Non-rigid object localization from color model using mean shift

This paper deals with non-rigid object localization in an image, from object colors. Our method allows detection in an image of all the objects, which correspond to a color model, without a priori information about their number. Our approach consists in creating a binary image, which represents the repartition of the most probable pixels to be part of the object. Considering this image as a cluster in R², the object localization is done by finding all the cluster modes. This search is carried out by applying a statistical method: the mean shift procedure. To illustrate our approach, we use sport images, from which we try to detect all the players.