GENDER IDENTIFICATION SYSTEM FOR CRIME SCENCE ANALYSIS USING FINGERPRINTS

Abstract

Gender identification or classification is a challenging task in computer vision as the biometrics of male and female such as fingerprints, face, vein have many variations. Among the various biometrics, fingerprints are commonly available in a crime scene. In this, study, gender identification system for crime scene analysis using fingerprints is presented. Initially, the fingerprints are de-noised by median filter and Otsu thresholding is employed to binarize the fingerprints in the preprocessing stage. Then, the features are extracted by Box-Cox transformation method. Finally, the classification is made by logistic regression classifier. A better classification accuracy of 96% is achieved by the gender identification system using Box-Cox transformation and logistic regression classifier.