Stella X. Yu : Research : thesis

Effortless perception is only possible through the popout of important information by our sensory system. We provide a computational account on how popout by novelty and popout by familarity can be implemented.

Our models are obtained by empowering current spectral graph theory with a richer representation of grouping cues, a grouping criterion that accounts for all cues simultaneously, and a computational solution that finds near-global optima efficiently.

Computational Models of Perceptual Organization

Stella X. Yu, University of California, Berkeley

Computer Vision Seminar, 10 June 2003

Computational Models of Perceptual Organization

Stella X. Yu

Ph.D. Thesis, Carnegie Mellon University, May 2003

Abstract | Paper | Slides