OrthoNet: Multilayer Network Data Clustering

Description: Network data comes with some information about the network edges. In some cases, this network information can even be given with multiple views or layers, each one representing a different type of relationship between the network nodes. Increasingly often, network nodes also carry a feature vector. We propose in this paper to extend the node clustering problem, that commonly considers only the network information, to a problem where both the network information and the node features are considered together for learning a clustering-friendly representation of the feature space.

Associated journal paper: M. El Gheche, G. Chierchia, P. Frossard, Orthonet: Multilayer network data clustering, IEEE Transactions on Signal and Information Processing over Networks, Vol. 6, Pages 13-23, Dec. 2020.

Disparity estimation from multicomponent images under illumination variation

Description: Disparity map can be recovered by estimating the relative position of features in the stereo pair. This process is called stereo matching. Looking for correspondences between stereo images is a difficult task, because of the presence of hidden areas (i.e. occlusions) and because of the fact that the light is reflected differently depending on the viewing angle. In this toolbox, we adress the problem of stereo matching of multi-component images (e.g. color images) by jointly estimating the disparity and the illumination variation. A convex energy function that takes into account the illumination variation model is derived by resorting to a relaxation based on a first-order Taylor approximation around an initial estimate. This energy is then minimized while taking into consideration various convex constraints arising from prior knowledge and observed data. The proposed method allows us to incorporate various convex distances and it relies on the extension of the Parallel ProXimal Algorithm (PPXA).

Associated journal paper: C. Chaux, M. El Gheche, J. Farah, J.-C. Pesquet, and B. Pesquet- Popescu, A parallel proximal splitting method for disparity estimation from multicomponent images under illumination variation, Journal of Mathematical Imaging and Vision, vol. 47, no. 3, pages 167-178, 2013.

Description: In this toolbox, we consider the problem of disparity estimation from gray level stereo images without illumination variation.

Associated paper: M. El Gheche, J.-C. Pesquet, J. Farah, M. Kaaniche and B. Pesquet-Popescu, “Proximal splitting methods for depth estimation,” in ICASSP 2011, Prague, Czech Republic, 22-27 May 2011.