Preprocessing network for multiple binary objects

This paper describes a simple network for a selective attention mechanism for multiple binary objects. The network is based on a concept similar to the region growing approach and is able to focus attention on one object at a time from multiple objects for a recognition process. The network consists of a growing network and an attention network. The growing network focuses on one object among others; while the attention network provides the "seed" for the region network to grow. Finally, a modification for "weightless" implementation is suggested.