Research On Key Techniques Of Slice Imaging In Optical Scanning Holography

Optical scanning holography is a real-time,high-resolution digital holography technology.It can obtain three-dimensional image information of fluorescent specimens and has broad application prospects in the field of biomedical imaging.In this thesis,we will explore the imaging problem of sliced objects in optical scanning holography through the theoretical derivation,software simulation and experimental verification methods.It mainly contains the following aspects.(1)The wave equation and Fresnel diffraction theorem are deduced by two-dimensional Fourier transform,convolution and correlation principles.The theory of recording and reconstruction of holographic image in optical scanning holography system are then studied,in which the key problems of sliced object imaging in optical scanning holography system are indicated.(2)The key technologies of sliced object imaging is introduced,including autofocusing problem,the research process of auto-focusing technology,two forms of defocusing noise,research progress of speckle noise elimination technology,etc.In addition,the two forms of defocus noise are analyzed in detail.(3)In this part,the commonly used image evaluation index functions is introduced,including the standard deviation correlation function,the energy of the image Laplacian,etc.The characteristics of reconstructed images in the blind reconstruction process of holographic images is then analysed.The auto-focusing algorithm based on image evaluation index function and the improved auto-focusing algorithm based on edge gray difference function are proposed.The effectiveness of the proposed algorithm has been verified through simulation as well as the experiment.(4)The concept of machine learning and its development process are briefly introduced in this part.The origin,characteristics and training methods of an unsupervised learning algorithm SOM neural network are introduced.According to the characteristics of speckle noise and the clustering characteristics of SOM neural network,a defocusing noise elimination algorithm based on SOM neural network is proposed.The effectiveness of the algorithm is validated by combining with simulation data and noise elimination evaluation.