The Accurate Reconstruction Of Light Field Based On Multiple-Intensity Image Phase Retrieval

Phase retrieval technique is regarded as a tool to use intensity images to reconstruct the complex amplitude of sample, which consisit of holography and iterative phase retrieval. Within more precise implentmentation and no reference beam, iteratve phase retrieval is paid more and more attention. The early iterative phase retrieval, such as GS algorithm and HIO algorithm, is hinder by ambiguity and slow convergence. For this propose, multiple-image phase retrieval is proposed, which speed up the convergence by diffraction diversity and iterative computation.As for multiple-image radial phase retrieval, the basic theory and application of PIE algorithm are extensively developed. On the contrary, convergence condition and performance of recovered phase for multiple-image axial phase retrieval are not discussed in detail. In consideration of this, my motivation is to bulid a more complete theory for multiple-image axial phase retrieval, which is followed as:Firstly, the domestic and overseas advance on iterative phase retrieval is discussed. The Basic theoretical model of phase retrieval algorithm based on diffraction computation is summarized. Moreover the mechanism of multiple-intensity image phase retrieval algorithm(MIIPRA) is analyzed. Here every image is from a computational unit. In MIIPRA, the all computational units are combined with different schemes.Secondly phase retrieval algorithm can be used in the imaging field to recover the amplitude distribution. Therefore amplitude reconstruction is regarded as a point of penetration. The convergence condition of MIIPRA is discussed at the aspects of diffraction transmission computation, evaluation of initial value and supporting region. For the reconstruction of phase information, the thesis focuses on two cases, continuous style and discrete style. The convergence precision of APR algorithm(Amplitude-phase retrieval,APR) and other MIIPRA methods at axial direction.Finally the recovered result of MIIPAR is disturbed due to the distance error at axial direction. We analyze the influence of the error on APR quantificationally. With the help of curve fitting, the distance error distribution function having uniform error distribution is calculated. The error evaluation function is also designed for improving APR method.