Far-field Diffuse Reflection Synthetic Aperture Super-Resolution Imaging Method Based on Stacked Reconstruction

Abstract

The present invention discloses a far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction to solve the technical problems in traditional imaging methods, such as limited imaging distance, large energy loss in spherical wave illumination mode, small reconstruction field of view, and low-quality imaging caused by noise of diffuse reflection objects. The present invention uses plane waves obtained by lens group conversion to perform reflective imaging on the target. By moving a two-axis precision displacement stage to drive the camera to collect the backscattered light information under the sub-apertures at corresponding positions, and then reconstructing a high-resolution image through the Fourier ptychography algorithm, and introducing total variation regularization for speckle denoising. The method proposed by the present invention can effectively improve the far-field detection distance, and introduce a denoising algorithm during the reconstruction process to alleviate the influence of speckle noise on the reconstructed image, and greatly improve the reconstruction quality.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical synthetic aperture imaging, and particularly relates to a far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction. Background Technique

[0002] In recent years, computational imaging has developed rapidly in the field of optical imaging, especially in far-field detection and remote sensing applications. Wider, clearer, and more accurate imaging results have always been the goals pursued in optical imaging. However, the resolution of traditional optical imaging systems highly depends on the aperture size of the imaging lens, and the manufacturing and processing of large-aperture lenses require excessive production costs. Therefore, how to achieve large field of view and high resolution under low-cost conditions is a key issue concerned in the field of far-field detection.

[0003] In order to break through the optical diffraction limit existing in the above traditional optical system, several research groups in the field of optical imaging have conducted relevant research. Among them, some scholars were inspired by the synthetic aperture technology in radar imaging and attempted high-resolution imaging by mainly synthesizing the aperture in the constructed optical imaging system. However, since the size of the aperture synthesis is based on the number of reflectors, the resolution improvement effect cannot be increased by more than 2 times.

[0004] In 2013, the team of Guoan Zheng from the University of Connecticut in the United States proposed the Fourier ptychography technique, which combines phase retrieval and synthetic aperture techniques and applies them in the field of microscopic imaging, obtaining high-resolution results for microorganisms such as cells and fluorescence imaging. However, the Fourier ptychography system used by Zheng et al. employed an LED array as the illumination source. Since the working distance of the microscopic platform is short and the imaging target is small, it can be approximately regarded as a partially coherent imaging system. In real far-field applications, a laser is needed to replace the LED array to meet the requirements of the imaging system for the energy and coherence of the light source. At the visible light wavelength scale, most objects are diffuse reflection objects with rough surfaces, and such rough surfaces are composed of a large number of randomly distributed surface elements. When a laser is used to detect a rough object, the light rays from the surface elements will be superimposed and interfere at the same image point, forming a speckle phenomenon. These speckles overlap with the image and will distort the retrieved result. Therefore, in far-field imaging, although the speckles contain information about the target to be guessed, during the high-resolution reconstruction process, the speckle noise will affect the reconstruction result. In subsequent work, Zheng et al. obtained high-resolution imaging results for fluorescence imaging with two turbid layers. The speckle effect formed by the first turbid layer is similar to the speckles formed by diffuse reflection objects in macroscopic imaging. Through the principle of imaging through the turbid layer, the high-frequency information of the object to be measured is transferred to the speckles generated by the turbid layer, and then the spectral information is solved iteratively. This method achieved a resolution improvement of up to 13 times, with the fluorescent sample sandwiched between the two turbid layers. An unshaped light beam is used to illuminate the turbid layer, and an unknown pattern is generated over a wide field of view on the target plane. Then, by tilting the input wavefront and scanning the unknown speckle pattern by moving the second turbid layer, the corresponding low-resolution fluorescence image is captured through the turbid layer, which provides an idea for how to obtain high-resolution and large field of view in macroscopic imaging. At the same time, in the field of macroscopic Fourier ptychography, J.S Holloway from Rice University applied Fourier ptychography in the active imaging mode and achieved super-resolution results for reflective macroscopic Fourier ptychography. To achieve active imaging, this technique obtains a high-resolution diffuse reflection image by collecting the reflected light of the object to be measured. This method scans the sub-aperture spectrum by translating the aperture position, and finally, the spectral information of the high-resolution object to be measured is retrieved through the Fourier ptychography algorithm. However, since this method also requires the light source to provide a spherical wave and uses the reflected light converging at the camera to obtain the spectral information after the light beam reaches the object, its imaging distance can only be about 1 m, still unable to meet the application scenarios of far-field imaging. And due to the diffusion problem of the spherical wave, even if one wants to increase the imaging distance, the laser power needs to be increased geometrically, and there are many factors affecting the imaging quality of the reconstructed image. Summary of the Invention

[0005] To solve the above technical deficiencies in the prior art, the present invention proposes a far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction.

[0006] The technical solution for achieving the object of the present invention is: a far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction, and the specific steps are as follows:

[0007] Step 1: Use the far-field diffuse reflection synthetic aperture super-resolution imaging system to form backscattered light containing laser speckles on the surface of the object to be measured. The backscattered light is converted into a spherical wave, and the spectral diagram formed at the camera is intercepted by the camera aperture to obtain a single low-resolution speckle image.

[0008] Step 2: Use a two-axis displacement stage to move on a square array to achieve aperture scanning, and linearly collect the low-resolution speckle images under the sub-apertures corresponding to the object to be measured row by row until the sampling of all the sample points in the entire square array is completed.

[0009] Step 3: Accumulate and average the intensity values of the low-resolution speckle image sequence collected after scanning, and expand the averaged image to the size of an ideal high-resolution image by bicubic interpolation as the initial high-resolution image.

[0010] Step 4: Update the spectral information of the high-resolution image by Fourier ptychography reconstruction for the low-resolution speckle images under different apertures, and then obtain the high-resolution image with low speckle noise according to the total variation regularization iteration.

[0011] Preferably, the far-field diffuse reflection synthetic aperture super-resolution imaging system includes a laser, a positive lens 1, a positive lens 2, a positive lens 3, a positive lens 4, a positive lens 5, an imaging lens, a camera, and a two-axis precision displacement stage. The object to be measured is on the far-field optical axis. The laser, the positive lens 1, the positive lens 2, and the positive lens 3 are on the same optical axis, and the positive lens 4, the positive lens 5, and the imaging lens are on the same optical axis. The beam emitted by the laser is a spherical wave, which is modulated by the positive lens 1, the positive lens 2, and the positive lens 3 to form an ideal plane wave to irradiate the object to be measured. When the reflected light carrying the target information passes through the positive lens 4 and the positive lens 5 after diffuse reflection, the plane is modulated and converted into a spherical wave. The convergence point between the positive lens 4 and the positive lens 5 serves as a new point light source carrying the target information. The imaging lens is installed on the camera to receive the converted spherical wave information, and the camera is installed on the two-axis precision displacement stage.

[0012] Preferably, the spectrum O Init of the initial high-resolution image in the x,y coordinates is expressed as:

[0013]

[0014] where I iis the intensity distribution of the i-th image in the low-resolution sequence diagram, and k is the number of low-resolution sequence diagrams participating in the initialization of the high-resolution image spectrum. is the two-dimensional Fourier transform.

[0015] Preferably, the spectral information of the high-resolution image is updated by Fourier ptychography reconstruction of the low-resolution speckle images under different apertures, and the high-resolution image with low speckle noise is obtained by iterating according to the total variation regularization. The specific steps are as follows:

[0016] Step 4.1: Update the complex amplitude in the spectral domain using the intensity of the i-th image collected by the camera, and replace the updated complex amplitude at the sub-aperture position corresponding to the initial high-resolution spectrum until the spectral replacement of all sub-images is completed;

[0017] Step 4.2: Introduce the total variation regularization method to constrain the laser speckle noise caused by the rough surface of the object and obtain a high-quality reconstructed image.

[0018] Step 4.3: Calculate the error function of the current iteration number. If the error function is greater than the pre-set value, repeat steps 4.1 to 4.2 until the error function converges to the pre-set value to obtain the final high-resolution denoised image.

[0019] Preferably, the update formula for the complex amplitude in the spectral domain in step 4.1 is:

[0020]

[0021] In the formula, is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera, and the superscript c represents actual shooting. is the complex amplitude distribution under the current aperture in the spectrogram, and the superscript e represents that the current spectrum and complex amplitude distribution are estimated from the high-resolution spectrum.

[0022] Preferably, the specific steps of introducing the total variation regularization method to constrain the laser speckle noise caused by the rough surface of the object and obtain a high-quality reconstructed image are as follows:

[0023] Using the total variation regularization method, the image speckle noise problem is transformed into a problem of solving an optimization problem, and the following functional is specifically solved:

[0024]

[0025] Among them, b is the image with speckle noise, u is the denoised image, A is the unit matrix linear blur operator, λ is the regularization parameter, E(u) is the energy function, and the denoised image u is obtained when taking the minimum (i.e., taking min). F(Au, b) is the data fidelity term, which ensures that the solution of the problem will not differ too much from the initial value and preserves the important information in the image. Represents the gradient information of the image, As the total variation constraint regularization term, let The regularization term is expressed as:

[0026]

[0027] where x and y represent the horizontal and vertical directions of the two-dimensional image respectively, For the denoised image u k The differentials in two directions, and the subscript k is the pixel index of the image.

[0028] Due to the gradient Being non-differentiable at zero, introduce the variable α as a constant, and the solution for the denoised image can be expressed as:

[0029]

[0030] where:

[0031]

[0032] Use the second-order central difference to obtain the denoised image in each coordinate system, introduce the time variable t, and obtain the overall denoised image for specific solution:

[0033]

[0034] Obtain the denoised image u k , that is, the optimal solution of the total variation regularization denoising objective function;

[0035] Determine the termination threshold threshold, when ||u k - u k－1 || / ||u k－1 || < threshold, the total variation regularization terminates, and at this time

[0036] Preferably, the error function of the current iteration number is:

[0037]

[0038] where, Is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera, and the superscript c represents actual shooting, Is the complex amplitude distribution under the current aperture in the spectrogram, and the superscript e represents that the current spectrogram and complex amplitude distribution are estimated from the high-resolution spectrum.

[0039] Compared with the prior art, the significant advantages of the present invention are as follows: In view of the problems existing in traditional active high-resolution imaging, such as short detection distance and large energy loss in spherical wave illumination mode, the present invention uses the generated plane wave to detect objects, which can greatly improve the high-resolution detection distance of target objects; the reflective Fourier ptychography method applied to rough surface objects used in the present invention can perform far-field high-resolution imaging on diffuse reflection objects and achieve far-field detection of external targets of the system; on the basis of reconstructing high-resolution images by the Fourier ptychography method, the present invention introduces total variation regularization to further denoise the high-resolution images, alleviate the influence of speckle noise on the reconstructed images, and greatly improve the reconstruction quality.

[0040] Other features and advantages of the present invention will be described in the following specification, and some of them will become obvious from the specification or be understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained through the structures specifically pointed out in the written specification, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings are only for the purpose of illustrating specific embodiments and are not considered as limitations of the present invention. Throughout the drawings, the same reference signs represent the same components.

[0042] Figure 1 It is a schematic diagram of the corresponding optical path for a far-field diffuse reflection imaging system.

[0043] Figure 2 It is a flowchart of the Fourier ptychography reconstruction algorithm and total variation regularization denoising.

[0044] Figure 3 They are schematic diagrams of the spectra and reconstruction results with overlap ratios of 40%, 60%, and 72% respectively.

[0045] Figure 4 It is a combined diagram of a sequence of low-resolution speckle images generated by a resolution target in a reflective imaging system under simulation conditions, the spectrum diagram reconstructed by the Fourier ptychography imaging method of the present invention using the sequence diagram, and the low-resolution speckle images at different positions under sub-apertures.

[0046] Figure 5 It is a low-resolution speckle image of a resolution target collected by an actual camera, and the imaging results after reconstruction by the Fourier ptychography imaging method and denoising by total variation regularization in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, those of ordinary skill in the art can imagine various implementation manners of the present invention. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as all of the present invention or as a limitation or restriction on the technical solution of the present invention. On the contrary, the purpose of providing these embodiments is to enable those skilled in the art to understand the present invention more thoroughly. The preferred embodiments of the present invention will be specifically described below with reference to the drawings, where the drawings form a part of this application and are used together with the embodiments of the present invention to illustrate the innovative concept of the present invention.

[0048] The inventive concept is a far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction, and the specific steps are as follows:

[0049] Step 1: Turn on the laser in the far-field diffuse reflection synthetic aperture super-resolution imaging system. After adjusting the system, align the plane wave generated by the lens group with the object to be measured. At this time, the camera obtains the low-resolution result under the current aperture.

[0050] In a further embodiment, as Figure 1 shown, the far-field diffuse reflection synthetic aperture super-resolution imaging system includes a laser, positive lens 1, positive lens 2, positive lens 3, positive lens 4, positive lens 5, an imaging lens, a camera, and a two-axis precision displacement stage, and the target to be measured is on the far-field optical axis. The laser, positive lens 1, positive lens 2, and positive lens 3 are on the same optical axis, and the positive lens 4, positive lens 5, and imaging lens are on the same optical axis. The beam emitted by the laser is a spherical wave, which is modulated by positive lens 1, positive lens 2, and positive lens 3 to form an ideal plane wave. The plane wave propagates approximately parallel during the propagation process and irradiates the target to be measured. When the reflected light carrying the target information passes through positive lens 4 and positive lens 5 again after diffuse reflection, the plane is modulated and converted into a spherical wave. At this time, the convergence point between positive lens 4 and positive lens 5 serves as a new point light source carrying the target information. The imaging lens is installed on the camera to receive the converted spherical wave information, and the camera is installed on the two-axis precision displacement stage.

[0051] In a further embodiment, the difference between the imaging system used in the present invention and the conventional macroscopic Fourier ptychography is as follows: In the conventional macroscopic Fourier ptychography, when the light source emits a spherical wave and passes through the transmissive object, the information of the object is carried forward and finally collected by the camera. In the whole process, the transmitted light of the spherical wave passing through the object is used to obtain the imaging result. Since the object to be measured is inside the overall optical path, this scheme cannot be applied in the fields of remote sensing, far-field detection, etc. In the present invention, the laser is placed on the side of the camera, and the light beam is emitted at a small angle with respect to the straight line direction of the camera shooting the object to be measured. After passing through the lens group, a plane wave is formed to keep the beam divergence size approximately the same, and it propagates to the far field. After that, it is diffusely reflected by the rough object. The information carried by the backscattered light is recorded in the camera on the same horizontal plane as the light source to form the result of active imaging. And using plane wave illumination to ensure the imaging distance and imaging range, the working distance of active imaging can be extended to the far-field range.

[0052] Step 2: The camera uses a two-axis precision displacement stage to move on the square array to achieve aperture scanning, linearly and sequentially collect the low-resolution speckle images corresponding to the sub-apertures of the object to be measured until the sampling of all the sample points of the entire square array is completed;

[0053] In a further embodiment, the two-axis precision displacement stage moves from the upper left corner point of the square array and scans to the right. When the acquisition of this row is completed, it returns to the first aperture point of the next row on the left and continues the acquisition until the acquisition process of the entire square array is completed, so as to achieve fine scanning. The number of low-resolution speckle images collected here is 441. In order to obtain a high-resolution reconstructed image, the acquisition overlap rate between apertures during scanning is not less than 60% to constrain the iterative update process. The reconstruction results with different overlap rates are as Figure 2 shown.

[0054] Step 3: Accumulate and average the intensity values of the low-resolution speckle image sequence collected after scanning, and then expand the averaged image to the size of the ideal high-resolution image by bicubic interpolation as the initial high-resolution image;

[0055] In a further embodiment, the 441 sub-aperture low-resolution images collected in Step 2 are intensity-accumulated and averaged, and the obtained intensity-averaged image is interpolated and enlarged to the required reconstructed image size. The spectrum of this image is used as the initial high-resolution spectrum. The initial high-resolution spectrum O Init (x,y) can be expressed as:

[0056]

[0057] where I i is the intensity distribution of the i-th image in the low-resolution sequence diagram, k is the number of low-resolution sequence diagrams participating in the initialization of the high-resolution image spectrum, is a two-dimensional Fourier transform.

[0058] Step 4: Update the spectral information of the high-resolution image by Fourier ptychography reconstruction of the low-resolution speckle images at different apertures, and then obtain the high-resolution image with low speckle noise by iteratively using total variation regularization. The specific steps are as follows:

[0059] Step 4.1: Update the complex amplitude in the spectral domain using the intensity of the i-th image collected by the camera. The specific formula is as follows:

[0060]

[0061] In the formula, is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera. The superscript c represents actual shooting. is the complex amplitude distribution at the current aperture in the spectrogram. The superscript e represents that the current spectral and complex amplitude distributions are estimated from the high-resolution spectrum. The updated complex amplitude is The superscript u represents the result updated by this method. Replace the result spectrum at the sub-aperture position corresponding to O Init (x, y) until the spectral replacement of all sub-images is completed;

[0062] Step 4.2: After completing the spectral replacement, introduce the total variation regularization method to constrain the laser speckle noise caused by the rough surface of the object and obtain a high-quality reconstructed image. Using the total variation regularization method, transform the image speckle noise problem into an optimization problem. The specific solution is as follows for the functional:

[0063]

[0064] where b is the image with speckle noise, u is the denoised image, A is the unit matrix linear blur operator, λ is the regularization parameter, E(u) is the energy function. When taking the minimum value (i.e., taking min), the denoised image u is obtained. F(Au, b) is the data fidelity term, which ensures that the solution of the problem does not differ too much from the initial value and preserves the important information in the image. represents the gradient information of the image. As the total variation constraint regularization term, let The regularization term is expressed as:

[0065]

[0066] where x and y represent the horizontal and vertical directions of the two-dimensional image respectively. is the differential of the denoised image u k in two directions, and the subscript k is the pixel index of the image.

[0067] Since the gradient It is not differentiable at zero. Introduce the variable α as a constant. The denoised image obtained by solving can be expressed as:

[0068]

[0069] Where:

[0070]

[0071] Use the second-order central difference to obtain the denoised image in each coordinate system. Introduce the time variable t to obtain the overall denoised image for specific solution:

[0072]

[0073] Obtain the denoised image u k , that is, the optimal solution of the total variation regularization denoising objective function;

[0074] Determine the termination threshold threshold. When ||u k -u k-1 || / ||u k-1 || < threshold, the total variation regularization terminates. At this time

[0075] Preferably, the error function of the current iteration is:

[0076]

[0077] Where, is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera. The superscript c represents actual shooting, is the complex amplitude distribution under the current aperture in the spectrogram. The superscript e represents that the current spectrogram and complex amplitude distribution are estimated from the high-resolution spectrum.

[0078] If the error function is greater than the pre-set value, repeat steps 4.1 - 4.2 until the error function converges to the pre-set value. At this time, the entire iterative update is completed, and the final high-resolution denoised image is obtained.

[0079] The main difference in the imaging results between the present invention and the previous macroscopic Fourier ptychography is that for the stitching result of the square array diagram and the reconstructed spectrum of the traditional macroscopic Fourier ptychography, when the detected object exists in the system, far-field imaging beyond 1 m cannot be truly achieved. Moreover, since the transmitted object does not consider the diffuse reflection situation, there are differences between bright and dark fields in the low-resolution sub-images obtained by aperture scanning, and the resulting high-resolution result after reconstruction does not consider the influence of speckles and corresponding noises. However, the super-resolution imaging method for far-field diffuse reflection objects proposed in the present invention can be applied to detection targets outside the system. In the simulation case, the low-resolution speckle result of the USAF resolution target as a far-field diffuse reflection object is as shown in Figure 4 Figure a. Due to the existence of speckles, the spectral information is scattered and there is no distinction between bright and dark fields. Using the reflective Fourier ptychography algorithm for reconstruction, the sub-apertures corresponding to the low-resolution speckle images are stitched in the frequency domain to obtain a wide-range synthetic aperture spectral information. The larger the spectrum corresponding to the synthetic aperture, the more target information can be obtained. After completing the spectral stitching, total variation regularization is introduced to denoise the reconstructed image, and a high-resolution target imaging result with suppressed speckles is obtained.

[0080] To verify the effectiveness of the present invention, an actual far-field detection experiment was carried out on a standard USAF resolution target using the system of the present invention. The specific experiment is as follows:

[0081] In the experiment, a USAF resolution target with a rough surface after chrome plating was used as the target to be measured. It was placed 3.6 m away, and the laser plane wave generated by the system of the present invention was used to irradiate the target to be measured to generate reflected light. By moving the two-axis precision displacement stage, a low-resolution sequence distribution diagram was obtained. The imaging result of the central small aperture in the case of object diffuse reflection imaging is as shown in Figure 4 Figure. At this time, due to the simultaneous existence of diffraction limitation and speckle noise, the imaging quality is further reduced, and only a low-resolution speckle image with a granular distribution can be seen. The F# of the corresponding aperture is 20, and the resolvable line pairs are 1.78 p / mm. After collecting the sub-images, these distribution diagrams are used as the data source for iterative reconstruction. The result after reconstruction using the reflective system Fourier ptychography reconstruction algorithm in the present invention and speckle denoising using total variation regularization is as shown in Figure 4 Figure. The resolvable line pairs in the recognizable area of the resolution target are increased to 7.13 p / mm, and the corresponding aperture F# is 5, indicating that the reconstructed result reaches a 4-fold super-resolution result.

[0082] The above is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0083] It should be understood that, in order to streamline the present invention and assist those skilled in the art in understanding various aspects of the present invention, in the above description of the exemplary embodiments of the present invention, various features of the present invention are sometimes described in a single embodiment or with reference to a single figure. However, the present invention should not be construed as meaning that the features included in the exemplary embodiments are all essential technical features of the claims of this patent.

[0084] It should be understood that the modules, units, components, etc. included in the device of an embodiment of the present invention can be adaptively changed to be arranged in a device different from that embodiment. Different modules, units or components included in the device of the embodiment can be combined into one module, unit or component, or they can be divided into multiple sub-modules, sub-units or sub-components.

Claims

1. A far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction, characterized in that The specific steps are as follows: Step 1: Use a far-field diffuse reflection synthetic aperture super-resolution imaging system to form backward scattered light containing laser speckles on the surface of the object to be measured. The backward scattered light is converted into a spherical wave, and the spectral diagram formed at the camera is intercepted by the camera aperture to obtain a single low-resolution speckle image. The far-field diffuse reflection synthetic aperture super-resolution imaging system includes a laser, positive lens 1, positive lens 2, positive lens 3, positive lens 4, positive lens 5, an imaging lens, a camera, and a two-axis precision displacement stage. The object to be measured is on the far-field optical axis. The laser, positive lens 1, positive lens 2, and positive lens 3 are on the same optical axis. The positive lens 4, positive lens 5, and imaging lens are on the same optical axis. The beam emitted by the laser is a spherical wave, which is modulated by positive lens 1, positive lens 2, and positive lens 3 to form an ideal plane wave to irradiate the object to be measured. After diffuse reflection, the reflected light carrying the target information is modulated and converted into a spherical wave when passing through positive lens 4 and positive lens 5. The convergence point between positive lens 4 and positive lens 5 is used as a new point light source carrying the target information. The imaging lens is installed on the camera to receive the converted spherical wave information, and the camera is installed on the two-axis precision displacement stage; Step 2: Use the two-axis precision displacement stage to move on a square array to achieve aperture scanning, linearly acquire the low-resolution speckle images under the sub-apertures corresponding to the object to be measured row by row until the sampling of all the sample points in the entire square array is completed; Step 3: Accumulate and average the intensity values of the low-resolution speckle image sequence collected after scanning, and expand the averaged image to the size of an ideal high-resolution image through bicubic interpolation as the initial high-resolution image; Step 4: Update the spectral information of the high-resolution image by Fourier ptychography reconstruction of the low-resolution speckle images under different apertures, and then obtain a high-resolution image with low speckle noise according to the total variation regularization iteration. The specific steps are as follows: Step 4.1: Update the complex amplitude in the spectral domain using the intensity of the i-th image collected by the camera, and replace the updated complex amplitude at the sub-aperture positions corresponding to the initial high-resolution spectrum until the spectral replacement of all sub-images is completed; in the x,y coordinates, the update formula for the complex amplitude is as follows: In the formula, is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera. The superscript c represents actual shooting. is the complex amplitude distribution under the current aperture in the spectrogram. The superscript e indicates that the current spectrogram and complex amplitude distribution are estimated from the high-resolution spectrum. Step 4.2: Introduce the total variation regularization method to constrain the laser speckle noise caused by the rough surface of the object and obtain a high-quality reconstructed image. The specific steps are as follows: Using the total variation regularization method, transform the image speckle noise problem into an optimization problem, and specifically solve the following functional: Among them, b is the image with speckle noise, u is the denoised image, A is the unit matrix linear blur operator, λ is the regularization parameter, E(u) is the energy function, and the denoised image u is obtained when it takes the minimum value (i.e., taking min). F(Au, b) is the data fidelity term, which ensures that the solution of the problem does not differ too much from the initial value and preserves the important information in the image; represents the gradient information of the image, As the total variation constraint regularization term, let The regularization term is expressed as: where x and y represent the horizontal and vertical directions of the two-dimensional image respectively, for the denoised image u k the differentials in the two directions, and the subscript k is the pixel point index of the image; Due to the gradient not being differentiable at the zero point, introduce the variable α as a constant, and the denoised image obtained by solving can be expressed as: Where: Use second-order central difference to obtain the denoised image in each coordinate system, introduce the time variable t, and obtain the overall denoised image to be specifically solved: Obtain the denoised image u k , which is the optimal solution of the total variation regularization denoising objective function; Determine the termination threshold threshold. When ||u k - u k-1 || / ||u k-1 || < threshold, the total variation regularization terminates. At this time Step 4.3: Calculate the error function of the current iteration. If the error function is greater than the pre-set value, repeat Steps 4.1 - 4.2 until the error function converges to the pre-set value to obtain the final high-resolution denoised image.

2. The far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction according to claim 1, characterized in that Initial high-resolution image spectrum O in the x, y coordinates Init (x, y) is expressed as: Among them I i is the intensity distribution of the i-th image in the low-resolution sequence diagram, and k is the number of low-resolution sequence diagrams participating in the initialization of the high-resolution image spectrum. is the two-dimensional Fourier transform.

3. The far-field diffuse reflection synthetic aperture super-resolution imaging method based on stacked reconstruction according to claim 1, characterized in that The error function ε for the current k-th iteration k is as follows: Among them, is the intensity distribution of the i-th low-resolution speckle image actually recorded by the camera, and the superscript c represents actual shooting. is the complex amplitude distribution under the current aperture in the spectrogram, and the superscript e indicates that the current spectrogram and complex amplitude distribution are estimated from the high-resolution spectrum.

Images