Connected component labeling based on the EVM model
Dolors Ayala, J. Rodríguez, A. Aguilera · 2002
This paper presents a new approach to achieve connected component labeling on both binary images and volumes by using the Extreme Vertices Model (EVM), a representation model for orthogonal polyhedra, applied to digital images and volume datasets recently. In contrast with previous techniques, this method does not use a voxel-based approach but deals with the inner sections of the object. This approach allows us to build data size-independent algorithms and work with volumes from range data or solid modeling field indistinctly. Our proposal, also can be applied to manifold as well as non-manifold data. The algorithm actually detects non-manifold zones and permits to break or not the objects at these zones by an user-specified parameter.