In this paper we present a biorealistic model for the first part of the earlyvision processing by incorporating memristive nanodevices. The architecture ofthe proposed network is based on the organisation and functioning of the outerplexiform layer (OPL) in the vertebrate retina. We demonstrate that memristivedevices are indeed a valuable building block for neuromorphic architectures, astheir highly non-linear and adaptive response could be exploited forestablishing ultra-dense networks with similar dynamics to their biologicalcounterparts. We particularly show that hexagonal memristive grids can beemployed for faithfully emulating the smoothing-effect occurring at the OPL forenhancing the dynamic range of the system. In addition, we employ amemristor-based thresholding scheme for detecting the edges of grayscaleimages, while the proposed system is also evaluated for its adaptation andfault tolerance capacity against different light or noise conditions as well asdistinct device yields.
展开▼