The study on the quality of simplified ghost imaging

• Main Authors: Tsung-Sheng Lu, 呂宗晟
• Other Authors: Ching-Hsu Chen
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/10730560682622496696

碩士 === 國立嘉義大學 === 電子物理學系光電暨固態電子研究所 === 102 === We study the quality of ghost imaging by using pseudo-thermal light source, which is illuminated by a 532nm laser on a rotating grounded glass or a plastic diffuser. The optical path resembles a Mach-Zehnder interferometer that has two optical paths. The first path is called the reference beam that impinges directly on the right-half of the charge coupled device (CCD) and the second path is the object beam that passes through an object (a double slits) and then impinges on the left-half of the CCD. By calculating the cross-correlation of the two signals from a single CCD we are able to create the ghost image. We use the visibility and the contrast-to-noise ratio (CNR) to describe the image quality. Moreover, we use the statistical methods to analyze the probability density function of the sum of the speckle intensity. When we use a rotating grounded glass as our pseudo-thermal light source, we find that the CNR decreases with increase of the distance between the object and the CCD. We can set a convergent lens between the object and the CCD to recover the CNR. When we use a rotating plastic diffuser as our pseudo-thermal light source, the probability density function for the data at a single pixel inside the object beam region obeys the negative exponential decay but the total intensity of the object beam does not coincide with the complex circular Gaussian statistics. Besides, the total intensity of the object beam exhibits a periodic time variation. Thus, a rotating plastic diffuser is a non-ideal pseudo-thermal light source and it produces a lower-CNR ghost image. Its visibility is hard to be determined. However, by eliminating the frequencies in the fast Fourier transform of the data series of the sum of the left-half signal, the probability density function matches with the partial polarized complex circular Gaussian statistics. Using the FFT-treated data, we can obtain a well-defined visibility and a largely improved CNR.