Building Geometric Models using Three-dimensional Laser Rangefinder Data

Gerhard Roth
Institute for Information Technology
National Research Council of Canada

Abstract

At the Institute for Information Technology of the National Research Council of Canada we have been developing laser rangefinders and applying them to a number of practical problems for over a decade. Such sensors directly obtain three-dimensional surface information by using a scanning laser beam. In this talk we will first describe the principles of operation of our sensors and show its ability to capture both geometry and colour at a high resolution.

Then we will discuss how we take a number of range images and build a geometric model of an object or environment. Such models are useful in the application areas of virtual reality, robotics and manufacturing. We are able to take registered range images and make both parametric models, and mesh models. A parametric model consists of spline surfaces while a mesh model consists of many small triangular patches.

The web page http://www.iitsg.nrc.ca/~roth shows some of the models that have been built, and http://www.autsrv.iitsg.nrc.ca/choice.html gives an overview of the entire project along, with some textured models. Most of the models are in vrml form and can be displayed with most vrml viewers.

Biographical sketch

Gerhard has a B.Math from the University of Waterloo and a Masters of Computer Science from Carleton University, where his advisor was Prof. W. Lalonde. Once at NRC he became interested in computer vision, and earned a Ph.D. from the EE department of McGill. His advisor was Prof. M. Levine who is the head of MCARSIM, a research center for intelligent machines. His research involved the application of evolutionary algorthms and robust statistics to the problem of finding curves and surfaces. Gerhard is currently an adjunct Professor at the School of Computer Science of Carleton University, and is also an adjunct Professor in the Computer vision group of Laval University. At the NRC, he is currently working on creating geometric models (triangle meshes and spline surfaces) from range data of objects and environments.