Office of Science & Technology - Introducing Cornelia Fermüller: The Science of Seeing
Menu Content/Inhalt

Podcast

This is the subscription link for bridges podcasts.

podcast
Please find more information in the podcast section .
Introducing Cornelia Fermüller: The Science of Seeing Print E-mail
bridges vol. 13, April 2007 / News from the Network: Austrian Researchers Abroad
by Roland Schneider


mp3 download

Robots nowadays can do a lot of things that normally only humans can do. They mow the lawn, build cars, and some of them can walk like we do, or even play soccer (maybe even better than we do ...). Over the past 50 years, robots were "taught" by their human inventors to imitate their creators' abilities. Still, they lack one of our most important abilities through which we recognize and interact with our surroundings - Vision.

"More than 50 percent of the human brain is dedicated to vision, that gives a pretty clear indication of its complexity," Prof. Cornelia Fermueller explains while sitting in her office at the University of Maryland in College Park (UMD).

fermueller_vor_buechern1.jpg
Cornelia Fermüller
Ms. Fermueller, who holds a PhD in the area of Vision and Geometry from the Vienna University of Technology (1993), has been working as a research scientist at the Center for Automation Research in the Computer Vision Laboratory at UMD since 1994. For more than 10 years, Ms. Fermueller's research has focused on one particular issue: understanding how the human vision system works, and applying those findings to create an artificial vision system that has navigational capabilities and, eventually might even have recognition capabilities.

Discovering the secrets of vision
One might frivolously assume that vision is a rather simple task. After all, humans do it every day so effortlessly, and even camera-equipped cell phones are able to take pictures of objects.

A very popular fallacy - the exact opposite is the case: While it is indeed fairly simple to record images of our surroundings, it is very hard to understand those images and react accordingly. Fermueller explains in basic terms what her field of visual navigation in computer vision and robotics is all about: "Imagine a robot that wanders through the world and as humans have eyes the robot has cameras. Just as a human, or animal, acquires images with his eyes, and the brain processes these images to arrive at an interpretation of the visual world, the cameras on the robot acquire images, which a computer processes. And my task now is to investigate and to understand what these computations are."

Vision: physiological process of distinguishing, usually by means of an organ such as the eye, the shapes and colours of objects.
[Encyclopedia Britannica]
The complexity of vision arises from the fact that while the world has a three-dimensional geometry, images are two-dimensional. Human vision works with the light bouncing off surfaces to interpret the world. The human "imaging device" is the innermost layer of the eye, the so-called retina, which is light-sensing. It converts light into electrical impulses that are interpreted by the brain as vision by decoding these images into information. Fermueller explains where she and her research group face the challenge: "Although it is well understood from the neurosciences that different parts of the brain are devoted to different visual modalities, such as color, shape, movement, or boundaries, we still have to figure out how to combine these different sources of information to arrive at an interpretation of the world."
Access to the full article is free, but requires you to register. Registration is simple and quick – all we need is your name and a valid e-mail address. We appreciate your interest in bridges.
 
Back to Top