Faculty Advisor

Looft, Fred J.

Abstract

For decades, NASA has been interested in the imaging of Earth to monitor terrestrial change as well as to use scientific observations to predict future occurrences. To continue its missions, NASA has invested in the development of new technologies to satisfy its current and future scientific needs. The images retrieved from the orbiting satellites are processed to determine the orientation of one image relative to another. This process, known as image registration, scans multiple satellite images in search of matching patterns. The system's weakness is that clouds pose a problem for a clear examination of the Earth's surface by preventing a satellite from capturing clear images that can be processed by the image registration software. In response to the image obstruction problem, an improved cloud detection system was created for the Landsat 7 mission. The system was composed of several algorithms implemented in software with C. Although the system was able to detect clouds properly, the process of sequentially scanning multiple image channels caused the system to be slow. As a result, it was determined that a new method for the implementation of the cloud detecting algorithms was needed to improve the image processing time. The new cloud detection system was implemented with the innovative reconfigurable computing technology. The new system consists of two different prototype hardware algorithms using different data interface methods. Each prototype offers distinct advantages for implementation of image registration and both prototypes demonstrate increased performance for a fraction of the cost of other systems.

Publisher

Worcester Polytechnic Institute

Date Accepted

January 1999

Major

Electrical Engineering

Project Type

Major Qualifying Project

Accessibility

Restricted-WPI community only

Advisor Department

Electrical and Computer Engineering

Share

COinS