The use of biometrics, such as iris, thumbprints, voice, face, etc, for authentication have become commonplace. In this research, we are attempting to classify human faces using CTFM (Continuously Transmitted Frequency Modulated) sonar in place of the normally used computer vision. However, for this preliminary paper, the main objectives are to find out the range relationships between human facial features and how they are represented in echoes, and to test the quality of echo features for three faces from a single orientation. The tested features are features that were effective in past research into classifying objects from ultrasonic echoes. We measure the quality of these features using a minimum Euclidean distance criterion. Actual classification of faces will be attempted at a later phase.
Yoong, K. & McKerrow, P. J. 2005, 'Face recognition with CTFM sonar', in C. Sammut (eds), Australasian Conference on Robotics and Automation, Australian Robotics and Automation Association, Sydney, pp. 1-10.