Adaptive spatiotemporal filtering by a neuromorphic model of the vertebrate retina

We propose a theoretical framework of the adaptive control of visual sensitivity performed by the vertebrate retina. Instead of a logarithmic function, the photoreceptor transfer function is modelled with a Michaelis-Menten law which has a more plausible biophysical correlate. We show that the neural and functional architecture of the retina supports the requirements for an optimal transcoding of non-stationary visual information. This control of visual sensitivity is done by using an adaptive transfer function whose parameters are spatiotemporally and locally estimated by the subsequent retinal circuit and fed back at the level of photoreceptors. Although the use of the resulting model remains limited in the context of digital image processing, it provides a good structural framework for an analog VLSI implementation of an adaptive spatiotemporal neuromorphic retina