Dental Biometric Identification Scheme Based On Complex Auto Regression Model-An Aid to Forensic Dentistry

Abstract

Automating the Forensic dental identification process is receiving increased attention. The main role of dental forensic experts is to identify individuals using the information extracted from their dental characteristics especially during major catastrophes like tsunamis, fire-accidents where the conventional biometric traits are usually unavailable or completely destroyed. In this paper, a dental biometric identification scheme is presented that extracts dental features from mandible using orthopantomograms. This system first traces the outer boundary coordinates of mandible and obtains a time series from these coordinates. Complex Auto regression (CAR) model is fitted to the obtained time series to get the CAR coefficients, which represent the mandibular structure. These coefficients are subsequently used for identification of individuals using clustering. The experimental results indicate that the proposed scheme exhibits better performance at CAR order six with low equal error rate of 19.850, and top-1 retreival accuracy of 0.80, thereby establishing its use in forensic applications. It is to mention that the results obtained using mandible bone as biometric identifier are not only at par with those obtained using the teeth and dental work based identification schemes but also the use of mandible bone as a biometric identifier overcomes the limitations associated with the teeth and dental work based identification schemes.