|
|
|
“Stereo Panorama Virtual Environment”
by Chiao Wang
May 2008
In this thesis, we describe the creation of virtual environment systems that can capture and display stereo images. The system can use a variety of display formats including auto-stereoscopic (AS) displays, shutter glasses, and head-mounted display (HMDs), and is suitable for virtual tours of outdoor or indoor sites such as a city tour, museums or galleries. We describe techniques for stereo panoramic image capture, present a general image capture model, specify geometrical parameters, and summarize how various physical parameters affect the perceived depth and image quality. We then describe an efficient stitching algorithm that corrects dynamic exposure variation and removes moving objects without manual selection of ground-truth images. We also present expressions for the horizontal and vertical disparity of the captured image and describe disparity measurement techniques. We review various disparity adjustment algorithms, and develop an object-based horizontal disparity adjusting algorithm that changes the disparities to enhance or reduce the stereo visual effect for a selected object region rather than for columns of the images. We describe a disparity morphing tool that enables users to specify the region of interest, select the disparity adjusting methods, and see the results directly on a AS display or with shutter glasses. We also describe the results of several subjective tests that compare system performance using different interaction tools and various displays such as HMD, shutter glasses, or AS displays. Future extension may use a large screen partial or fully panoramic display. In these systems, smaller auto-stereoscopic (AS) panels may be merged for a large screen or panoramic effort, or passive glasses-based technology (polarizing or anaglyph) may provide the stereo effect.