Thorax biometrics from millimetre-wave images

Millimetre-wave imaging is an emerging technique for detection and discrimination of objects based on their radiometric signature. Recent applications focus on near range detection of metallic objects, such as weapons concealed under clothes and far range detection of vehicles or people. For outdoor scenarios, the human body is particularly well visible, with sufficient details that can be used for the purpose of biometrics. This paper discusses a method for human identification based on its signature in millimetre wavelength, specifically by analyzing the appearance of the upper body and thorax. Multilinear analysis is applied, which has been proposed for pose and expression invariant face recognition in visual images. Recognition by multilinear analysis of the thorax and face regions will be compared to recognition by a set of anthropometric distance measures. The thorax region is shown to give promising results on a dataset of 50 persons. Finally, the paper discusses a system for stand-off identification, combining visual localization and millimetre-wave biometrics.