Distance maps for shape classification

The presented works show how to compute a parameter estimating the similarity degree of a given planar shape to some particular objects, by using distance maps. The particular objects are reference shapes which have to be detected in the images under study. This shape parameter works on well-defined 2D objects and allows their classification in different subsets, each of these shape families being defined according to a reference shape. The method is based on the computation of distance maps in order to determine two scale ratios, the one defining the greatest homothetic set of the reference shape included in the object under study and the other the greatest homothetic set of the object under study included in the reference shape. The ratio of these two scale ratios gives the estimation of the similarity degree between the two shapes. Furthermore, this method can also be extended to the third dimension, in order to classify 3D objects.