Volume Simplification and Segmentation by 3D Connected Operators

An extension to connected operator filtering is proposed here for 3D still images (e.g. biomedical volumes). The "surface preservation" property of this kind of morphological processing can be exploited to obtain a "simplification" effect on the data, deriving from the cancellation of some undesired 3D structures; this is achieved by some filtering criteria which can be related to different objectives, varying from denoising to real structure removal purposes. We also show how the simplification process constitutes a useful pre-processing step to improve the performance of an automatic and/or assisted morphological segmentation of volumetric objects in the case of noisy, natural and complex data, such as MR and CT volumes.