Authors

Xiao XiaoFollow

Date of Completion

9-3-2014

Embargo Period

8-27-2016

Keywords

3D imaging, 3D displays, Integral imaging

Major Advisor

Bahram Javidi

Associate Advisor

Rajeev Bansal

Associate Advisor

Quing Zhu

Field of Study

Electrical Engineering

Degree

Doctor of Philosophy

Open Access

Campus Access

Abstract

The physical world around us is three-dimensional (3D) and 3D perception is inherent in human vision. Unfortunately, traditional imaging sensors can only provide two-dimensional (2D) images and cannot acquire depth information. Due to the lack of 3D information, our perception and understanding of the complexity of real-world objects are greatly limited. In the past several decades, new technologies for 3D imaging and 3D display have been extensively researched for diverse applications. The growing interest in the field of 3D technologies has been stimulated by advances in high-resolution and relatively inexpensive electronic imaging sensors, and increasing computational power.

In this dissertation, we focus on novel aspects of multi-view 3D imaging systems and auto-stereoscopic 3D displays. We organize the dissertation into two separate parts. In the first part, an estimation algorithm of sensor positions in several types of multi-view imaging systems is proposed. In particular, in photon starved multi-view imaging systems, in order to estimate sensor positions, we utilize a total variance expectation maximization technique to recover more details of 2D images captured in low level light. In addition, a 3D polarimetric computational imaging method is presented to distinguish polarization signatures of objects in 3D space. Techniques for depth estimation and occlusion removal in 3D multi-view imaging systems are also discussed.

In the second part of this dissertation, two new techniques for auto-stereoscopic 3D displays are developed. Optical experiments for a high resolution holographic display using dense ray sampling and integral imaging capture are first presented. Second, in order to improve the quality of the displayed images, a multiple-planes pseudoscopic-to-orthoscopic conversion method is proposed for 3D integral imaging display systems. Display results are presented to demonstrate the viability of the proposed techniques.

COinS