My research focuses on the problem of extracting structural and spatial information from a messy, unorganized, raw visual signal, something human beings accomplish effortlessly all the time. In particular, I am interested in what computational tools can be applied to the problem of visual structure and perceptual organization, with specific focus on techniques that offer simplicity, versatility, and intuition. Several of the visual problems I have studied are listed below.
An intuitive model framework for computational implementation of classical Gestalt principles, such as grouping by proximity, grouping by similarity, and good continuation.
A novel technique for inflating 2D silhouettes in to 3D shapes, which offers insight into the perception, representation, and manipulation of 2D and 3D shapes.