| Traditional optical imaging systems are difficult to achieve both large field of view and high-resolution imaging at the same time.Therefore,the Fourier Ptychographic Imaging technique has gained much attention due to its advantages of balancing large field of view and high-resolution imaging,and has been applied in long distance imaging fields.It obtains a series of low-resolution images by moving the camera,and then performs frequency domain stitching and fusion to reconstruct high-resolution images that exceed the diffraction limit of the optical system.However,this technique is currently not mature enough,and there are still some problems in practical applications,such as noisy low-resolution images that result in poor reconstruction image quality,and camera position misalignment in the process of acquiring low-resolution images.This paper focuses on these problems and proposes corresponding solutions.The main research contents of this paper are as follows:A Fourier ptychographic imaging denoising algorithm based on L2 norm is proposed to solve the noise problem in Fourier ptychographic imaging.By solving the overall objective function,the frequency spectrum update method is improved from single aperture update to batch-style overall update,allowing all frequency spectrum apertures to participate in high-resolution reconstruction at the same time.In addition,L2 norm is introduced into the cost function as a constraint to increase the weight of the high-resolution frequency spectrum,avoiding the influence of the extremely small value of the high-resolution frequency spectrum in the reconstruction process and effectively suppressing noise and improving the quality of the reconstructed image.For Gaussian noise with a standard deviation of 0.01,compared with the traditional AP algorithm,the proposed algorithm reduces the MSE of the reconstructed amplitude image from 0.0028 to 0.0008,increases the PSNR from 21.8933 to 24.0646,and the SSIM from 0.8345 to 0.8691.When collecting low-resolution images with a moving camera,positional misalignment can lead to a decrease in image quality during image reconstruction.Therefore,an analysis of the positional misalignment model for Fourier ptychographic imaging technology was conducted,and a suitable positional misalignment model for macroscopic imaging systems was determined.To reduce the stringent requirements on scanning accuracy,a camera positional misalignment correction algorithm based on particle swarm optimization was proposed.The algorithm corrects the positional misalignment of each sub-aperture in the frequency domain,ultimately reconstructing a high-resolution image.To verify the performance of the algorithm,we conducted a simulation experiment using “Lena” and “Map” images as amplitude and phase images,respectively.When the low-resolution image set was noise-free,the proposed algorithm showed significant improvements over traditional algorithms,with the PSNR of the reconstructed amplitude image increasing from 16.2454 to 24.3670,and the SSIM increasing from 0.5718 to 0.9296.The sum of pixel difference between the uncalibrated spectral sub-aperture position and the actual position also decreased from1160 to 69.When the low-resolution image was contaminated with Gaussian noise with a standard deviation of 0.01,the proposed algorithm increased the PSNR of the reconstructed amplitude image from 15.2966 to 23.2073 and the SSIM from 0.4568 to0.7924,while the sum of the difference between the sub-aperture positions before and after calibration decreased from 1160 to 168 pixels.These results indicate that the proposed algorithm can effectively correct for positional misalignment even when the low-resolution image is noisy.The algorithm was also applied to real-world datasets,resulting in an increase in image resolution from 4.00 lp/mm to 5.04 lp/mm compared to the traditional Fourier ptychographic algorithm. |
PDF Full Text Request