Fuzzy attribute openings based on a new fuzzy connectivity class. Application to structural recognition in images

In problems such as image segmentation and recognition, the connectivity of target objects is a key feature. In the mathematical morphology framework, connected filters were derived from the classical connectivity theory but do not take into account the imperfections that can affect the image formation. The aim of this paper is twofold: (i) introduce a new class of connectivity for fuzzy objects and (ii) derive some associated attribute openings. We show also that the latters can be performed efficiently using a component-tree representation. We illustrate a potential use of these filters in a brain segmentation and recognition process.