A spatio-temporal model for estimation and efficient tracking of dynamic boundaries

We consider the problem of minimising the number of measurements that sensors report when they are tracking dynamic boundaries or contours. Toward this, we consider spatiotemporal model for the evolution of a dynamic boundary where the y-coordinates are assumed to vary like correlated Brownian motion with drift. Sensors sample the points on the contour and when queried, they report the y-coordinate as measure by them and also the some samples estimates of the model parameters between the querying instants. Using the estimates of the model parameters, the sink interpolates the contour samples in time. For a time t, the estimates and the measured values from the queries is used in a non parametric regression to interpolate in space. The sensors are queried when the confidence in the estimate is below a specified thereshold. The performance of this estimation algorithm is compared with the case when the estimation algorithm does not assume the correlations spatial between the samples on the contour.