Existing algorithms for topology extraction focus on only one topology feature, either skeleton or segmentation, in 2D or 3D sensor networks, most of which requiring complete boundary information. As boundary information is not easily obtained, especially in sparse 3D sensor networks, and extracting these two features separately is very expensive, in this study, we propose to simultaneously extract the line-like skeleton of 2D/3D sensor networks and decompose the network into nice pieces, by constructing the Reeb graph. The Reeb graph has been envisioned as a powerful tool for encoding the topology of an object, where the key is to select the right function f. Without using boundary information, we first construct a cut graph, and then regard the distance of a node to the nearest cut as the function f such that the corresponding Reeb graph is pose independent, based on which the skeleton extraction and network decomposition are simultaneously conducted. Some simulation results are presented to show the efficiency of the algorithm.
