Binocular, multiple-view and video-based 3D reconstruction is now well-understood and is one of the most active research areas in computer vision. Compelling 3D models are reconstructed for applications such as virtual tourism, cultural heritage preservation and visualization. Most stereo vision methods, however, fall short in two areas: they cannot handle dynamic scenes and they are completely unaware of semantic content. In the first part of the talk, I will present results on static 3D reconstruction highlighting some of the key choices that have to be made in the design of a stereo algorithm. In the second part, I will present preliminary results on dynamic scene reconstruction. Finally, I will show examples of semantic tasks on 3D data that have only been possible, so far, on datasets acquired by laser range scanners.
Philippos Mordohai is an Assistant Professor of Computer Science at the Stevens Institute of Technology, USA. Prior to that, he held postdoctoral researcher positions at the University of North Carolina and the University of Pennsylvania. He holds the Diploma in Electrical and Computer Engineering from the Aristotle University of Thessaloniki, Greece, and the MS and PhD degrees, both in Electrical Engineering, from the University of Southern California. His research interest include 3D reconstruction from images and video, range data analysis, perceptual organization and manifold learning.
Dr. Mordohai serves as an Associate Editor for the Journal of Image and Vision Computing and as a reviewer for numerous international journals and conferences. He has also organized several workshops and symposia. For more details visit http://www.cs.stevens.edu/~mordohai