Research On Reconstruction Algorithm Of Reconstructed Image Of Inline Holographic Microscope

Digital holography is a new three-dimensional imaging method,which has been taken intensive attentions in past decades due to the improvement of technology for computer and photoelectric sensors processing application.Digital holograms,differing traditional optics,are recorded and input by using photoelectric sensors,such as CMOS.The reconstruction and reproduction process of digital holography can be simulated by computer,where it stored,and also can realize the whole digitization from recording to reconstruction.The advantages of traditional holographic is not only suit for the viewing,non-contacting and phase information obtaining,but also has highly flexibility for computer digitization,accurate quantitative measurement,easy storage and easy processing.inline digital holography has been widely used due to its complicated physical optics,wide range of light modulators,low cost,simple principle,and high utilization efficiency of the photoelectric sensor's spatial bandwidth.However,because of the co-propagation object and reference light,the real image is overlapped during the reconstruction process,resulting in seriously effects on the reconstruction.Therefore,how to eliminate the influence of noise,such as twin-images,out-of-focus artifacts and improve the reconstruction quality,is a key problem to be solved by the in-line digital holography technology.In regard of the above problems,this project focus on research and discussion through theoretical analysis and simulation experiments.The main contents of the article are as follows:1.This paper proposes a compressive sensing algorithm based on sparse differences,which solves the problem of twin-image between the conjugate image and the real object image in the reconstruction of the in-line digital holographic microscope.The Fourier transform in the wave propagation process has the property of transform basis.Compressive sensing technology is introduced,and the twin-image is iteratively eliminated by total variation sparse constraint iteration according to the difference in the sharpness of the twin-image and the real image.Compared with the traditional phase retrieval method,the results of simulation experiments and hologramreconstruction experiments verify the effectiveness of the algorithm.2.Compressive sensing deconvolution 3D reconstruction algorithm is proposed to solve the problem that noise such as out-of-focus image and twin-image in holographic 3D reconstruction affect the reconstruction quality.Compressed perception tomography reconstruction is used to replace traditional anti-propagation reconstruction,and deconvolution technology is introduced to filter the reconstruction results.Through experiments on holograms of different objects,the experimental results prove that this method can obtain higher accuracy.3.An end-to-end in-line digital holographic reconstruction network based on convolutional neural network is proposed,which optimizes the holographic reconstruction denoising process with complex prior parameters and time-consuming and so on.By learning the imaging characteristics of the holographic reconstruction reconstruction image,the noise-free reconstruction image can be directly reconstructed directly from a single in-line digital hologram.And through the simulation of holography and equipment shooting to build training and testing data set.Experimental results show that the network has a good reconstruction effect.Compared with the traditional back propagation and iterative method of eliminating twin-images,the reconstruction time consumption is reduced.