Parameter Identification and Reconstruction Speed Improvement of Digital Off-axis Holographic Microscope

• Main Authors: Chih-wei Chang, 張致維
• Other Authors: Cheng-hsiung Chen
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/75247357245919242215

碩士 === 國立中正大學 === 光機電整合工程所 === 97 === Compare with other measurement tool, Off-axis digital holographic microscopy have the important characteristics of one-shot and non-contact. It can record the amplitude and phase information simultaneously, and then we can obtain 3D profile through numerical reconstruction. Combining digital image recording with computer operation technology shortened the time that deal with classic holograms. That makes digital holographic microscope becoming a potential tool for optical measurements. In this research, we developed a reconstruction program through MATLAB. We used the same commercial off-axis digital holographic microscopy but two different ways: the reconstruction program in this research and other commercial digital reconstruction software to reconstruct hologram. And then analyzing these two results proved the accuracy of the reconstruction in this study. In the reconstructed process, it really takes time to finding the right reconstruction distance parameters which can decide the image intensity. The quantifiable criterion of image intensity that based on the Fourier weight function can recognize the position of the reconstruction distance. And, the minimum of the focus value which corresponded to the reconstruction distance is the correct position of the reconstruction distance. However, owing to the change of amplitude contrast image in the judge process, our result can not correspond to this way. Using two-dimensional fast Fourier transform to calculate the Fresnel approximation of the classic holograms in order to promote the speed of fast Fourier transform. Utilizing GPU’s advantages and CUDA technique of the graphic card can verify that the rising velocity of 1024x1024 pixels gray value image which according to parallel computing is 45 times MATLAB. This can be conducive to speed up the reconstruction speed when using dynamic or real-time measurements if we need.