| Summary: | 碩士 === 逢甲大學 === 通訊工程學系 === 102 === Tomography Digital Holographic Microscopy provides a way to obtain fully three-dimeension imformation of the object based interference techniques. In this system, the coherence light source illuminates the rotating transparent micro-object and the transmitting wavefield is interfered with a reference plane wave. In digital holographic, one uses a charge-cover device (CCD) camera to record the hologram of the interference pattern and stored in the computer. For each rotating angle of the sample object, one can capture sectional information of the 3-D object. To integrate the holograms obtaining by the all rotating angles, one can reconstruct truly 3-D tomography information of the object.
We can use Fourier diffraction theory to simulate Tomography Digital Holographic Microscopy. Then, the sectional information of each rotating angle is sampling in the Fourier domain. The reconstruction algorithm then interpolates all data in the three-dimensional Fourier domain. The truly 3-D tomography information of the object can then be obtain by the 3-D inverse Fourier transform. However, by the diffraction theory, the data inside a region of the Fourier domain is loss by using the TDHM scanning technique, named as Apple core distortion, since the shape of the missing region looks like apple core. In this thesis, the compressive sensing algorithm is employed to accounts for the Apple-core distortion problem to achieve a better reconstruction.
Compressed sensing can be used as the image reconstruction for the inaccurate acquisition system. In this way, we have the sensing mechanism in the Fourier domain and the sparity of toatal variation. With this in mind, two-Step Iterative Shrinkage/Thresholding (TWIST) is developed for the reconstruction of the tomography data.
|
|---|
