Research On Adaptive Filtering And Reconstruction Method Of Off-Axis Digital Holography

The interference field of object wave and reference wave in digital holography is typically recorded by a photoelectric sensor,instead of that in optical holography by traditional holographic recording material.The reconstruction process is simulated by computer algorithm,and the amplitude and phase maps of the object are reconstructed.The recording process is based on the interference principle.Interference fringes are generated by coherent superposition of the object wave and the reference wave on holographic recording plane.The photoelectric sensor device records the amplitude and phase information of the object wave into these fringes.The reconstruction process is based on the principle of diffraction imaging,and the diffraction propagation process of the reconstructed wavefront is simulated by computer.The reconstructed object wavefront is obtained by numerical calculation using the diffraction integral formula,and the holographic reconstruction results are displayed as amplitude and phase images.Digital holography has the advantages of high sensitivity,high resolution,fast imaging speed and flexible recording and reproduction.With the development of digital holographic technology in recent years,the application scope of digital holography is expanding.It has been applied to many fields,such as morphology measurement,threedimensional image recognition,life science,material science and so on.With the wide use of digital holography,the application prospect of high-quality image reconstruction with high automation and low error rate is gradually emerging.Spectrum filtering is one necessary procedure in off-axis digital hologram reconstruction.Due to the significant differences in the spatial frequency distributions of different object waves,manual filtering will add manual workload and timeconsuming,especially for processing a large amount of data.On the other hand,it is difficult to ensure the consistently of the quality of the reconstructed images.Therefore,by optimizing the filtering process,it is expected to improve the automation of the reconstruction process and the image quality of the reconstructed amplitude and phase.Based on the principle of off-axis digital holography,the adaptive filtering reconstruction and the fractional Fourier transform reconstruction are explored,to improve the automation filtering and the imaging quality in off-axis digital holographic reconstruction by optimizing the reconstruction algorithm.It mainly includes two parts:In the first part,the adaptive spectrum filtering combining K-means clustering method and automatic interception algorithm is studied to improve the adaptability of hologram reconstruction.Firstly,the adaptive reconstruction algorithm of off-axis digital holography based on K-means clustering is studied,includes evaluating the advantages and feasibility of unsupervised learning in spectrum filtering algorithm and completing the clustering adaptive reconstruction method.Then,based on the basic principle of K-means clustering and according to the spatial spectrum distribution characteristics of off-axis digital hologram,the spectrum distribution of off-axis digital hologram is adaptively clustered into the zero-order spectrum,the positive first-order spectrum and the negative first-order spectrum.Based on the common attenuation rate of communication signal and the spectrum intensity of the positive first-order spectrum center,the radius of the filter window is automatically determined.Combined with the position of the positive first-order spectrum center and the radius of the filtering window,the filtering range of the off-axis digital holographic spectrum is adaptively selected.The K-means clustering filtering algorithm is given.Then,combined with the diffraction propagation formula,the amplitude and phase images can be reconstructed from off-axis digital holograms by the adaptive method.The reconstruction of different off-axis digital holograms with the K-means adaptive filtering method is realized.The results show that the images reconstructed by the adaptive reconstruction algorithm based on K-means is not inferior to the images reconstructed by manual filtering algorithm.Using the adaptive reconstruction algorithm could reduce the manual error and time-consuming.In the second part,the resolution and image quality of amplitude and phase images reconstructed by filtering in fractional Fourier transform domain for off-axis digital holograms are studied.Firstly,the fractional Fourier transform of off-axis digital holograms and the characteristics are described.Analyzing the possibility and advantages of fractional Fourier transform filtering of off-axis digital hologram.Secondly,the amplitude and phase reconstructions by fractional Fourier transform domain filtering for the computer generated digital holograms are performed.The image reconstructed by fractional transform domain filtering is compared with that reconstructed by conventional Fourier spectrum filter.It is verified that imaging reconstruction by filtering in the fractional transform domain can improve the resolution of amplitude and phase images.The influence of the zero-order spectrum on fractional-transform-domain filtering is analyzed.After the elimination of the zeroorder term of the hologram,the resolution of reconstruction imaging can be improved by fractional-transform-domain filtering.The reconstruction characteristics with the different orders of fractional Fourier transform are compared.Further,in the experiment,the hologram with the zero-order-term elimination is generated by the subtraction of two holograms,and the reconstruction of amplitude and phase of this resultant hologram is achieved by fractional Fourier transform domain filtering.The results show that by using the proper order of fractional Fourier transform,the amplitude and phase reconstruction have higher resolution and image quality than that reconstructed by conventional spectrum domain filtering.In particular,for the zero-elimination hologram,the fractional Fourier transform domain filtering can improve the resolution of the reconstructed images more significantly.