Visual Person Localization with Dynamic Neural Fields: towards a Gesture Recognition System

For any visually-based interaction between persons and acting systems within a real-world environment the localization of a user by the system is a necessary condition. The presented work deals with this visual loca-lization problem of a user concretely referred to the autonomous mobile robot system MILVA of our department. Since this system is applied under real-world conditions especially for the localization some proper techniques are needed which have an adequate robustness. In our opinion, this requires the combination of several components of saliency towards a multi-cue approach, consisting of structure-and color-based features 2]. This paper introduces one of them: the localization based on a typical shape of contour. A simple contour shape prototype model consists of an arrangement of oriented lters doing a piecewise approximation of the upper shape (head, shoulder) of a frontally aligned person. Applying such lter arrangement in a multiresolution manner, this leads to a robust localization of frontally aligned persons even in depth. The central problem of selecting the most promising (salient) image region is treated by means of a three-dimensional dynamic neural eld performing a dynamic winner-take-all process (WTA, 1, 6]). After a successful localization of a person one can start a more detailed analysis of the gesture's meaning: besides the recognition of static gestures we also concentrate on the acquisition and later the recognition of dynamic gestures.