Edge Detection Using Orientation-Independent Operators in Large-Scale Sensor Networks

Detecting and tracking boundaries of a physical process by a large-scale sensor networks are often desirable or even required in various applications. The first step in the detection and tracking procedures is edge detection. Most of the edge detection schemes proposed for sensor networks require information on sensor locations. Also, quantitative measures by which performance of an edge detector for sensor networks can be analyzed have not been available. In this paper, edge detectors which do not require the absolute or relative location of each sensor node are proposed along with a set of metrics used for performance analysis. Detailed performance analysis results obtained through simulation, varying parameters such as sensor density, communication range, and noise level, are provided.