Date of Award

Summer 7-1999

Document Type


Degree Name

Master of Electrical Engineering (MEE)


Electrical Engineering

First Advisor

Dr. Edward Doering

Second Advisor

Dr. Michael Mcinerney

Third Advisor

Dr. Bruce Black


This thesis describes a new approach for fingerprint identification that will be shift and rotation independent. Detailed descriptions of directional filtering, foreground and background segmentation, feature extraction, and matching based on structural correlation are the main topics of this thesis. The fingerprint identification system consists of image preprocessing, feature extraction, and matching which run on a PC platform. The preprocessing step includes histogram equalization, block-based directional filtering, thinning, and adaptive thresholding to enhance the original images for successful feature extraction. The features extracted will be stored in the database for matching. The matching algorithm presented is a modification and improvement of the structural approach. A two-step process of local feature matching and global feature matching guarantees the correct matching results.