Research On Remote Imaging Via Fourier Ptychography

Resolution is one of the most important indicators in optical imaging.Due to the diffraction of light waves,the imaging resolution is mainly limited by the aperture size of the imaging lens,Therefore,the weight,size and cost of imaging systems are rapidly increasing when pursuing high-resolution imaging.Fourier Ptychographic Microscopy(FPM)is a coherent aperture synthesis imaging technique that has gradually developed in the past decade.It can achieve high magnification aperture synthesis of targets,which has very broad development prospects.This technique uses phase recovery to restore the phase information of small apertures,achieving the goal of digitally synthesizing large aperture images,and it has seen rapid development in the field of microscopy.However,Fourier Ptychographic Imaging still faces many problems in long-distance applications.Therefore,this dissertation conducts research on the model and method of long-distance Fourier Ptychographic Microscopy.The dissertation proposed a method of long-distance Fourier Ptychographic Microscopy based on divergent spherical wave illumination.This method uses divergent spherical waves to illuminate distant targets and utilizes the diffuse reflection characteristics of rough targets to destroy the influence of the secondary phase in the imaging process.The linear phase shift required for spectral shift is provided by the spherical wave moving on the target surface,achieving the scanning of the target spectrum.Compared to previous models,the divergent beam can expand the target illumination range,increasing the imaging field of view of Fourier Ptychographic Microscopy.By setting up an experimental system and conducting comparative experiments under the same conditions as previous models,the method obtained more than a 10-fold increase in the field of view.To verify the system’s long-distance imaging capability,we performed scanning synthesis on a 1m×0.7m target at a distance of 10 meters,achieving a 6-fold synthetic aperture capability and significantly improving the resolution.Second,we have studied the resolution criterion of the reflection coherent imaging system.When a reflective coherent imaging system captures images of rough targets,speckle is formed,resulting in a decrease in image quality and resolution.This dissertation introduces the mechanism and properties of speckle formation in reflective coherent imaging systems,based on the Rayleigh criterion saddle to peak ratio index and the object image superposition relationship of coherent imaging processes,a resolution criterion for reflective coherent imaging systems is proposed to quantitatively measure the system’s resolution improvement ability.We use the Fourier Ptychographic Imaging system to synthesize speckle images of different aperture sizes to validate the proposed resolution index.The experimental results show that the resolution of synthesized images under different apertures is consistent with the index.Then,we have researched a speckle suppression method based on dual-zone Gaussian for speckle images.By transforming the speckle noise into dual-zone Gaussian noise suitable for Gaussian denoising algorithms using a transformation function,and designed an unbiased inverse transformation method to reduce the intensity difference caused by the transformation process,This method can suppress the reduction in image resolution caused by speckle in reflective coherent imaging systems.When using the method presented in this dissertation to suppress speckle,the visual effect of the images is significantly improved,and the signal-to-noise ratio is increased by more than 2d B without causing resolution loss.The last,we have studied the low-resolution Fourier Ptychographic reconstruction algorithm,By establishing a generalized Fourier Ptychographic imaging model with pixel sampling,we can analyze the impact of the sampling process on data acquisition and image reconstruction.When the sampling rate of the system is too low,the speckle structure of the collected image will be damaged,resulting in a decrease in the quality and resolution of the reconstructed image.This dissertation proposes a new cost function that utilizes forward estimation of the target spectrum to balance the weight between estimated sample strength and collected sample strength,in order to avoid spectrum leakage and ambiguity caused by intensity based cost functions.Through simulations and experiments,we verify that the algorithm can compensate for data loss caused by the sampling process.At a sampling rate of0.75,high-quality reconstruction is achieved,reducing the data volume to one twenty-fifth of the original.