In this paper, we present a general frame for a system of automatic modelling and recognition of 3D polyhedral objects. Such a system has many applications for robotics : recognition, localization, grasping,...Here we focus upon one main aspect of the system : when many images of one 3D object are taken from different unknown viewpoints, how to recognize those of them which represent the same aspect of the object ? Briefly, it is possible to determine automatically if two images are similar or not ? The two stages detailed in the paper are the matching of two images and the clustering of a set of images. Matching consists in finding the common features of two images while no information is known about the image contents, the motion or the calibration of the camera. Clustering consists in regrouping into sets the images representing a same aspect of the modeled objects. For both stages, expermiental results on real images are shown.