Resolution Optimization Of Off-axis Digital Holopraghy Based On Image Interpolation

Digital holography is a three-dimension imaging technology,which has the advantages of non-contact,large depth of field,real-time and fast imaging.Digital holography use CCD to record the complex amplitude field formed by the interference between object light and reference light.The hologram is reconstructed and processed by computer simulation of diffraction propagation process,and the amplitude and phase information of the object are obtained.It has a good application prospect in biomedical imaging and microstructure morphology detection.The size and resolution of the hologram are limited by the size of the sensitive area and the pixel size of CCD camera.Because the pixel size affects the sampling frequency of the image,the sampling frequency affects the imaging effect of the reconstructed image.The higher the sampling frequency,the much more the details of the image are.The area size of the recorded hologram and its pixel size will affect the resolution of the reconstructed image.Although the pixel size of CCD device is shrinking,improving the imaging resolution is always an important driving force for the development of digital holography technology.As research developing,a variety of effective optical system and reconstruction methods are proposed.In this thesis,the optimization of off-axis digital holographic reconstruction image based on multi-frame super-resolution is studied,and specifically the characteristics of multi-frame hologram resolution improvement under Bicubic interpolation and Kronecker product interpolation are focused on.The first part of this thesis is to optimize the reconstruction image of off-axis super-resolution digital holography with Bicubic interpolation method.Firstly,an offaxis digital holography system is built in the experiment.The CCD mounted a twodimensional translation is controlled to move along the X and Y directions,and several groups of low resolution hologram sequences with sub-pixel micro displacement are recorded.By using MATLAB software,Keren motion estimation algorithm is used to accurately calculate the micro-displacement of each low resolution hologram.One of the low resolution holograms is interpolated with Bicubic interpolation method to obtain a high-resolution guessing image.By using the image as a reference,the motion estimation results of the original low resolution image sequence are used to downsample the high-resolution guessing image for many times,until the high-resolution guessing image is down-sampled to the spatial position of each frame of the original image.Then,by comparing it with the original image,the residual and standard deviation are calculated,and used as feature information.The feature information of the low resolution hologram sequence is mapped to the high resolution network one by one,and a super-resolution hologram is generated and reconstructed.Experimental imaging results show that this method can effectively improve the resolution of reconstructed images without increasing the complexity of the algorithm.In the second part,the super-resolution off-axis digital holography based on Kronecker product interpolation is studied.Herein,the single frame reconstruction based on Kronecker product interpolation is applied to combine with multi-frame superresolution reconstruction of off-axis holograms,to increase the high-frequency information on the basis of expanding the spectrum filtering range.Firstly,the low resolution hologram sequence is obtained by experiment.After motion estimation,the3×3 unit matrix is used to interpolate the hologram sequence by Kronecker product,and each pixel of the hologram is expanded into one 3×3 interpolation matrix,via which a sequence of new holograms are obtained.Finally,the multi-frame image superresolution processing introduced in the previous chapter is used to iterate this new hologram sequence,to generate one super-resolution hologram.Then,Fourier transform on the super-resolution hologram is performed to obtain its spatial spectrum with eight aliasing terms.The eight aliasing terms are added together,and their sum is placed back to the position of the original aliasing term.Filtering interception on one positive first-order spectrum among the sum aliasing term is performed,and then holographic imaging reconstruction is achieved by using this spatial-frequency spectrum.The results show that this method can extract more amplitude and phase information of the hologram.