Image segmentation has many applications which range from machine learning to\nmedical diagnosis. In this paper, we propose a framework for the segmentation\nof images based on super-pixels and algorithms for community identification in\ngraphs. The super-pixel pre-segmentation step reduces the number of nodes in\nthe graph, rendering the method the ability to process large images. Moreover,\ncommunity detection algorithms provide more accurate segmentation than\ntraditional approaches, such as those based on spectral graph partition. We\nalso compare our method with two algorithms: a) the graph-based approach by\nFelzenszwalb and Huttenlocher and b) the contour-based method by Arbelaez.\nResults have shown that our method provides more precise segmentation and is\nfaster than both of them.\n