Mei Han
School of Computer Science, Carnegie Mellon University
|
|
I am a Ph.D. student at Robotics Institute. My research interests are in computer vision and computer graphics. My advisor is Takeo Kanade. I was studying in Department of Computer Science, Tsinghua University from 1987-1995. |
Research Interests
Structure from
Motion
Scene Reconstruction with Multiple Moving Objects
Currently I am working on the reconstruction of a scene containing multiple moving objects. Given a monocular image sequence, we recover the scene structure, the trajectories of the moving objects and the camera motion simultaneously. The number of the moving objects is automatically detected without prior motion segmentation.
Scene Reconstruction with Uncalibrated Cameras
I am also working on a factorization-based method to reconstruct Euclidean shape and motion from multiple perspective views with uncalibrated cameras. The method first performs a projective reconstruction using a bilinear factorization algorithm, and then converts the projective solution to a Euclidean one by enforcing metric constraints.
Homography Based 3D Scene Analysis
My interest is on 3D scene analysis of video sequences. I propose a framework to recover projective depth based on image homography and apply it to motion detection and three dimensional mosaicking. The techniques include a robust homography algorithm which incorporates contrast/brightness adjustment and robust estimation into image registration, a camera motion solver to obtain the ego-motion and the real/virtual plane position, and temporal integration of dense projective depth map over video sequences.
Ink Jet Printer Inspection System
I was working with Reconfigurable Vision Machine group in the project of building ink jet printer inspection system. Considering the structure of ink jet printing process, we designed the affine dot printing algorithm for efficient image registration.
Summer Intern
Interactive 3D Modeling from Panoramic Mosaics
During the summer of 1997, I was working with Harry Shum and Richard Szeliski at Vision Technology group in Microsoft Research. We built an interactive modeling system that constructs 3D models from one or more panoramas. Panoramas help to decouple the modeling problem into a zero baseline problem (building panoramas from images taken with rotating camera) and a wide baseline structure from motion problem (recovering 3D model from multiple panoramas). Interactive system directly incorporates the regularities present in the environment so that accurate 3D models can be constructed.
Email: meihan@cs.cmu.edu