Spatial acuity modulation of an address-event imager

We have implemented a foveated vision system that uses an 80×60 address-event neuromorphic image sensor. Its effective spatial acuity is varied throughout the visual field to use bandwidth more efficiently. Low resolution pixels are created by pooling spikes from neighboring photoreceptors using overlapping Gaussian kernels. The pooling is effected by routing address-events through a memory-based projective field mapping, and new pixels are created by summing inputs on an array of integrate-and-fire neurons. The fovea can be relocated without physically moving the imager by performing a simple computation in the address domain. We provide an example of the system in operation and compute a figure of merit to quantify foveation savings.