Dental Biometric Identification System using AR Model

Dental biometry has a leading role in forensic human identification. Identifying a person in mass disasters and major catastrophes, which have frequently happened due to airplanes crashes, tsunamis, fire accidents, etc is a challenging problem if conventional biometrics, e.g., face, fingerprint, iris, etc. is not available. Dental characteristics of persons are naturally unique and can be used to identify individuals based on their dental radiographs. In this paper, we present a new and workable method for identifying humans, that extracts the dental mandibular information from panoramic dental radiographs which is used as a biometric identifier. The system segments the mandible from dental panoramic X-ray images to obtain the outer mandibular contour coordinates. Time series is then obtained from the extracted contour coordinates which gives the structural information of the mandible. AR model is then fitted to this time series and the AR coefficients thus obtained form the feature vector representing each mandible. These feature vectors are later used for matching and identification using the Euclidian distance classification criteria. In order to assess the proposed system, we have carried out experiments at three different orders of AR model using a database of 120 ante-mortem and 90 post mortem panoramic dental images, obtained from 30 individuals. The experimental results show that the proposed system is effective in identifying individuals and exhibits better results at order 3 of AR model with a Recognition rate (RR) up to 75.52%, low Equal error rate (EER) of 23% and a rank-1 identification rate of 76.66%.