Author

Krause

Date of Award

12-2007

Degree Name

Master of Science

Department

Electrical and Computer Engineering

First Advisor

Dr. Ikhlas Abdel-Qadar

Second Advisor

Dr. Liang Dong

Third Advisor

Dr. Osama Abudayyeh

Fourth Advisor

Dr. Sherif Yehia

Access Setting

Masters Thesis-Open Access

Abstract

Ground Penetrating Radar is a valuable tool for infrastructure condition evaluation as well as in archaeology and geology. The image produced by a GPR scan can be difficult to interpret due to the weak reflection amplitude, the overlap of reflections, and due to outside interference. This thesis proposes an algorithm to automate GPR scan interpretation. This will enable maintenance engineers to read GPR scans quickly and accurately.

The proposed algorithm framework uses an optimization-based deconvolution technique followed by a segmentation process. Each column of the image is deconvoluted and thus reducing them to a list of ordered pairs of reflection amplitude and delay time. The ordered pairs appearing in multiple columns are grouped to form an arc and line objects using a via similarity measures. The algorithm was successfully tested using synthetic and real GPR data.

Share

COinS