Multi sensors based ultrasonic human face identification: Experiment and analysis

Abstract

This paper presents an ultrasonic sensing based human face identification approach. As a biometric identification method, ultrasonic sensing could detect the geometric structure of faces without being affected by the illumination of the environment. Multi ultrasonic sensors are used for data collection. Continuous Transmitted Frequency Modulated (CTFM) signal is chosen as the detection signal. High Resolution Range Profile (HRRP) is extracted from the echo signal as the feature and a K nearest neighbor (KNN) classifier is used for the face classification. Data fusion is applied to improve the performance for identifying faces with multi facial expressions. Experimental results show a success rate of more than 96.9% when the test database includes 62 persons and 5 facial expressions for each person. The results prove that multi sensors ultrasonic sensing could be a potential competent face identification solution for many applications.