Human Identification based on Tooth Contour Guided SIFT Descriptor

Nourdin Al-sherif, Ayman Abaza, and Hany Ammar


Dental Identification , Tooth Contour, SIFT Descriptor


Automated dental identification is one of the best candidates for postmortem identification. With the large number of victims encountered in mass disasters (e.g., September 11, and Asian tsunami), automating the dental identification process would enhance the scalability of this biometric. However, archiving and retrieving dental records from large databases is a very challenging task and has received inadequate attention in the literature. This paper presents a new approach of searching large databases to find a candidate list. This candidate list is short in length compared to the original database size. The basic idea of this approach is to compare SIFT descriptors of the subject tooth and the reference tooth so as to calculate tooth-to-tooth matching score. Then the various tooth-to-tooth matching scores are integrated into record-to-record matching score. The experimental results on a database of over 100 dental records, encountering more than 1200 teeth comparisons, show that the proposed approach achieves 74% accuracy compared to 43% using previous methods suggested in the literature.

Important Links:

Go Back