Performance of a Robust Filter-based Approach for Contour Detection in Wireless Sensor Networks

A robust filter-based approach is proposed for wireless sensor networks for detecting contours of a signal distribution over a 2-dimensional region. The motivation for contour detection is derived from applications where the spatial distribution of a signal (such as temperature, soil moisture level, etc.) is to be determined over a large region with minimum communication cost. The proposed scheme applies multi-level quantization to the sensor signal values to artificially create an edge and then applies spatial filtering for edge detection. The spatial filter is localized and is based on an adaptation of the Prewitt filter used in image processing. Appropriate mechanisms are introduced that minimizes the cost for communication required for collaboration. Simulation results are presented to show the error performance of the proposed contour detection scheme and the associated communication cost (single-hop communications with immediate neighborhood in average) in the network.