Date of Graduation

2016

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Lane Department of Computer Science and Electrical Engineering

Committee Chair

Bojan Cukic

Committee Co-Chair

Jeremy Dawson

Committee Member

Matthew Valenti

Abstract

Gender estimation from fingerprints have wide range of applications, especially in the field of forensics where identifying the gender of a criminal can reduce the list of suspects significantly. Although there have been quite a few research papers in the field of gender estimation from fingerprints most of those experiments used a lot of features but were only able to achieve poor classification results. That being the motivation behind the study we successfully proposed two different approaches for gender estimation from fingerprints and achieved high classification accuracy.;In this study we have developed two different approaches for gender estimation from fingerprints. The dataset used consists of 498 fingerprints of which 260 are male and 238 are female fingerprints. The first approach is based on wavelet analysis and uses features obtained from a six level discrete wavelet transform (DWT). Classification is performed using a decision stump classifier implemented in weka and was able to achieve a classification accuracy of 95.38% using the DWT approach. The second approach uses wavelet packet analysis and extracted the Shannon entropy and log-energy entropy from the coefficients of wavelet packet transform and provided a classification accuracy of 96.59% on the same dataset using decision stump classifier implemented in weka.

Download

Share

COinS