Enhancement of viewing characteristics of depth-fused display with autostereoscopic display and projection-type display technologies

Abstract

In this dissertation, various methods for enhancing the viewing characteristics of the depth-fused display are proposed with combination of projection-type displays or integral imaging display technologies. Depth-fused display (DFD) is one kind of the volumetric three-dimensional (3D) displays composed of multiple slices of depth images. With a proper weighting to the luminance of the images on the visual axis of the observer, it provides continuous change of the accommodation within the volume confined by the display layers. Because of its volumetric property depth-fused 3D images can provide very natural volumetric images, but the base images should be located on the exact positions on the viewing axis, which gives complete superimpose of the images. If this condition is not satisfied, the images are observed as two separated images instead of continuous volume. This viewing characteristic extremely restricts the viewing condition of the DFD resulting in the limited applications of DFDs. While increasing the number of layers can result in widening of the viewing angle and depth range by voxelizing the reconstructed 3D images, the required system complexity also increases along with the number of image layers. For solving this problem with a relatively simple configuration of the system, hybrid techniques are proposed for DFDs. The hybrid technique is the combination of DFD with other display technologies such as projection-type displays or autostereoscopic displays. The projection-type display can be combined with polarization-encoded depth method for projection of 3D information. Because the depth information is conveyed by polarization states, there is no degradation in spatial resolution or video frame in the reconstructed 3D images. The polarized depth images are partially selected at the stacked polarization selective screens according to the given depth states. As the screen does not require any active component for the reconstruction of images, projection part and reconstruction part can be totally separated. Also, the projection property enables the scalability of the reconstructed images like a conventional projection display, which can give immersive 3D experience by providing large 3D images. The separation of base images due to the off-axis observation can be compensated by shifting the base images along the viewers visual axis. It can be achieved by adopting multi-view techniques. While conventional multi-view displays provide different view images for different viewers positions, it can be used for showing shifted base images for DFD. As a result, multiple users can observe the depth-fused 3D images at the same time. Another hybrid method is the combination of floating method with DFD. Convex lens can optically translate the depth position of the object. Based on this principle, the optical gap between two base images can be extended beyond the physical dimension of the images. Employing the lens with a short focal length, the gap between the base images can be greatly reduced. For a practical implementation of the system, integral imaging method can be used because it is composed of array of lenses. The floated image can be located in front of the lens as well as behind the lens. Both cases result in the expansion of depth range beyond the physical gap of base images, but real-mode floating enables interactive application of the DFD. In addition to the expansion of depth range, the viewing angle of the hybrid system can be increased by employing tracking method. Viewer tracking method also enables dynamic parallax for the DFD with real-time update of base images along with the viewing direction of the tracked viewers. Each chapter of this dissertation explains the theoretical background of the proposed hybrid method and demonstrates the feasibility of the idea with experimental systems.