Anuj Srivastava

Associate Professor, Coordinator of the Laboratory for
Computational Vision
D.Sc., 1996, Washington University

anuj@stat.fsu.edu
 

Associate Editor, Journal of Statistical Planning and Inference, 2001-04

STATEMENT OF RESEARCH INTERESTS SELECTED PUBLICATIONS

 
 
 
 
 
STATEMENT OF RESEARCH INTERESTS
Statistics plays a fundamental role in many engineering applications. One such area that interests me is “image understanding”. Image understanding aims at developing automated systems that can match human abilities in analyzing images. Understanding and modeling “human vision” is a grand challenge; this active area of research involves scientists from neuroscience, psychology, computer science, engineering, and statistics.

Our goal is to develop efficient representations and algorithms that can capture and analyze variability associated with the targets of interest. These targets can be human faces, military vehicles, anatomical parts, robotic objects etc, depending upon the application. Our models are motivated by physics: we capture target shapes using a 3D scanner, textures using digital cameras, motion using video camera, and temperatures using an infrared camera. These physical attributes are then used in understanding the images that contain the targets of interest. Ingredients for this multi-disciplinary research come from computational statistics, differential geometry, and computer graphics.

3D Representation of a face (right) by registering its surface (left) with its texture (middle). 
SELECTED PUBLICATIONS
• On Advances in Statistical Modeling of Natural Images, Journal of Mathematical Imaging and Vision, 18, 17-33, January 2003 (with A.B. Lee, E.P. Simoncelli, and S-C. Zhu).
    • Stochastic Models for Capturing Image Variability, IEEE Signal Processing Magazine, 19, issue 5, 63-76, September 2002. 

    • Universal Analytic Forms for Modeling Image Probabilities, IEEE Transactions on Pattern Analysis and Machine Intelligence, 24, no. 9, 1200-1214, September 2002.

    • Monte Carlo Extrinsic Estimators for Manifold-Valued Parameters, special issue of IEEE Transactions on Signal Processing on Monte Carlo methods, 50, no. 2, 299-308, February 2002.
PREVIOUS PAGE          BROCHURE CONTENTS               NEXT PAGE