Linear Texture Segmentation Using Elastic Model Matching. Application for Geographic Map Understand

R. Mariani, M.P. Deseilligny, J. Labiche and R. Mullot

MATIS, IGN/DT/SR, 2, Av. Pasteur 94160 Saint Mande FRANCE
LACP, Faculté des Sciences de Caen, 14000 Caen
La3i-LACIS, UFR Sc et Tech., Universite de Rouen, 76281 Mont Saint Aignan

We develop an algorithm used to extract the linear textures present in a geographic map. Textures are explicitly described with attributed grammars and the segmentation of an observed input signal, provides the several textures it contains. Elastic model matching is used, including a priori knowledge management. Finally we provide an a posteriori verification tool.


GREC'97 program