Euclidean graphs are widely used in various fields, such as computing sciences, space exploration, neural networks, etc. In many cases, the majority of these graphs have vertices randomly deployed. This random aspect can make the graph divided into disjoint connected components with a topology that needs to be built. We propose a new technique to find the connected components and their boundaries using only connectivity information between a vertex and their neighbors. We show by an extensive simulation that the algorithm gives good results with a random generation of Wireless Sensors Networks.