Detection of people and animals using non-imaging sensors

Abstract

Non-imaging sensors offer low power and long lasting solutions for perimeter, border crossing, and forward operating base protection. In this paper, we study the utility of acoustic, seismic, and ultrasonic transducers for detection and identification of people and animals. Various algorithms will be developed for them, which are computationally less intensive and amenable to implement on sensor network. We identify the physics-based phenomenology associated with the targets and the features selected for classification are based on the phenomenology. We fuse the results from various sensor modalities to achieve higher probability of correct classification.