Full-focusing super-resolution imaging method based on deconvolution

Abstract

The invention discloses a deconvolution-based full-focusing super-resolution imaging method, and relates to the technical field of full-focusing super-resolution imaging, and the method specifically comprises the steps: 1, carrying out the data collection based on a full-matrix collection mode, and carrying out the initial image reconstruction through a full-focusing method, and 2, carrying out the reconstruction of an initial image through the analysis of a point spread function of an imaging system, the method comprises the following steps: step 1, establishing a physical convolution model to describe a fuzzy effect in an imaging process, and step 2, by taking minimization of an imaging error as a target and introducing sparse constraint, promoting generation of non-zero reaction only at a position where a defect actually exists in a reconstructed image and inhibiting background noise and artifacts. A point spread function modeling imaging process is introduced, sparse deconvolution solution is carried out on defect distribution by utilizing a fast iteration threshold method, and a super-resolution imaging scheme which has a physical basis and is high in robustness and calculation efficiency is provided for guided wave ultrasonic imaging.

Description

Technical Field

[0001] The present invention belongs to the technical field of full-focusing method super-resolution imaging, and particularly relates to a full-focusing method super-resolution imaging method based on deconvolution. Background Art

[0002] In many fields such as industrial production, aerospace, and building structures, the quality inspection of materials and components is of vital importance. The existence of defects may seriously affect the safety, reliability, and service life of the structure. Therefore, the development of high-precision and high-resolution non-destructive testing technologies has always been a research hotspot.

[0003] Ultrasonic array imaging technology is an important means in non-destructive testing and structural health monitoring. It relies on a multi-channel transducer array to collect internal reflection signals of the structure and restores the defect distribution through an imaging algorithm. Common imaging algorithms include delay and sum (DAS) and full-focusing method (TFM). Among them, TFM performs point-by-point focusing based on full matrix capture (FMC) data, has good imaging quality and applicability, and has become the mainstream method in array imaging.

[0004] However, limited by the system aperture and operating frequency, the resolution of traditional TFM is still limited by the Rayleigh criterion, and it is difficult to accurately identify multiple defects with sub-wavelength spacing. To overcome this limitation, researchers have tried to introduce subspace methods such as TR-MUSIC to achieve super-resolution, but it is extremely sensitive to noise, and the reconstruction process depends on the accurate estimation of the signal subspace dimension.

[0005] Therefore, there is an urgent need for a full-focusing method super-resolution imaging method based on deconvolution to provide a super-resolution imaging scheme with physical basis, strong robustness, and high computational efficiency for guided wave ultrasonic imaging. Summary of the Invention

[0006] The purpose of the present invention is to provide a full-focusing method super-resolution imaging method based on deconvolution, which is used to solve the technical problems in the prior art that it is difficult to find a physical basis, poor robustness, and low computational efficiency in guided wave ultrasonic imaging.

[0007] To achieve the above purpose, the present invention adopts the following technical solutions: A full-focusing method super-resolution imaging method based on deconvolution, comprising: Step 1: Perform data acquisition based on the full matrix capture mode and perform initial image reconstruction using the full-focusing method; Step 2: Analyze the point spread function of the imaging system and establish a physical convolution model to describe the blurring effect during the imaging process; Step 3: With the goal of minimizing the imaging error, by introducing a sparse constraint, promote the reconstruction image to produce non-zero responses only at the positions where defects actually exist, and suppress background noise and artifacts.

[0008] Furthermore, data acquisition is performed based on the full matrix acquisition mode. The specific method is as follows: According to the size, shape of the detection area and the expected detection accuracy, a limited number of sensors are arranged to cover the entire detection area. An array system containing multiple transducers is built based on all the arranged sensors. A narrowband pulse or a sine wave modulated by a Gaussian envelope is used as the excitation signal. The center frequency is set according to the thickness of the detection object and the target mode. The sampling frequency is greater than or equal to ten times the center frequency. The sampling time window is calculated according to the propagation distance and the wave velocity. The single excitation-reception signal is processed by multiple superposition averaging. The high-precision time synchronization technology is adopted. Each transducer in the array is used as the excitation source in turn. When a certain transducer is set as the excitation source, all the other transducers are used as the receiving ends at the same time. This operation is performed on each transducer in the array in turn until all transducers have completed the task of being an excitation source once, and finally the original full matrix acquisition data is formed.

[0009] Furthermore, the full focus method is adopted for initial image reconstruction. The specific method is as follows: The full focus method is used to process the original full matrix acquisition data to generate the initial defect image. According to the propagation path from the exciter to each pixel point in the imaging area and the receiver, combined with the group velocity of Lamb waves in the material and the defect position coordinates, when the position of the defect is aligned with the focus, the array elements will experience a time delay. Using the formula represents the time delay, where represents the coordinates of the defect position, represents the coordinates of the array element i, represents the group velocity of Lamb waves in the material. According to the propagation path lengths from each exciter and receiver in the sensor array to the target pixel point, combined with the propagation velocity of Lamb waves in the material, the propagation time delay corresponding to each excitation-reception combination is calculated. Subsequently, interpolation processing is performed on the original discrete sampling signal. For each pixel point, the received signal is adjusted according to the calculated propagation time delay. By delaying and aligning the signals of each channel and performing amplitude accumulation or energy accumulation, the amplitudes of each channel at the delay point are squared and then accumulated, which is equivalent to accumulating the energy response. The accumulated result is subjected to amplitude normalization processing, or the defect signal is highlighted by dynamic range limitation to suppress background stray noise, and the centralized focusing of the wave field information is completed.

[0010] Furthermore, by analyzing the point spread function of the imaging system, a physical convolution model is established to describe the blurring effect in the imaging process. The specific method is as follows: In experimental or simulation data, set the distance between the center of the scatterer and the nearest other scatterer as d. When the distance satisfies d ≥ 2PSF, it is determined that the scatterer satisfies isolation. Set the amplitude at the center of the scatterer as a. When the amplitude satisfies a ≥ 3bac, it is determined that the scatterer satisfies intensity. Select an isolated and intense single scattered echo region. Taking the center of this region as the origin, intercept a small surrounding imaging sub-region. The amplitude or energy distribution within this sub-region is directly used as the numerical expression of the point spread. Based on the array geometric parameters, wave velocity information, and excitation signal characteristics, the spatial response function of the system is theoretically derived to obtain the expected point spread function pattern. The point spread function is fitted using a two-dimensional Gaussian distribution or an Airy disk function. The PSF matrix obtained through the fitting formula is the complete response model of the imaging system for a point source. The convolution relationship is established between the PSF and the actual observed image.

[0011] Furthermore, the point spread function is fitted using a two-dimensional Gaussian distribution or an Airy disk function. The specific method is as follows: Using the formula to fit the point spread function, where represents the coordinate of the center position of the isolated point scatterer, A is the amplitude normalization constant of represents the control of the diffusion width, represents the spatial coordinate of the local coordinate system plane established with the center of the isolated point scatterer as the origin.

[0012] Furthermore, with the goal of minimizing the imaging error, by introducing a sparse constraint, it is promoted that only non-zero responses are generated at the positions where actual defects exist in the reconstructed image. The specific method is as follows: Regard the blurred image generated by the initial full focus as the result of the convolution of the true defect distribution and the system point spread function, and superimpose noise interference. With the goal of minimizing the imaging error, construct an objective function, introduce a sparsity constraint, that is, sum the absolute values of all pixel amplitudes, and add this sparse term to the total objective function to obtain a point source localization deconvolution algorithm based on the accelerated gradient projection method, denoted as the fast iterative shrinkage threshold algorithm. Based on the currently estimated defect image, calculate the gradient direction of the convolution error, and perform iterative updates with step size control along the gradient direction to reduce the reconstruction error. Apply a shrinkage operation to the updated image to suppress small amplitude noise to zero and only retain significant defect responses. Introduce a momentum acceleration mechanism to improve the convergence speed. After several iterations, until the reconstruction error converges to a preset threshold or reaches the maximum number of iterations, output the final high-resolution defect image.

[0013] Furthermore, with the goal of minimizing the imaging error, construct an objective function. The specific method is as follows: Using the formula represents the objective function, where represents the target imaging, represents the image generated from the original data, represents the 2D Fourier transform, represents the 2D inverse Fourier transform, represents the Hadamard product, where D is the first-order difference matrix, represents the non-negativity constraint, using the formula to represent, where λ represents the weight parameter of the sparse regularization term.

[0014] Furthermore, the fast iterative shrinkage threshold algorithm, the specific method is as follows: Using the formula represents the fast iterative shrinkage threshold algorithm, where is the data consistency term, is the sparse regularization term. The weight parameter λ of the sparse regularization term is used to trade off and adjust between the imaging error and the sparsity requirement, balance the reconstruction accuracy and the sparsity degree. FISTA approximates the optimal solution by combining the shrinkage threshold operation and momentum acceleration after each gradient descent step.

[0015] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are as follows: 1. The present invention extracts the diffusion effects of the array aperture, propagation mode, and system bandwidth on imaging through the imaging response of isolated point scatterers, establishes a PSF physical convolution model, provides a mathematical basis for super-resolution reconstruction, and uses a two-dimensional Gaussian distribution or Airy disk function to fit the PSF to accurately describe the system spatial response, solving the problem that the blur effect is difficult to quantify in traditional methods; 2. The present invention models the TFM blurred image as the convolution of the real defect and the PSF plus noise, constructs an objective function through the least square criterion to realize the quantitative optimization of the imaging error, introduces a first-order difference matrix to constrain the spatial smoothness, avoids the over-oscillation of the reconstructed image, improves the accuracy of defect localization, realizes the spatial sparsity of the defect image by sparse regularization, only retains the significant defect response, suppresses background noise and artifacts, and combines momentum acceleration and soft threshold processing through the FISTA algorithm. While maintaining sparsity, the iteration speed is improved compared with the traditional gradient descent method, shortening the calculation time. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1Shows a step diagram of a super-resolution imaging method based on deconvolution full focusing method; Figure 2 Shows a schematic diagram of the Lamb wave group velocity dispersion curve in the present invention; Figure 3 Shows a schematic diagram for verifying the implementation results of the present invention. Specific implementation manner

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Such as Figure 1 、 Figure 2 、 Figure 3 As shown, a super-resolution imaging method based on deconvolution full focusing method specifically includes the following steps: Step 1: Perform data acquisition based on the full matrix acquisition mode, and use the full focusing method to reconstruct the initial image.

[0020] According to the size, shape of the detection area and the expected detection accuracy, arrange a limited number of sensors to ensure that these sensors can cover the entire detection area. Based on all the arranged sensors, build an array system containing multiple transducers. Use a narrowband pulse or a sine wave modulated by a Gaussian envelope as the excitation signal. The center frequency is set in the range of 100 - 500 kHz according to the thickness of the detection object and the target mode (such as A0 or S0 mode). The sampling frequency is greater than or equal to ten times the center frequency. The sampling time window is calculated according to the propagation distance and the wave speed. The single excitation-reception signal is processed by multiple superimposed averaging to improve the signal-to-noise ratio. Use a high-precision time synchronization technology, such as GPS synchronization or atomic clock synchronization, to accurately synchronize the excitation and reception times of each transducer. Take each transducer in the array as the excitation source in turn. When a certain transducer is set as the excitation source, all the remaining transducers act as the receiving end at the same time. The excitation source transducer emits a signal, and the receiving end transducer records the reflected signal. Perform this operation on each transducer in the array in turn until all transducers have completed the task of acting as the excitation source once, and finally form the original full matrix acquisition data; After the data acquisition is completed, use the full focusing method to process the original full matrix acquisition (FMC) data to generate an initial defect image. According to the propagation path from the exciter to the receiver to each pixel point in the imaging area, combined with the group velocity of Lamb waves in the material and the defect position coordinates, when the position of the defect is aligned with the focus, the array elements will experience a time delay. The formula for the distance relative to the focus is as follows: ; wherein, represents the coordinates of the defect position, represents the coordinates of array element i, represents the group velocity of Lamb waves in the material. The group velocity is obtained from the theoretical value of the dispersion curve. According to the propagation path lengths from each actuator and receiver in the sensor array to the target pixel point, combined with the propagation velocity of Lamb waves in the material, the propagation time delay corresponding to each pair of actuator-receiver combinations is calculated. The time delay is obtained by dividing the geometric path by the wave velocity.

[0021] Subsequently, high-precision interpolation processing (such as using linear interpolation, cubic spline interpolation, or sinc interpolation) is performed on the original discrete sampling signal. For each pixel point, the received signal is adjusted according to the calculated propagation time delay. By delaying and aligning the signals of each channel and performing amplitude accumulation or energy accumulation. For example: if the calculated propagation time delay of a certain channel is Δt, then the signal of this channel is shifted to the right by Δt (if Δt is positive) or to the left by |Δt| (if Δt is negative) on the time axis. The instantaneous amplitudes of each channel after delay alignment are extracted, and then directly summed to obtain the amplitude response of this pixel point. The amplitudes of each channel at the delay point are squared and then accumulated, which is equivalent to accumulating the energy response. The accumulated result is subjected to amplitude normalization processing, or the defect signal is highlighted by dynamic range limitation to suppress background stray noise, and the concentration and focusing of wave field information are completed.

[0022] Step 2: By analyzing the point spread function of the imaging system, a physical convolution model is established to describe the blurring effect in the imaging process.

[0023] By analyzing the response of an isolated point scatterer (such as a tiny reflection source artificially introduced in a defect-free area or a preset reference defect) in the imaging area in the TFM imaging result, the spatial expansion form of this point scatterer under the imaging system is obtained. The response of this isolated point shows an energy diffusion characteristic due to the influence of the array aperture, propagation mode, and system bandwidth during the imaging process. The point spread function of the imaging system is obtained by extracting this characteristic. The specific method includes: In experimental or simulation data, when the distance between the center of the scatterer and the nearest other scatterer is set to d, and the distance satisfies d≥2PSF, it is determined that this scatterer satisfies the isolation property, where PSF is the theoretical resolution of the system (such as the full width at half maximum FWHM), ensuring that there is no interference source around the target scatterer. When the amplitude of the scatterer center is set to a and the amplitude satisfies a≥3bac, it is determined that this scatterer satisfies the strong property; Select an isolated and strong single scattered echo region. Taking the center of this region as the origin, intercept a small surrounding imaging sub-region. The amplitude or energy distribution within this sub-region is directly used as the numerical expression of the point spread. Based on the array geometric parameters (such as sensor spacing, aperture size), wave speed information (such as the group velocity of each mode of Lamb wave), and excitation signal characteristics (center frequency and bandwidth), the spatial response function of the system is derived theoretically to obtain the expected point spread function pattern. The point spread function is fitted using a two-dimensional Gaussian distribution or an Airy disk function: ; Among them, represents the central position coordinates of the isolated point scatterer, A is the amplitude normalization constant of, represents the control diffusion width, represents the spatial coordinates of the plane of the local coordinate system established with the center of the isolated point scatterer as the origin.

[0024] The PSF matrix obtained through the fitting formula is the complete response model of the imaging system for the point source. By establishing a convolution relationship between the PSF and the actual observed image, the diffusion blur effect in the imaging process can be effectively described, providing a physical basis and calculation basis for subsequent sparse deconvolution to solve the defect image.

[0025] Step 3: With the goal of minimizing the imaging error, by introducing sparse constraints, promote non-zero responses only at the positions where actual defects exist in the reconstructed image, and suppress background noise and artifacts.

[0026] Regard the blurred image generated by the initial total focusing method (TFM) as the result of the convolution of the true defect distribution and the system point spread function (PSF), and superimpose noise interference. With the goal of minimizing the imaging error, construct an objective function. This objective function is used to measure the difference between the currently estimated defect image after convolution with the PSF and the observed blurred TFM image. The imaging error adopts the least squares criterion, that is, minimizing the Euclidean distance between the predicted image and the observed image. The specific formula of the objective function is as follows: ; Among them, represents the target imaging, represents the image generated from the original data, represents the 2D Fourier transform, represents the 2D inverse Fourier transform, represents the Hadamard product, D is the first-order difference matrix, represents the non-negativity constraint, using the formula represents, λ represents the weight parameter of the sparse regularization term.

[0027] To promote the reconstructed image to exhibit sparse characteristics in space, that is, most pixel values are close to zero and only respond at positions where defects actually exist, the present invention introduces a sparsity constraint, that is, sums the absolute values of all pixel amplitudes and adds this sparse term to the total objective function to obtain a point source localization deconvolution algorithm based on the accelerated gradient projection method, denoted as the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA). The specific formula of FISTA is: ; where, is the data consistency term, is the sparse regularization term. The weight parameter λ of the sparse regularization term is used to balance between the imaging error and the sparsity requirement, balancing the reconstruction accuracy and the sparsity degree. The weight value can be set empirically according to the actual noise level or defect size, or optimized by methods such as cross-validation. FISTA can quickly approach the optimal solution while maintaining the sparsity feature by combining the shrinkage threshold operation and momentum acceleration after each gradient descent step.

[0028] The specific implementation steps of FISTA are as follows: Based on the currently estimated defect image, calculate the gradient direction of the convolution error, perform iterative updates with step size control along the gradient direction to reduce the reconstruction error, apply a shrinkage operation (i.e., soft thresholding) to the updated image to suppress small-amplitude noise to zero and only retain significant defect responses, introduce a momentum acceleration mechanism so that each iteration not only depends on the current step size but also combines the information of the previous iteration to improve the convergence speed. After several iterations, until the reconstruction error converges to a preset threshold or reaches the maximum number of iterations, output the final high-resolution defect image.

[0029] Through this sparse deconvolution process, it is possible to effectively strip off the blur and noise interference caused by the system diffusion effect and restore a defect image close to the true distribution, thereby achieving super-resolution imaging beyond the diffraction limit.

[0030] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

[0031] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only the specific embodiments. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A super-resolution imaging method based on deconvolution full focus method, characterized in that Including: Step 1: Perform data acquisition based on the full matrix acquisition mode and perform initial image reconstruction using the full focus method; Step 2: By analyzing the point spread function of the imaging system, establish a physical convolution model to describe the blurring effect during the imaging process; Step 3: With the goal of minimizing the imaging error, by introducing sparse constraints, promote non-zero responses only at the positions where defects actually exist in the reconstructed image, and suppress background noise and artifacts.

2. The super-resolution imaging method based on deconvolution full focusing method according to claim 1, wherein Perform data acquisition based on the full matrix acquisition mode. The specific method is as follows: According to the size, shape of the detection area and the expected detection accuracy, arrange a limited number of sensors to cover the entire detection area. Based on all the arranged sensors, build an array system containing multiple transducers. Use a narrowband pulse or a sine wave modulated by a Gaussian envelope as the excitation signal. The center frequency is set according to the thickness of the detection object and the target mode. The sampling frequency is greater than or equal to ten times the center frequency. The sampling time window is calculated according to the propagation distance and the wave speed. The single excitation-reception signal is processed by multiple superimposed averaging. Adopt high-precision time synchronization technology. Take each transducer in the array as the excitation source in turn. When a certain transducer is set as the excitation source, all the other transducers act as the receiving end at the same time. Perform this operation on each transducer in the array in turn until all transducers have completed the task of being the excitation source once. Finally, form the original full matrix acquisition data.

3. The super-resolution imaging method based on deconvolution full focusing method according to claim 1, characterized in that, Perform initial image reconstruction using the full focus method. The specific method is as follows: The full focusing method is used to process the original full matrix acquisition data to generate an initial defect image. According to the propagation paths from the exciters to the receivers to each pixel point in the imaging area, combined with the group velocity of Lamb waves in the material and the defect position coordinates, when the position of the defect is aligned with the focus, the array elements will experience a time delay. Using the formula represents the time delay, where represents the coordinates of the defect position, represents the coordinates of array element i, represents the group velocity of Lamb waves in the material. According to the propagation path lengths from each exciter and receiver in the sensor array to the target pixel point, combined with the propagation speed of Lamb waves in the material, calculate the propagation time delay corresponding to each excitation-reception combination. Subsequently, perform interpolation processing on the original discrete sampling signal, adjust the signals received by each pixel point according to the calculated propagation time delay, and by delaying and aligning the signals of each channel and performing amplitude accumulation or energy accumulation, square the amplitudes of each channel at the delay point and then accumulate them, which is equivalent to accumulating the energy response. Perform amplitude normalization processing on the accumulated result, or highlight the defect signal by dynamic range limitation and suppress background stray noise to complete the centralized focusing of wave field information.

4. A super-resolution imaging method based on deconvolution full-focus method according to claim 1, characterized in that By analyzing the point spread function of the imaging system, establish a physical convolution model to describe the blurring effect during the imaging process. The specific method is as follows: In experimental or simulation data, set the distance between the center of the scatterer and the nearest other scatterer as d. When the distance satisfies d≥2PSF, it is judged that the scatterer satisfies isolation. Set the amplitude of the scatterer center as a. When the amplitude satisfies a≥3bac, it is judged that the scatterer satisfies intensity. Select an isolated and intense single scatter echo region. Take the center of this region as the origin and intercept a small surrounding imaging sub-region. The amplitude or energy distribution within this sub-region is directly used as the numerical expression of the point spread. Based on the array geometric parameters, wave speed information and excitation signal characteristics, theoretically deduce the spatial response function of the system to obtain the expected point spread function pattern. Use a two-dimensional Gaussian distribution or an Airy disk function to fit the point spread function. The PSF matrix obtained through the fitting formula is the complete response model of the imaging system for the point source. Establish a convolution relationship between the PSF and the actual observed image.

5. A super-resolution imaging method based on deconvolution full focusing method according to claim 4, characterized in that, Use a two-dimensional Gaussian distribution or an Airy disk function to fit the point spread function. The specific method is as follows: Using the formula to fit the point spread function, where represents the central position coordinates of the isolated point scatterer, A is the amplitude normalization constant of represents the control diffusion width, represents the spatial coordinates of the plane of the local coordinate system established with the center of the isolated point scatterer as the origin.

6. The super-resolution imaging method based on deconvolution full focusing method according to claim 1, wherein, With the goal of minimizing the imaging error, by introducing sparse constraints, promote non-zero responses only at the positions where defects actually exist in the reconstructed image. The specific method is as follows: The blurred image generated by the initial full focus is regarded as the result of the convolution of the true defect distribution and the system point spread function, and noise interference is superimposed. With the goal of minimizing the imaging error, an objective function is constructed, and a sparsity constraint is introduced, that is, the sum of the absolute values of all pixel amplitudes is calculated, and the sum value is denoted as the sparse term, and this sparse term is added to the total objective function to obtain a point source localization deconvolution algorithm based on the accelerated gradient projection method, denoted as the fast iterative shrinkage threshold algorithm. Based on the currently estimated defect image, the gradient direction of the convolution error is calculated, and iterative updates with step size control are performed along the gradient direction to reduce the reconstruction error. A shrinkage operation is applied to the updated image to suppress small-amplitude noise to zero and only retain significant defect responses. A momentum acceleration mechanism is introduced to improve the convergence speed. After several iterations, until the reconstruction error converges to a preset threshold or reaches the maximum number of iterations, the final high-resolution defect image is output.

7. A super-resolution imaging method based on deconvolution full focusing method according to claim 6, characterized in that With the goal of minimizing the imaging error, an objective function is constructed, and the specific method is as follows: Using the formula represents the objective function, where represents the target imaging, represents the image generated from the original data, represents the 2D Fourier transform, represents the 2D inverse Fourier transform, represents the Hadamard product, D is the first-order difference matrix, represents the non-negativity constraint. Using the formula represents, where λ represents the weight parameter of the sparse regularization term.

8. A super-resolution imaging method based on deconvolution full-focus method according to claim 6, characterized in that The fast iterative shrinkage threshold algorithm, and the specific method is as follows: Using the formula represents the fast iterative shrinkage threshold algorithm, where is the data consistency term is the sparse regularization term. The weight parameter λ of the sparse regularization term is used to make a trade-off adjustment between the imaging error and the sparsity requirement, balancing the reconstruction accuracy and the sparsity level. The fast iterative shrinkage threshold algorithm approaches the optimal solution by combining the shrinkage threshold operation and momentum acceleration after each gradient descent step.

Images