Research On Image Preprocessing Techniques Needed For Long-distance Camera Scanning Fourier Ptychographic Imaging

Achieving super-resolution imaging is the constant pursuit in the optical imaging field,because more details of the images can be obtained.Fourier ptychographic imaging was first proposed as a super-resolution imaging method to overcome the tradeoff between the high resolution and the large field of in microscopic imaging.For long-distance imaging such as remote sensing and aerospace,because of the diffraction limit,the aperture of the lens must be increased to reduce the effect of diffraction blur to obtain high-resolution images,but this will inevitably increase the weight,complexity and cost of the imaging system.The computational imaging method is considered to solve the problem of limited diffraction.Therefore,fourier ptychographic imaging technology is used as a new image reconstruction technology in long-distance imaging,which includes synthetic aperture and phase recovery technology.Compared with microscopic imaging,the imaging distance has been expanded hundreds to thousands of times in long-distance fourier ptychographic imaging,and there will be interfered by many uncertain factors which are brought by system and the environment,these all factors greatly affect the original images collected and even influence the final quality of the reconstructed image.Therefore,this article mainly studies the preprocessing technology of long-distance camera scanning fourier ptychographic imaging,which with the space target imaging as the background,analyzes the image degradation factors in the imaging process,and suppresses these influence factors to improve the quality of the final image reconstructed.The main works of the paper is as follows:1.Taking the microscopic imaging as a guide,the imaging mechanism,mathematical model and the optical design of the fourier ptychographic microscopy imaging are analyzed;then it is extended to the long-distance camera scanning fourier ptychographic imaging.And the imaging system model,propagating process and reconstruction algorithm of the far-field fourier ptychographic imaging system are introduced in detail to realize the overall grasp of the long-distance camera scanning fourier ptychographic imaging;2.The possible image degradation factors in long-distance camera scanning fourier ptychographic imaging are analyzed in detail,and the corresponding solutions are put forward according to the specific factors.For the uneven brightness of the captured target image which caused by the strong directivity of the laser light source,the gamma transformation is selected as the final calibration algorithm because it is simpler and more effective after comparing two brightness calibration algorithms;For the dark current noise,Poisson noise,Gaussian noise and other noise brought by the experimental device and environment,this article adapts a dark field calibration method firstly to remove the dark current noise,and then a Poisson-Gaussian mixture noise model is proposed to estimate noise parameters and denoise accurately.So it exactly solves the problem which the high-frequency information is overwhelmed by noise;For the problem of image spatial position deviation caused by camera scanning,two registration algorithms are compared: image registration method based on gray information and sub-pixel registration method based on phase correlation.Finally,the sub-pixel registration method is selected because of the higher registration accuracy.And the sub-pixel registration method effectively solves the artifacts in the reconstructed image caused by the position error;For the motion blur and defocus blur caused by camera and object motion,this article compares two deblurring algorithms: Wiener filtering and Blind Richardson-Lucy algorithm.Finally,the Blind Richardson-Lucy algorithm with better deblurring effect is selected.And through simulation,it is verified that these preprocessing algorithms can effectively improve the resolution of the reconstructed image.Compared with the reconstructed image,which with Poisson parameter of 0.005,the Gaussian parameter of 0.0015,the horizontal displacement deviations of 0.52465,the vertical displacement deviations of0.8372 and the Gaussian blur,the resolution of the reconstructed image after preprocessing is increased by 4 times,and the signal-to-noise ratio is improved by10.56 d B.3.On the basis of the above-mentioned preprocessing method,certain constraints are imposed during the image reconstruction process and some calibration processing is performed to further improve the resolution of the final reconstructed image.It mainly includes the selection of cost function,regularization constraints,and aberration calibration;by comparing three different types of cost functions,the cost function based on amplitude with a more ideal reconstruction effect is finally selected and the total variational regularization constraint is imposed,so that the reconstruction algorithm is more robust to noise;then,the aberrations which may exist in the imaging system are calibrated by updating the pupil function.4.Completed the selection of experimental devices and the construction of the experimental system,and collected the image data acquisition.Finally,the preprocessing algorithm and the improved reconstruction algorithm proposed in this paper were used to reconstruct the experimental data,and the resolution was improved by 3.8 times at the imaging distance of 1.6m?...