Represents the sampling matrix to be reduced, Indicates the downsampling factor, This indicates the downsampling method, where the downsampling factor is... Downsampling uses nearest neighbor interpolation. ; Represents a matrix Perform a Fourier transform operation. Represents the estimated high-resolution image. This represents the estimated low-resolution image. This represents the intensity coefficient matrix at low resolution.

5. The variable aperture coded imaging system according to claim 1, characterized in that, Frequency domain increment The specific formula is: in, This represents updating the estimated light intensity distribution of the image. This represents the image spectral increment.

6. The variable aperture coded imaging system according to claim 1, characterized in that, Updated high-resolution object spectrum Specifically: in, This represents the updated high-resolution object spectrum. This represents the high-resolution object spectrum from the previous iteration. Indicates the update step size. This represents the regularization factor.

7. The variable aperture coded imaging system according to claim 1, characterized in that, Updated high-resolution object light intensity distribution Cost function of the low-resolution image after exposure compensation Specifically: in, Represents a matrix Find the square of the absolute value.

8. A super-resolution method based on variable aperture coding, characterized in that, The specific steps are as follows: (1) Based on the wavelength of the current imaging system and the pixel size of the detector and the number of codes required Calculate the cutoff aperture number and equal intervals to reduce the aperture number ; (2) Adjust the aperture blades of the variable aperture, and reduce the aperture blades by equal spacing to obtain the desired aperture number. Different coding patterns , record Serial number pattern and its corresponding optical transfer function ,in ; (3) Take pictures using an image acquisition device Different coded patterns Corresponding real-world low-resolution images All real-shot low-resolution images With the selected reference image Exposure compensation is applied; here, the last coded pattern is selected. Corresponding real-world low-resolution images As the reference image; obtain the gain coefficient matrix. After obtaining exposure compensation, low-resolution images were captured. ; (4) The aperture value after exposure compensation Real-world low-resolution images Upsampling is performed to obtain high-resolution objects As the initial value for iteration, a Fourier transform is performed on it to obtain the initial high-resolution object spectrum. ; (5) in the In the next iteration, the aperture number is selected. Corresponding optical transfer function To extract the spectrum of high-resolution objects Information is used to obtain an estimated high-resolution image spectrum. After performing an inverse Fourier transform on the spectrum of the high-resolution image and downsampling, an estimated low-resolution image is obtained. ; (6) Estimated low-resolution image Low-resolution images captured after exposure compensation Divide to obtain a low-resolution intensity coefficient matrix. For low-resolution intensity coefficient matrices Upsampling is performed to obtain a high-resolution intensity coefficient matrix. The light intensity distribution used to update the estimated high-resolution image , to obtain updated high-resolution images ; (7) Update the high-resolution image spectrum With the estimated high-resolution image spectrum Subtraction yields the frequency domain increment. The increment is then subjected to Wiener filtering to obtain the updated high-resolution object spectrum. ; (8) Calculate the updated high-resolution object light intensity distribution Cost function of the low-resolution image after exposure compensation ; (9) Order And repeat steps (5)-(8) until Low-resolution images taken after exposure compensation Iterate through them all once, and... Cost function The average cost function is obtained by summing and averaging. ; (10) Order Then proceed to the next iteration, repeating steps (5)-(9) until the average cost function is reached. Less than the error threshold .

Images