ALGORITHMS FOR THE DETECTION OF CONNECTEDNESS AND THEIR NEURAL IMPLEMENTATION
Pieter R. Roelfsema, Sander M. Bohté, Henk Spekreijse · Series in Mathematical Biology and Medicine · 1999
Introduction Typical natural visual scenes contain many objects, which need to be segregated from each other and from the background. Components of a single object are usually related to each other by numerous perceptual grouping criteria, including colinearity and connectedness. The visual system is well adapted to exploit these grouping criteria in order to recover image regions occupied by one of the objects. Indeed, the visual system readily groups image regions together that are connected to one another @1 , as is illustrated in Fig. 1. Image elements that are connected to one another are easily segregated from other image regions. This is important, since other functions of the visual system, like object recognition, depend on proper segmentation of the visual image. In this chapter we will consider the physiological mechanisms that allow the visual system to detect connections among image regions. Minsky and Papert were among the first to recognize that connectedness detec