A critical review of analogue image processing

Mead (1989) and co-workers have developed analogue circuit models of several neurobiological systems. Prominent amongst these models is one based upon the outer plexiform layer of the mammalian retina. This model mimics both the logarithmic response and the centre-surround receptive field observed in the retina. Although the analogue circuit was primarily based upon the mammalian retina, these two general features are shared by many and various retinas, including those of the Limulus and the fly. The widespread use of a common information processing strategy suggests that it leads to general benefits, which may also be useful in artificial systems. Our aim is to identify the potential benefits, and difficulties, in employing imaging sensors which share this common strategy. The starting point is the existing silicon retina architectures