Biometric Identification System using Panoramic Dental Radiograms based on CAR Model

Abstract

Forensic Odontology is as a means of human identification in forensics using dental records of individuals. A dentist extracts the information by manual comparisons between the antemortom and postmortom dental features, which is a very time consuming process. Thus, an Automatic Dental identification System is needed which can save time in identifying people especially in major catastrophes like tsunamis, airplanes crashes, fire accidents etc if conventional biometrics such as fingerprints, face, iris, palm print etc. are unavailable. This paper presents a new technique for person identification that extracts the features from mandibular bone using panoramic dental x-rays. The system first obtains the outer mandibular contour coordinates and a time series is later acquired from these extracted coordinates, which provides information about the mandibular structure. Complex Autoregression (CAR) model is then fitted to the acquired time series and the CAR coefficients thereby obtained represent the mandible features. These feature vectors acquired from mandible are then used for identification of individuals. From the experiments, it can be perceived that the performance of the system in identifying individuals using panoramic dental radiograms is good with a Recognition rate upto 79.3% and an identification rate of 80%.