Noël Bonnet

Learn More
Whereas estimating the number of clusters is directly involved in the ®rst steps of unsupervised classi®cation procedures , the problem still remains topical. In our attempt to propose a solution, we focalize on procedures that do not make any assumptions on the cluster shapes. Indeed the classi®cation approach we use is based on the estimation of the(More)
Although most histogram-based image segmentation methods rely on the identiÿcation of a good threshold, we show that thresholding is not mandatory. Instead, we propose the association of grades of membership to each individual pixel, in order to perform probabilistic relaxation in the image space (which realizes some kind of regularization) and ÿnally to(More)
At the present time, one of the best methods for semi-automatic image segmentation seems to be the approach based on the fuzzy connectedness principle. First, we identify some deficiencies of this approach and propose a way to improve it, through the introduction of competitive learning. Second, we propose a different approach, based on watersheds. We show(More)
We investigated whether cluster formation by noninvasive cells can be explained by a global attractive potential. Indices quantifying the persistence of migration in experimental conditions were compared to the same indexes computed from simulations with a density-based cellular automaton. The results indicate that the attractive potential hypothesis must(More)
A complementary approach is proposed for analysing series of electron energy-loss spectra that can be recorded with the spectrum-line technique, across an interface for instance. This approach, called blind source separation (BSS) or independent component analysis (ICA), complements two existing methods: the spatial difference approach and multivariate(More)