Silicon retina with correlation-based, velocity-tuned pixels

Abstract

I report the first functional two-dimensional silicon retina that computes a complete set of local direction-selective outputs. The chip motion computation uses unidirectional delay lines as tuned filters for moving edges. Photoreceptors detect local changes in image intensity, and the outputs from these photoreceptors are coupled into the delay line, where they propagate with a particular speed in one direction. If the velocity of the moving edges matches that of the delay line, then the signal on the delay line is reinforced. The output of each pixel is the power in the delay fine signal, computed within each pixel. This power computation provides the essential nonlinearity for velocity-selectivity. The delay line architecture differs from the usual pairwise correlation models in that motion information is aggregated over an extended spatiotemporal range. As a result, the detectors are sensitive to motion over a wide range of spatial frequencies. I have designed and tested functional one- and two-dimensional silicon retinas with direction-selective, velocity-tuned pixels. A velocity-selective detector requires only a single delay element and nonlinearity for each tuned velocity, and is sensitive to both light and dark contrasts. The use of adaptive photoreceptors and compact circuits makes for a well-conditioned input signal and small circuit offsets, resulting in robust operation. All circuits work in subthreshold, resulting in low power consumption. Pixels with three hexagonal directions of motion selectivity are approximately (225 mum)2 area in a 2-mum CMOS technology, and consume less than 5 muW of power.