Ultrasound time mapping based human face identification system

Abstract

Some technologies on biometric application have implemented different methods and techniques to solve various problems on face recognition solely on computer vision domain. One of them is face recognition using ultrasound transducer. The ultrasound transducer has been used to solve some problems found in face recognition using computer vision technologies where the latest method is limited to many external factors such as the light effect surrounding the object. The earlier researches have proved that ultrasound transducer could be used to detect human face. However, those researches have not been developed to identify a person through his/her face. Therefore, this research has proposed a new human face recognition system using ultrasound transducer and feature to detect and identify a person through his/her face with mapping of ultrasound travel time. This face recognition application has been designed to move horizontally (left to right) and vertically (top to down) in order to scan the whole face and to create matrix data. The matrix data in 15 × 15 has been composed from the signal reflection from the face, thus this reflection also represents the distance of the sensor and the object. Later, the matrix data were analyzed using ANN and conventional template matching to identify the person/object. The research has proved that ANN has lower accuracy than conventional template matching.