Study On Twin Image Removal And Phase Reconstruction Of Lens-less In-line Digital Hologram

Lens-less in-line digital holography(LIDH)is a microscopic imaging technology that can simultaneously obtain object amplitude and phase information.It has the advantages of large imaging field,high resolution,high space-bandwidth utilization,and no need to dye samples.Meanwhile,owing to its simple optical path structure without the complex optical modulators and optical lenses,it is also advantageous in portability and low cost,which makes it particularly suitable for quantitative information statistics of microbial samples.However,the object light and the reference light propagating in the same direction of LIDH brings about the twin image interference to seriously affect the quality of reconstructed object.In addition,the phase information obtained directly from the reconstructed object light field is wrapped,so the phase unwrapping algorithm is necessary to restore the original phase of the object.Therefore,twin image removal and accurate phase unwrapping algorithms are vital issues associated with LIDH.This thesis focuses on these two aspects to carry out the systematic algorithm study,simulation analysis and experimental tests,with the aim to improve the quality of the reconstruction images of LIDH and the accuracy of phase recovery.Firstly,the related imaging and reconstruction theories of LIDH were reviewed.Based on the existing algorithms,an improved dual-wavelength angular spectrum iterative removal of twin image reconstruction algorithm and an improved K-value rounding two-way shear phase unwrapping algorithm combined with regional discrimination were proposed.Then,MATLAB platform was adopted to simulate the interference recording process of LIDH.Three types of the object optical field,including the pure amplitude,pure phase,hybrid of amplitude and phase,were constructed by simulation.Moreover several different wrapped phase values were selected from the simulated optical field of single and mixed objects values were selected,and then used to simulate and analyze the improved twin image removal algorithm and unwrapping algorithm respectively.Finally,the LIDH images of particles in different sizes,red blood cells and white blood cells were obtained via.experiments.Furthermore,the twin image removal effect and phase unwrapping effect of the improved algorithm were analyzed and verified by A combination of qualitative analysis and quantitative comparative analysis based on image sharpness evaluation parameters and statistical parameters were applied to verify the effects of the improved algorithm the twin image removal and phase unwrapping.The simulation and experimental results indicated that the dual-wavelength angular spectrum iterative reconstruction algorithm performs significantly better than the traditional angular spectrum reconstruction algorithm in the three image sharpness evaluation parameters including the energy gradient of the reconstructed amplitude image,the Variance of gray-scale and the weighted logarithm of the Fourier spectrum.Moreover,for different types of objects such as pure amplitude,pure phase and amplitude & phase hybrid,the simulated reconstructed object light field and the original object light field have the smallest root mean square error values in both amplitude and phase.What's more,they are the closest in appearance.In addition,the root mean square error between the unwrapped phase of the single or mixed sample or the mixed sample obtained by the improved phase unwrapping algorithm and the original phase is smallest.Neither pulling nor failure to unwrap phenomenon is observed.Besides this,the experimentally measured thickness values of the particles and cells are also consistent with the standard ones.Therefore,the improved algorithms proposed in this thesis can effectively remove the twin image and recover the amplitude and phase information of the object more accurately.They are suitable for the reconstruction of LIDH images,and would be helpful to promote the applications of LIDH in the fields of biomedical morphology observation and microscopic particle field detection.