Computer Vision Talks at Columbia University

Layered Reconstruction of 3D Scenes from Multiple Images
 
P. Anandan
Microsoft Research
 
3:00 pm, April 3rd , 2000 
Inter School Labs,  7th floor CEPSR, Schapiro Building, Computer Science

Host: Shree K. Nayar

 

Abstract

A central problem in Vision is the reconstruction of 3D scenes from multiple 2D images of that scene.   One of the most powerful visual cues useful in this process is the coherence of visual motion over space of time.  Since their introduction about 10 years ago, layered motion models have been a powerful way to describe and recover multiple coherent motions in image sequences.   Although they have proved promising for multiple motion analysis, their full use in scene reconstruction remains an open problem.  In this talk, I will describe our research effort aimed towards obtaining layered descriptions of a scene from multiple images.   Our recent work has focused on three aspects of this problem: The first  is an approach for modeling the appearance and geometry of  rigid, static 3D scenes from multiple views of that scene.  We model the scene as a collection of layered 2.5D sprites. Each sprite corresponds approximately to a "cardboard-cutout" description of a portion of the scene together with a "parallax" component, which describes the finer variation of the shape of that region.  A semi-automatic technique is used for recovering the layered description of a scene from a given set of input images.   Our second effort is aimed at automatically initializing the layer segmentation process using a statistical approach.  A Bayesian formulation of the problem is used to automatically determine the number of layers and an initial segmentation of the scene into layers.  Our most recent effort focuses on decomposing multiple images of a scene containing reflections and transparency into component layer images.  All of the ideas will be illustrated with real-image examples.