Forensic Human Identification using Dual Cross Patterns of Dental Panoramic Radiographs

Abstract

Dental biometrics plays a very vital role in identifying the victims in natural and human-made disasters. The
survivability and diversity of dental radiographs make them excellent alternatives over traditional biometric
techniques using fingerprints, face, and iris. The main focus of this study is to deal with missing and unidentified
person identification from a guide to automated based on dental panoramic radiographs using Dual Cross Pattern
features that are complicated to be assessed only by visual examination. This paper seeks to identify an appropriate
classifier amongst Feedforward Neural Network (FNN), Multiclass Support Vector Machine (m-SVM), k-Nearest
Neighbor (k-NN) and Classification Tree (CT) based on retrieval accuracy of 10 adult subjects with 100 panoramic
radiographs. The preliminary results on a small dataset are encouraging.