Multiple objects segmentation with fuzzy rule-base trained topology adaptive active membrane

Abstract

Segmentation of multiple objects in a scene from the single initialization of an active membrane is always advantageous compared to separate initialization of active contour for segmenting each of the multiple objects. This proposal, however, is not robust in segmenting poorly-contrasted touching objects especially when pixel groups belonging to a single object can have spectral signatures similar to the background pixels. In this paper we have used fuzzy rule based learning scheme to record the spectral signature of the objects and background and spatial information of the topology of an active membrane segmenting the objects. The learning scheme helps in splitting the active membrane for segmenting multiple objects and integrates the topology adaptive property of the active membrane with its architecture and evolution. The evolution of this membrane is tested in a challenging application domain of estimation of sizes of oil sand.