design element
RIACS banner
 
section image top Outreach       
section image bottom     design element

 
Seminar
Home › Outreach › Seminars › Seminar
Title:
 Classification and content-based retrieval of images for planetary exploration
Date:
 March 9, 2006
Speaker(s):
 Celine Meyer
Affiliation(s):
 Autonomous Systems Laboratory - Swiss Federal Institute of Technology / Lausanne
Abstract:

Much of the scientific analysis of images from Mars still requires tedious and slow manual processing. Next generation planetary and lunar rovers will traverse larger distances, operate for longer durations, and have access to higher bandwidth communication. In order to effectively make use of the increased access to science data, automated image analysis tools are required.

Onboard planetary rovers and landers, these tools will enable data triage: allowing selection and prioritization of science observations. On the ground, these tools will enable efficient and timely analysis of large quantities of images, as well as automatically finding relationships between similar images in large collections.

In this talk, I will describe how computer vision analysis of geological samples can be used to:

  • automatically cluster image textures present in a data set into a set of related natural characteristics
  • segment new images and classify the segmented regions
  • organize related textures by caching and matching texture statistics for fast retrieval of similar images.
Speaker's Bio:
Celine Meyer is a diploma candidate in the Autonomous Systems Laboratory of the Swiss Federal Institute of Technology / Lausanne. Her current research focuses on techniques for image texture classification and segmentation. She previously developed control methods for a microgravity free-flyer experiment as part of the European Space Agency's student parabolic flight campaign.

 

 design element
design element
 design element
design element