Issue No. 05 - October (1996 vol. 16)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/40.540078
Retinomorphic vision systems use neurobiological principles to accomplish all four major operations performed by biological retinae: (1) continuous sensing for detection, (2) local automatic gain control for amplification, (3) spatiotemporal bandpass filtering for preprocessing, and (4) adaptive sampling for quantization, and they perform all four operations at the pixel level. The retinomorphic system I describe here uses a random-access communication channel to read out asynchronous pulse trains from a 64 64 pixel array in the retinomorphic chip, and transmits them to corresponding locations on a second chip that has a 64 64 array of integrators. Both chips are fully functional. I compare and contrast retinal design principles with the standard practice in imager design. I argue that neurobiological principles are best suited to perceptive systems that go beyond reproducing the dynamic scene, like a conventional video camera does, to extracting salient information in real time.
K. A. Boahen, "A Retinomorphic Vision System," in IEEE Micro, vol. 16, no. , pp. 30-39, 1996.