Classifying still faces with ultrasonic sensing

The echo of a chirp of ultrasonic energy from an object contains information about the geometry of that object: the relative depth of its surfaces and the approximate area of those surfaces. A human face has complex geometry that produces a distinctive echo. In this paper, we report the initial research into whether there is sufficient information in the echo to recognize a still face. Potential features for classification are identified using a facial model. The classification results for 10 faces encourage future research with a larger number of faces and with moving faces.