This dissertation presents a new method of digital holography for numerical reconstruction of a 3D object. Unlike the results of reconstructing an off-axis hologram by Fresnel transformation method, the numerically reconstructed images with our algorithm are free of conjugate images and free of zero order of diffraction. And since Fresnel transformation is merely the numerical simulation of the physical process of diffraction, any two-dimensional images obtained directly from Fresnel transformation, are only the projections of the 3D object onto a 2D plane, so they cannot provide details of each layer of the 3-D object. Our proposed method is an exploratory effort to obtain detailed information of all the layers of the 3D object....[ Read more ]

This dissertation presents a new method of digital holography for numerical reconstruction of a 3D object. Unlike the results of reconstructing an off-axis hologram by Fresnel transformation method, the numerically reconstructed images with our algorithm are free of conjugate images and free of zero order of diffraction. And since Fresnel transformation is merely the numerical simulation of the physical process of diffraction, any two-dimensional images obtained directly from Fresnel transformation, are only the projections of the 3D object onto a 2D plane, so they cannot provide details of each layer of the 3-D object. Our proposed method is an exploratory effort to obtain detailed information of all the layers of the 3D object.

First, our algorithm can separate the object-wave spectrum from the hologram. This object spectrum can be numerically used to form a series of 2-D sectioning images of the object as physically using an "imaging system" to focus on it. But these images contain both the in-focus information of the exactly focused layer and the vague information caused by all the other out-of-focus layers of the object. Thus PLI-detect-technique is applied in spatial domain to extract the PLI (information that is exactly on the focused layer), which is essential for later reconstruction. Finally, by using an iterative image-processing algorithm that performs both in the spatial and the frequency domains, the details of the 3D model of the transparent object can be obtained. In our algorithm, the constraint condition is satisfied and the noise distribution of each reconstruction cycle can be adjusted so that statistics theory is used to improve the reconstruction quality.

Unlike the conventional technologies to acquire data in point such as laser scanner, or in a frame such as vision, the proposed research provides a powerful tool to acquire data in a volume (3D) at once. This technology may have a profound impact on various fields related to measurement, manufacturing, and control application, or any further applications in the non-destructive detection of the industrial structures & components, medical image diagnosis such as cancer detection.

Keywords: Digital holography, numerical reconstruction, layer information