We consider the situation where a set of similarly looking particles is analysed and where it may be assumed that the observed particles have been obtained from a common axial-symmetric parametric prototype form class by individual random deformation. An identification rule is presented, which allows to choose between two different given shape-types based on a few easily to perform measurements. It turns out that the misclassification probabilities may be evaluated asymptotically provided that the number of particles in the set is large.