My general areas of research are in Computer Vision, Image Processing, and Machine Learning. For my thesis, I explored the detection and use of occlusion boundaries in short video clips. More information on this project is available on this page, and in my publications.
My previous research and work experience is listed below.
I do most of my work in Matlab, which I find to be an excellent environment for rapid prototyping and visualization of ideas. I also often incorporate C/C++ code using Matlab's MEX capabilities.
I spent the summer applying image processing and machine learning techniques to ChemImage's multi-spectral chemical data (e.g. Raman spectral imagery) for prostate cancer detection and auto-targeting.
I was an intern in the Computer Vision Group at NASA's Jet Propulsion Laboratory in Pasadena, CA. There, I worked on attenuating the "pixel-locking" phenomenon in sub-pixel stereo disparity estimation, and to a lesser extent on evaluating Symmetric Stereo using Belief Propagation.
I was an intern at Intel's research lab in Pittsburgh, where I began development of an online structure from motion (SFM) system to reconstruct environment geometry and camera motion from video sequences.
Before initiating my current thesis research, I was working with Sebastian Thrun (before he moved to Stanford) on structured light range imaging and on the construction of 3D object models from range data.
Before coming to CMU, I earned my bachelor's degree in Electrical Engineering and my master's degree in Electrical & Computer Engineering at Georgia Tech.
As an undergrad, I was involved with the Robotics Team, which competed in a regional IEEE-sponsored contest each year.
As a graduate student, my main project was the development of a real-time (109 fps) Bayesian tracking system for the Air Force's Airborne Laser missile defense system. Surprisingly, the code I developed also proved useful for blood vessel segmentation from medical imagery. I also worked with active contours and level set methods for image segmentation and tracking.