Computing Image and Motion with 3-D Memristive Grids

摘要: In this paper, we first present a biorealistic model for the part of early vision processing by incorporating memristive nanodevices. The architecture proposed network is based on organisation and functioning Outer Plexiform Layer (OPL) Inner (IPL) in vertebrate retina. non-linear adaptive response devices make them excellent building blocks realizing complex synaptic- like architectures that are common human We particularly show how hexagonal grids can be employed faithfully emulating smoothing effect occurring OPL to enhance dynamic range system. A memristor-based thresholding scheme detecting edges grayscale images, while evaluating system’s adaptability different lighting conditions fault tolerance capacity. then extend our work computing relative motion objects, which an important navigation task vertebrates routinely perform relying inherently unreliable biological cells Here, novel facilitates detection moving introduced. addition, double-layered 3-D modeling computations take place both IPL enables on-center off-center transient responses. Applying results, it shown possible generate estimation speed direction object.