1-bit walker recognition with distributed binary pyroelectric sensors

This paper presents a wireless compressive pyroelectric sensor system for walker recognition. The research aims to make wireless distributed pyroelectric sensors a low-cost, low-data-throughput alternative to the expensive infrared video sensors in behavioral biometric applications. The compressive measurements are achieved by using both (1) non-uniform sensor sampling structures and (2) random projection based sensing protocols. As a result, the gait biometric feature can be represented by a 1-bit data stream. Both decentralized and centralized data fusion schemes are developed to improve the recognition speed and accuracy. A prototype of wireless pyroelectric sensor system has been developed to demonstrate the discrimination capability of the proposed 1-bit walker recognition system for a small group of subjects.