University of Houston
Department of Computer Science
In partial fulfillment of the Requirements for the Degree of
Master of Science
Victor Negrete
will defend his thesis
Crater Image Classification using Classical Methods and Ontologies
Abstract
By identifying and classifying craters in images of planets and moons, planetary scientists are able to assign relative geologic ages, describe what kinds of asteroids and comets have hit those celestial bodies, study surface properties, and determine if a celestial body ever had weather or flowing water. This kind of study is an example of a task that is time-consuming to produce, difficult to automate, and scientifically important.
In this thesis we present two approaches to automate the classification of gray scale crater images into three classes based on geologic age. The first approach is based on classical computer vision methods, where we use texture analysis and histograms to create feature vectors that are used in conjunction with a neural network to classify a crater image into the three crater classes. The second approach uses domain ontologies, and classifies a crater from its parts.
The two approaches were tested and evaluated for image database from the NASA Voyager Orbiter Mission to Mars. From the results in the experiments, we found that the method using the histogram to classify craters in images was superior to other texture based analysis for the classical methods. Moreover, the classification based on ontologies had a slightly better performance in comparison with the histogram method.
Date: Monday, November 25, 2002
Time: 4:30 PM
Place: 550-PGH
Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Christoph F. Eick