Fuzzy Features for Facial Shape Classification on Panoramic Dental Image

Abstract

Research on human face shape identification can assist forensic teams in reconstructing an unidentified victim’s facial features. Human face shape identification using panoramic dental imaging is suitable for use by forensic teams in identifying a large number of victims due to the teeth’s ability to withstand heat of up to 1,000°C. This study proposes an application for face shapes classification on panoramic dental image using fuzzy features and decision tree method. It will be used to assist forensic scientists in reconstructing unidentified people identifying numerous victims from their facial features. There are three classifications of human face shapes, namely oval, tapered, and square. The steps in this study are digitizing panoramic dental images into files; segmenting the upper jaw’s incisor teeth; then extracting the features by area, parameter, width, length, width-to-length ratio, area-to-parameter ratio, center_x and center_y. Fuzzy theory is used to convert numeric features into category features, while decision tree method will be used for training features. The experimental results show that the proposed method obtain accuracy 67% of 42 panoramic dental image.