Chapter 6 Video-rate Environment Capture with Mirror Pyramid-based Multiview List of Figures

To Sandra iii Acknowledgements I would like to thank my advisor Professor Narendra Ahuja for sharing his wisdom with me, and the members of the Beckman Institute Computer Vision and Robotics Lab for creating the stimulating and supportive environment in the group. I am also very grateful to Professors David Kriegman, Michael Garland, and Yizhou Yu for their suggestions and criticisms that helped me tremendously in many ways. I owe a great debt of thanks to my fiancee Sandra, for being my biggest fan and primary source of inspiration. Lastly, my deepest gratitute goes to my parents and two sisters, for their unwavering and unconditional support that gave me the courage to embark on this journey. 1.1 A typical scene in a video game rendered with texture-mapped 3D models. 1.2 A typical scene in a video game rendered predominantly with image-based techniques. (Scene taken from Onimusha 1 and 2, images courtesy of Capcom.) 3 2.1 Mapping onto the covariance-weighted space. (a) Original geometric entities. (b) Mapped into covariance-weighted space. Notice that the isocontour for cluster 1 is now a circle.. 2.2 Alpha channel estimation experiment with the " lighthouse " image. (a) Original image. (b) Foreground/Background/Mixture map. (c) Estimated alpha channel. (d) Color quantization map. (e) Extracted foreground, composited against a black and a white background.. Alpha channel estimation experiment with the " lion " image. (a) Original image. (b) Foreground/Background/Mixture map. (c) Estimated alpha channel. 3.1 Example of objection selection. (a) Original image, shown with sketch. (b) Resulting selection. An example of vector graphics editing in GRANDMA. (a)-(c) A series of gestures for creating a number of geometric objects, including a rectangle, an ellipse, and a line segment. (d) Shows the selection operator that involves encircling the objects desired. (e)-(f) Manipulation operations on the selected group of objects: duplication(e), rotation and scale(f), and deletion(g). .