1.IntroductionHigh resolution (HR), large field of view (FOV), and quantitative phase imaging have long been the goals of optical microscopy. However, conventional microscopy struggles to combine these objectives, which significantly limits its practical application.1,2 Fourier ptychographic microscopy (FPM) is a representative new computational imaging technique that employs angularly varying illumination to obtain a series of low-resolution (LR) images corresponding to different spatial frequencies and frequency domain stitching to broaden the spectrum of the sample, computationally breaking the diffraction limit of the objective lens.3 The technique combines the principles of phase recovery4,5 and synthetic aperture6,7 to recover the intensity and phase information of a sample under a low numerical aperture (NA) objective with a large FOV, increasing the spatial bandwidth product of the microscope by at least an order of magnitude.8 However, there are several inevitable imaging process aberrations far from the edge of the central FOV, such as defocus aberrations caused by the uneven thickness and inaccurate focus of the sample, spherical and coma aberrations of the microscope objective, and the image dispersion of the off-axis LED beam refracted by the lens, which cannot be focused at one point.9–11 A hybrid of these aberrations has serious effects on the reconstruction quality of the FPM. The FPM proposed in 2013 by Zheng et al. has received extensive attention and is researched worldwide.3 FPM reconstruction frameworks combined with correction methods have been proposed to further optimize the performance of FPM and reduce the impact of wavefront aberrations on the quality of FPM reconstruction. The existing optimization schemes fall into two main categories: one is to pre-estimate the imaging system aberrations and correct them in the phase recovery algorithm, while the other is to optimize the reconstruction algorithm directly to correct the imaging system aberrations without pre-estimation. To estimate the imaging system aberrations in advance, Zheng et al.9 proposed a generalized pattern search algorithm based on phase retrieval to measure aberration coefficients, followed by aberration correction of the acquired images using a deconvolution method, which was highly sensitive to component movement in the imaging system. It inaccurately estimated higher-order aberrations and had a high computational cost. Furthermore, attempts to combine the aberration correction algorithm with the reconstruction algorithm were made. Ou et al. proposed an embedded optical pupil function recovery algorithm (EPRY-FPM),10 which could recover the optical pupil function of the imaging system using the acquired LR image dataset alone without estimating the imaging system aberration in advance. EPRY-FPM can better correct aberrations, and it is currently considered a mainstream method. However, the recovery results were not satisfactory when hybrid aberrations were encountered owing to the influence of spectrum and pupil function update strategy because EPRY-FPM used the traditional FPM reconstruction framework, which was combined with an alternating iterative reconstruction algorithm (AP algorithm).3 Subsequently, some researchers used deep learning methods to optimize aberrations. Sun et al. proposed a neural network framework with optical pupil function recovery,12 where the complex amplitude image of a sample was obtained by minimizing the loss function. However, the framework did not consider the effect of spatially varying aberrations, and its generalization was low considering the requirement of a long time to retrain for different samples and the high demand for hardware processing. Therefore, the development of a correction method for the existing FPM imaging system that does not increase the computational complexity has a fast convergence rate, and is more robust to aberrations is necessary. An AA algorithm was used in our previous study to optimize the spectral update strategy of the reconstruction algorithm for an improved FPM reconstruction framework.13 The algorithm improved the convergence speed of the reconstruction algorithm and its robustness to noise, but the impact of various potential imaging system aberrations on the reconstruction quality was not considered. Therefore, this study proposes an FPM aberration correction reconstruction algorithm (AA-P) based on an improved phase recovery strategy. The AA-P optimizes the updating process of the sample spectral and optical pupil functions by adaptively selecting a modulation factor from the acquired LR image dataset. Compared to conventional aberration correction methods such as EPRY-FPM, AA-P can reconstruct better results with faster convergence, and the recovered optical pupil function can better characterize the imaging system aberrations. Furthermore, the effectiveness and advantages of the AA-P are verified through simulations and experiments on biological sample datasets. The remainder of this paper is organized as follows. Section 2 briefly outlines the basic model of the conventional FPM, the effect of aberrations in an optical system on the FPM, and the specific procedure of the AA-P proposed in this study. Section 3 verifies the effectiveness of the AA-P through simulations and real experiments, while Sec. 4 summarizes and discusses the results of this study. 2.Materials and Methods2.1.FPM Forward ModelFigure 1 shows the model and imaging process of a typical FPM system. The model mainly consists of a programmable LED array with a low NA objective ( objective is used as an example in this study) for a general microscopic imaging system. The imaging process of FPM includes image acquisition and image reconstruction. During image acquisition, the spectral information of the sample beyond the bandwidth of the objective lens is moved within the NA of the objective lens and positioned on the back focal plane of the objective lens (i.e., the spectral plane of the objective lens) by sequentially lighting the LED array (shown in the lower right corner of Fig. 1) to provide angularly varying plane waves for illumination. As shown in the upper right corner of Fig. 1, the dashed circles are: several of the sub-spectral information corresponding to samples illuminated by plane waves of different angles, while the corresponding LR images are captured at the camera port, after which the completed LR image dataset is iteratively reconstructed using a phase recovery algorithm.14 The conventional microscopic illumination in the FPM image acquisition process is replaced by an array of programmable LEDs. Each lit LED provides an illumination sample of wave vector , which varies with an angle . for the ’th LED can be expressed as , where denotes the illumination wavelength of light in air and denote the angle of incidence of the ’th LED. The image sensor (CMOS) is limited by a sampling rate. Hence, to only capture the light wave intensity, the captured intensity image can be expressed as where represents the coordinates in a two-dimensional space domain, which signifies the position of each LED unit in the array; represents the spatial coordinates in the two-dimensional Fourier domain; represents the spectrum of the sample under illumination corresponding to the ’th LED; and and represent the Fourier transform and inverse transform, respectively. represents the optical pupil function of the imaging system, which is regarded as the coherence transfer function in the FPM system, and is regarded as an ideal circular domain function without aberration in the ordinary FPM algorithm. Specifically, it can be expressed as where NA is the numerical aperture of the objective lens and denotes the amplitude of the optical pupil function. According to Eqs. (1) and (2), corresponds to the low-pass filtering of light field information of a sample when a LED unit irradiates the sample through the FPM imaging system. Thus, a series of LR images captured during image acquisition represent the subspectral information of the sample in different spectral regions. The captured LR images are used as the null domain amplitude constraint, while is used as the frequency constraint in the Fourier domain to expand the spectral range of the sample by stitching.2.2.Aberrations in FPM Imaging Systems and Their EffectsThe forward imaging model of FPM in an ideal aberration-free case was discussed in Sec. 2.1; however, in practical experimental platforms and applications, various aberrations are inevitably introduced into the imaging system. It is well known that linear proportionality exists between aberration effect and imaging FOV,10 which is especially important for FPM imaging systems with a large FOV and HR imaging. Since aberrations are present and most microscope objectives are in circular domains with symmetric rotational properties, the optical pupil function of the microscope objective can no longer be expressed by Eq. (2). Hence, is expressed as11 where is the amplitude of the light pupil function shown in Eq. (2), and are the Zernike polynomials and their coefficients corresponding to the aberration, respectively.A real FPM imaging system has usually several types of aberrations, such as: tilt, defocus, spherical, primary and secondary astigmatism, primary coma, primary trefoil, primary tetrafoil, etc. If they exist simultaneously, they are called hybrid aberrations. In this study, we simulate the above types of aberrations using the first 15 terms of the Zernike polynomial, as shown in Table 1, where is the radial distance and is the azimuthal angle, and both variables are expressed in polar coordinates. Figure 2(a) shows a two-dimensional presentation corresponding to each aberration type in Table 1. In this study, each coefficient is simulated using values in the simulation of hybrid aberrations.12,15,16 Table 1First 15 terms of Zernike and the assumed coefficients of its simulation.
Figures 2(b)–2(d) illustrate the effects of hybrid aberrations on the captured LR images and reconstruction quality. The captured LR images showed several streaks because of the pupil aberration of the objective, as shown in Fig. 2(c). The appearance of phase residuals in Eq. (3) affected the spectral distribution of the sample to some extent, which eventually affected the reconstruction quality (here, the reconstruction is performed using AA algorithm, which has been proven to be generally stable and advantageous for FPM reconstruction), as shown in Fig. 2(d).13 Subsequent corrections and comparisons of various algorithms are further discussed in Sec. 3.1. Therefore, it is necessary to improve the robustness and quality of the FPM reconstruction algorithm for hybrid aberrations conditions. 2.3.AA-P Aberration Correction Algorithm Reconstruction FrameworkBased on the potential aberrations of the FPM imaging model and system, an FPM forward imaging model using the AA algorithm was implemented in this study. However, the reconstruction framework of the AA algorithm set the pupil function to an ideal value and the inaccuracy of the previously fixed pupil function could seriously degrade the reconstruction quality when hybrid aberrations occurred in the system. Therefore, the AA-P algorithm is proposed. The specific process and steps are discussed further, as shown in Fig. 3.
Thus, the spectrum function is updated as where is the regularization parameter, set to for updating sample spectrum function, and “*” denotes the complex conjugate operation. optimizes the process of spectrum updation to obtain better reconstruction quality and convergence speed, which is also used to update the sample spectrum of the AA algorithm. The optical pupil function in the previous case was kept constant throughout the iterative process, and the HR spectral information of the sample target subaperture was updated using the spectral difference before and after the intensity constraint. Similarly, the optical pupil function in this study is updated considering the aberrations present in the imaging system as where is set in the same manner as . is a constant used for optimizing the updated light pupil function. To obtain a better correction effect, in this study. Thus, the updating principles of the optical pupil function and the sample spectrum function are basically the same.
3.Experiments and Results3.1.Simulation and ResultsTo quantitatively analyze the effectiveness of the proposed AA-P algorithm for aberration correction and improving the quality of FPM reconstruction, the algorithm was first verified through a simulation, where the simulation parameters were consistent with the parameters of the subsequent experimental platform. For LED illumination, a green LED array placed 90.88 mm directly below the sample, having a 4-mm spacing between adjacent LED units was simulated with an incident wavelength of 531 nm. The simulated microscope objective was set to magnification, while the NA was set to 0.1, and the pixel size of the camera was . “Cameraman” and “Westconcordorthophoto” of size were used as the input HR amplitude and phase images, respectively, as shown in Figs. 4(a1) and 4(a2). Therefore, the synthesized is 0.49. In an actual optical system, it is not only affected by a single aberration but by a variety of hybrid aberrations. In this study, the first 15 Zernike polynomial coefficients are used to simulate the hybrid aberration of the imaging system with reference to the ideas in Sec. 2.2, as shown in Fig. 4(a3). The simulation generated 225 LR images with an overlap rate of 68.04% in the Fourier domain, and the convergence of all algorithms was guaranteed without any aberration.18 Then, four FPM reconstruction algorithms, AP, AA, EPRY-FPM, and the proposed AA-P, were iteratively reconstructed and compared in the simulated FPM imaging system with wavefront aberration, as shown in Fig. 4. The four algorithms were grouped into two categories, where the AP and AA algorithms were without an aberration correction function, while the EPRY-FPM and the proposed AA-P algorithms were with an aberration correction function. As shown in Figs. 4(b1) and 4(b2), the AP algorithm has the worst reconstruction quality and failed to perform normal convergence for multiple hybrid aberrations. Comparatively, the AA algorithm had an improved the reconstruction quality, owing to its superior performance, as shown in Figs. 4(c1) and 4(c2). However, the reconstruction quality of the AA algorithm was still unsatisfactory because the wavefront aberration repeatedly affected the reconstructed spectrum distribution, leading to a certain amount of crosstalk in the reconstructed intensity and phase information. The EPRY-FPM algorithm further improved the reconstruction quality, as shown in Figs. 4(d1) and 4(d2); however, it continued to suffer from the case of detail blurring with crosstalk between intensity and phase information.19 The proposed AA-P algorithm had the best reconstruction quality, as shown in Figs. 4(e1) and 4(e2), compared to the others and showed higher robustness to hybrid aberrations of the optical system owing to the improved optimization strategy of spectral and optical pupil function updates. Furthermore, by analyzing Table 1 and Figs. 4(a3), 4(d3), and 4(e3) for structural similarity index (SSIM) between the reconstructed phase image of optical pupil function and the simulated wavefront aberration, the proposed AA-P algorithm had a higher match with the SSIM than the EPRY-FPM algorithm. Thus, the phase image of the optical pupil function recovered by the proposed AA-P algorithm could better characterize the wavefront aberration of the FPM optical system. The convergence performance and reconstruction quality results of the four algorithms were further quantified and analyzed by comparing the root mean square error (RMSE) curves of the HR amplitude and phase intensity images reconstructed by the four algorithms under different iterations, and the RMSE values after iterative stable convergence were recorded in Table 2. As shown in Figs. 4(f) and 4(g), the AP algorithm was severely disturbed by the aberration, with the worst reconstruction quality, so it could not converge in normal iterations; the AA algorithm converged normally but its reconstruction quality still needs to be improved; although the EPRY-FPM algorithm corrected the aberration to some extent, comparatively, the AA-P algorithm converged faster and better reconstructed the HR amplitude and phase images, with an improved reconstruction quality of 82.6% and 54.6%, respectively. Table 2RMSE values of reconstructed intensity and phase for the four algorithms under iterative stable convergence conditions, and SSIM values of the phase of the optical pupil function compared to the simulated wavefront.
3.2.Real ExperimentThe effectiveness of the proposed AA-P algorithm was verified through simulations. Furthermore, the algorithm was validated on the experimental data of open-source biological samples provided by Zheng et al. to verify its performance.20 Experimental parameters in the experiment were set similar to that in the simulation experiment of Sec. 3.1. A biological sample of HE pathological tissue section was chosen, for which 225 LR images were captured under a green LED array illumination (central wavelength is 531 nm). Consistent with the simulation experiments, the LR image dataset was reconstructed in the presence of aberrations in an unknown imaging system. The LR image captured by the central LED array illumination was upsampled as the initial HR spectrum shown in Fig. 5(a), and the amplitude and phase images reconstructed by four FPM algorithms were compared. As shown in Figs. 5(b1)–5(b3), the reconstruction quality with the AP algorithm was the worst, with blurred and cloudy reconstructed amplitude and phase intensity image information, respectively. Figures 5(c1)–5(c3) show the HR images reconstructed using the AA algorithm. Comparatively, the clarity of the reconstructed amplitude intensity image by the AA algorithm was significantly improved, with distinct line contours but partially blended and some crosstalk in the recovery of phase information. Figures 5(d1)–5(d3) show the reconstruction results of the EPRY-FPM algorithm, where the phase image reconstruction quality was significantly improved. However, due to its optimization strategy, the lines of the phase image were not clear and distinct and the quality of the reconstructed amplitude image was inferior to that of the AA algorithm, with problems of low overall contrast and blurred local details. Conversely, the HR pathological tissue sections reconstructed by the proposed AA-P algorithm was identified in more detail, with clearly defined tissue lines and higher overall contrast, as shown in Figs. 5(e1)–5(e3). In addition, Figs. 5(d4) and 5(e4) show the fitting results of the phase of the optical pupil function recovered by the EPRY-FPM and AA-P algorithms, respectively. Combined with the simulation experiment results, the proposed AA-P algorithm can be used to further estimate the wavefront aberration of the FPM optical system accurately in real experiments. To further evaluate the imaging performance of the four algorithms, we compared the RMSE, number of iterations, final error, and run time of their reconstruction results were compared, and the results are listed in Table 3. The RMSE and iterative final error metrics show that the amplitude and phase images reconstructed using the AA-P algorithm had better quality. In addition, the AA-P algorithm had the least number of iterations and required less time, indicating a better convergence performance. Table 3Comparison of RMSE, number of iterations, final error, and run time of the AP, AA, EPRY-FPM, and AA-P algorithms.
To verify the stability of the AA-P algorithm for experimental data, additional biological samples were recovered from the FPM optical system. A reconstruction recovery on the kidney fibroblasts (MouseKidney) of a mouse was performed using the same experimental parameters used in the experiment of Sec. 3.2, as shown in Fig. 6. Raw image of a selected region is shown in Fig. 6(a). Figures 6(b) and 6(c) represent the intensity images reconstructed using an adaptive step reconstruction strategy21 (AS algorithm) and the AA algorithm, respectively, where the background was blended with the cells, and some distortions were observed when compared to Fig. 6(a), which illustrates that the presence of aberrations can have a serious impact on the fidelity of the FPM reconstructed images. Figure 6(d) shows the intensity image reconstructed using the EPRY-FPM algorithm, which effectively improved the problem of background turbidity, but the overall contrast was not very high. The cells appeared blurry and the reconstruction quality was still unsatisfactory. Comparatively, the reconstructed intensity image of the proposed AA-P algorithm, shown in Fig. 6(e), had a clearer background and cell separation with clean image details, and showed the best reconstruction quality. Furthermore, the curve trends of the normalized pixel values in the annotated regions of Figs. 6(b)–6(e) were compared and the results are shown in Fig. 6(f). The four colors represent the normalized curve trends at the annotation regions for the intensity images reconstructed using the four algorithms. On analyzing the curve trends, it can be concluded that the red curve has the highest upper and lower drops, with a highly obvious concavity and a high average value. Thus, the image reconstructed by the AA-P algorithm had a higher contrast, clearer cell outlines distinguished with more details, and a maximum fidelity to the original image. Thus, the feasibility of the proposed AA-P algorithm for quantitative phase imaging of different samples is further validated. Finally, we performed experiments under other optical systems, and we chose to perform reconstruction recovery on human blood smears. 225 LR images were taken using a central blue LED array under illumination (central wavelength of 475 nm), and other experimental parameters were consistent with previous experiments. The original images of the selected areas are shown in Fig. 7(a1). Similarly, we compared the reconstructed amplitude and phase results of the four algorithms (AP algorithm, AA algorithm, EPRY-FPM algorithm, and the AA-P algorithm proposed in this paper), and each algorithm was iterated to the convergence condition. Figures 7(b1) and 7(b2) show the reconstructed amplitude and phase images of the AP algorithm, which could not converge to obtain clear reconstruction results and produced serious distortion. Figures 7(c1), 7(c2), 7(d1), and 7(d2) are reconstructed using the AA algorithm and the EPRY-FPM algorithm. Although the amplitude reconstruction results have been greatly improved, the phase reconstruction results have lower contrast and the amplitude reconstruction still has some blurring. However, compared with the previous three algorithms, the amplitude and phase images reconstructed by the AA-P algorithm are better recovered, with clear cell contours and high phase contrast, without image distortion and crosstalk, which further verifies the stability and effectiveness of the AA-P algorithm in recovering different samples under different aberration conditions. Figure 7(f) shows the fitting results of the phase of the optical pupil function reconstructed by the AA-P method, which can reflect the wavefront aberration of the system, for which we performed the Zernike polynomial decomposition of the first 25 terms, as shown in Fig. 7(g). The analysis shows that the wavefront aberration of the system mainly comes from the tilt of modes 2 and 3, the defocus of mode 4, the astigmatism of modes 5 and 6, and the coma of modes 7 and 8, in addition to which the higher-order aberrations such as spherical, primary trefoil, and tetrafoil are still not negligible. 4.Conclusion and DiscussionIn this work, we propose an FPM aberration-correction reconstruction framework based on an improved phase-recovery strategy, referred to as the AA-P algorithm, which aims to improve the reconstruction quality of FPM. By optimizing the spectral function and optical pupil function update strategy, the algorithm effectively improves the iterative reconstruction quality and reduces the impact of hybrid wavefront aberration on the FPM reconstruction quality and occurrence of errors in the reconstruction process. The reconstruction quality of the proposed AA-P algorithm is better than the most advanced reconstruction algorithm (AA algorithm) and aberration correction method (EPRY-FPM), and had higher robustness to hybrid aberrations and convergence performance, as demonstrated in experiments with biological samples such as HE pathological tissue, MouseKidney and human blood smears. The strict requirement of wavefront aberration for FPM is further relaxed. In future, the AA-P algorithm can be used for more advanced application scenarios. For example, it is used for dynamic aberration correction of the observed live sample for high quality dynamic high-throughput quantitative phase imaging. This requires future work on improving the imaging efficiency of FPM. AcknowledgmentsThis work was supported by the National Natural Science Foundation of China (Grant No. 62075042), the Science and Technology Program of Guangdong Province (Grant No. X190311UZ190) and Foshan University 2022 Annual Student Academic Fund (Grant No. XSJJ202205KJB07). Code, Data, and Materials AvailabilityWe sincerely acknowledge the open source data set of the biological sample provided by Zheng et al.20 ReferencesG. Zheng et al.,
“Fourier ptychographic microscopy: a gigapixel superscope for biomedicine,”
Opt. Photonics News, 25
(4), 26
–33 https://doi.org/10.1364/OPN.25.4.000026
(2014).
Google Scholar
Y. Xiao et al.,
“High-speed Fourier ptychographic microscopy for quantitative phase imaging,”
Opt. Lett., 46
(19), 4785
–4788 https://doi.org/10.1364/OL.428731 OPLEDP 0146-9592
(2021).
Google Scholar
G. Zheng, R. Horstmeyer and C. Yang,
“Wide-field, high-resolution Fourier ptychographic microscopy,”
Nat. Photonics, 7
(9), 739
–745 https://doi.org/10.1038/nphoton.2013.187 NPAHBY 1749-4885
(2013).
Google Scholar
X. Ou et al.,
“Quantitative phase imaging via Fourier ptychographic microscopy,”
Opt. Lett., 38
(22), 4845
–4848 https://doi.org/10.1364/OL.38.004845 OPLEDP 0146-9592
(2013).
Google Scholar
K. Creath and G. Goldstein,
“Dynamic quantitative phase imaging for biological objects using a pixelated phase mask,”
Biomed. Opt. Express, 3
(11), 2866
–2880 https://doi.org/10.1364/BOE.3.002866 BOEICL 2156-7085
(2012).
Google Scholar
L. Granero et al.,
“Synthetic aperture superresolved microscopy in digital lensless Fourier holography by time and angular multiplexing of the object information,”
Appl. Opt., 49
(5), 845
–857 https://doi.org/10.1364/AO.49.000845 APOPAI 0003-6935
(2010).
Google Scholar
P. Song et al.,
“Synthetic aperture ptychography: coded sensor translation for joint spatial-Fourier bandwidth expansion,”
Photonics Res., 10
(7), 1624
–1632 https://doi.org/10.1364/PRJ.460549
(2022).
Google Scholar
A. Lohmann, R. Dorsch, D. Mendlovic, Z. Zalevsky and C. Ferreira,
“Space-bandwidth product of optical signals and systems,”
J. Opt. Soc. Am. A, 13
(3), 470
–473 https://doi.org/10.1364/JOSAA.13.000470 JOAOD6 0740-3232
(1996).
Google Scholar
G. Zheng et al.,
“Characterization of spatially varying aberrations for wide field-of-view microscopy,”
Opt. Express, 21
(13), 15131
–15143 https://doi.org/10.1364/OE.21.015131 OPEXFF 1094-4087
(2013).
Google Scholar
X. Ou, G. Zheng and C. Yang,
“Embedded pupil function recovery for Fourier ptychographic microscopy,”
Opt. Express, 22
(5), 4960
–4972 https://doi.org/10.1364/OE.22.004960 OPEXFF 1094-4087
(2014).
Google Scholar
C. Shen et al.,
“Computational aberration correction of VIS-NIR multispectral imaging microscopy based on Fourier ptychography,”
Opt. Express, 27
(18), 24923
–24937 https://doi.org/10.1364/OE.27.024923 OPEXFF 1094-4087
(2019).
Google Scholar
M. Sun et al.,
“Neural network model combined with pupil recovery for Fourier ptychographic microscopy,”
Opt. Express, 27
(17), 24161
–24174 https://doi.org/10.1364/OE.27.024161 OPEXFF 1094-4087
(2019).
Google Scholar
J. Luo et al.,
“Fast and stable Fourier ptychographic microscopy based on improved phase recovery strategy,”
Opt. Express, 30
(11), 18505
–18517 https://doi.org/10.1364/OE.454615 OPEXFF 1094-4087
(2022).
Google Scholar
P. Konda et al.,
“Fourier ptychography: current applications and future promises,”
Opt. Express, 28
(7), 9603
–9630 https://doi.org/10.1364/OE.386168 OPEXFF 1094-4087
(2020).
Google Scholar
J. Wu, F. Yang and L. Cao,
“Resolution enhancement of long-range imaging with sparse apertures,”
Opt. Lasers Eng., 155 107068 https://doi.org/10.1016/j.optlaseng.2022.107068 OLENDN 0143-8166
(2022).
Google Scholar
H. Gross et al., Handbook of Optical Systems, Volume 3: Aberration Theory and Correction of Optical Systems, Wiley-Vch(
(2007). Google Scholar
L. Hou et al.,
“Background-noise reduction for Fourier ptychographic microscopy based on an improved thresholding method,”
Curr. Opt. Photonics, 2
(2), 165
–171 https://doi.org/10.3807/COPP.2018.2.2.165
(2018).
Google Scholar
J. Sun et al.,
“Sampling criteria for Fourier ptychographic microscopy in object space and frequency space,”
Opt. Express, 24
(14), 15765
–15781 https://doi.org/10.1364/OE.24.015765 OPEXFF 1094-4087
(2016).
Google Scholar
Y. Zhang, W. Jiang and Q. Dai,
“Nonlinear optimization approach for Fourier ptychographic microscopy,”
Opt. Express, 23
(26), 33822
–33835 https://doi.org/10.1364/OE.23.033822 OPEXFF 1094-4087
(2015).
Google Scholar
G. Zheng et al.,
“Concept, implementations and applications of Fourier ptychography,”
Nat. Rev. Phys., 3 207
–223 https://doi.org/10.1038/s42254-021-00280-y
(2021).
Google Scholar
C. Zuo, J. Sun and Q. Chen,
“Adaptive step-size strategy for noise-robust Fourier ptychographic microscopy,”
Opt. Express, 24
(18), 20724
–20744 https://doi.org/10.1364/OE.24.020724 OPEXFF 1094-4087
(2016).
Google Scholar
BiographyRuofei Wu is currently working toward the ME degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His current research interests include Fourier ptychographic microscopy, computational microscopy imaging, and computer vision. Jiaxiong Luo is currently pursuing his ME degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interests include computer imaging and Fourier ptychographic microscopy. Jiancong Li is currently pursuing his ME degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interest is optical system design and telescope design. Hanbao Chen is currently pursuing his bachelor degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interests include Fourier ptychographic microscopy, structural design, and software interface design. Junrui Zhen is currently pursuing his bachelor degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interests include Fourier ptychographic microscopy and software interface design. Sicong Zhu is currently working toward his ME degree in School of Mechatronic Engineering and Automation, Foshan University, Foshan, China. His research interests include Fourier ptychographic microscopy. Zicong Luo is currently pursuing his ME degree at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interests include Fourier ptychographic microscopy and computer imaging. Yanxiong Wu received his PhD from the Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun, China, in 2015. He is currently a professor at the School of Physics and Optoelectronic Engineering, Foshan University, Foshan, China. His research interests include computational optical imaging, optical instrument design, and optical system design. |
CITATIONS
Cited by 5 scholarly publications.
Reconstruction algorithms
Image restoration
Pupil aberrations
Aberration correction
Imaging systems
Phase reconstruction
Lawrencium