Optimization Of Phase Reconstruction And Speckle Reduction In Digital Holography

With the visual requirements increasing,digital holography has attracted much attention as it could record and reconstruct three-dimensional(3D)information.The advantage of digital holography is that it could record the complex amplitude non-destructively so that the object could be reconstructed before people's eyes in three-dimensional form.In digital holography,the complex amplitude recorded in the hologram could be reconstructed by software on the computer,which is more convenient for people to view.However,digital holography still has many defects in the application of 3D display at this stage.Especially in lensless Fourier transform digital holography,due to the aberration in phase reconstruction,only the intensity information is extracted,and the phase information is usually ignored.In addition,the existence of speckle also affects the application of digital holographic 3D display.In this thesis,three algorithms are designed for the aberration in the reconstructed phase and speckle in the reconstructed complex amplitude.These algorithms are implemented and verified by Matlab.Finally,the phase is successfully reconstructed,and the better speckle reduction effect is obtained in the complex amplitude.This thesis focus on the following contents:(1)The basic principles of digital holography are introduced briefly which the off-axis holography is taken as an example.Then the basic principles of lensless Fourier transform digital holography is introduced in detail which includes the aberration in the reconstructed phase and the basic principles of speckle noise in the reconstructed image;(2)Based on introducing the basic principle of numerical compensation and the application of the principal component analysis method in the aberration compensation,we propose a more universal and anti-noise automatic compensation algorithm called “Rotation Principal Component Analysis”.By rotating the image,the principal component analysis method could obtain more information in each direction,so that the aberration surface be fitted accurately.(3)After the phase reconstruction,it is found that there is a lot of noise in the opaque area of the reconstructed phase which is called abnormal phase fluctuation in this thesis.To handle this type of noise,an abnormal phase fluctuation elimination method based on structural similarity is proposed.By inferring the object structure through the illumination model,the object in the reconstructed phase is extracted and preserved.The other region is full of aberration phase fluctuation and needs to be removed.(4)Besides,the speckle in the reconstructed amplitude also needs to be denoise.The problem of the ringing effect exists in the spatial domain mask method and redundant speckle elimination.Therefore,we proposed a method using the Gauss window instead of the rectangular window to extract the sub-hologram and add a gray-value compensation method to improve the contrast reduced by the redundant speckle elimination.The algorithms mentioned above have undergone principle analysis,simulation,and finally verified the feasibility in the experimental data obtained by the optical setup.