CAREER: Exploring the Multispectral Frontier in Computer Vision

Stevens Institute Of Technology, Hoboken NJ

Investigators

Abstract

The goal of this project is to develop a methodology for analyzing the wealth of information captured by imaging sensors that densely sample visible light. Methods will be developed for automated selection of an "ideal" set of filters for multi-spectral imaging. Algorithms will be explored for accurate interpretation and recreation of scenes with complex light interaction, in particular in the high-dimensional spectral domain. Techniques for extraction of geometric and photometric invariants from video images will be studied, and novel algorithms will be generated for visualization of "invisible" spectral information. The project's education component includes development of new cross-disciplinary undergraduate and graduate courses, appropriate modification of existing courses, and hands-on approach to coursework and research projects.