Stochastic Geometric Filter and Its Application to Shape Estimation for Target Objects

We investigated how to estimate the shape of a target object. For this problem, we propose pair-line composite sensor nodes consisting of multiple sensors on a pair of line segments, where each sensor generates binary information whether it detects the target object or not. We show that the proposed pair-line composite sensor nodes, which are randomly placed, can detect a certain range of angles; therefore, we also call them stochastic geometric filters. By random distribution of pair-line composite sensor nodes without GPS functions or careful placement at known locations, the information sent from the nodes enables us to estimate the boundary angles of the target object as well as its size and perimeter length. A composite sensor node can be conceptualized as between a sensor node equipped with GPS functions, or carefully placed sensors at known locations, and randomly deployed simple sensors without GPS functions.