| |
Abstract:
We present an analog VLSI cellular architecture implementing
a simplified version of the Boundary Contour System (BCS) for
real-time focal-plane image processing. Inspired by neuromorphic
models across several layers of visual cortex, the design
integrates in each pixel the functions of simple cells, complex
cells, hyper-complex cells, and bipole cells, each tuned to one of
three orientations and interconnected on a hexagonal grid. Analog
current-mode CMOS circuits are used throughout to perform edge
detection, local inhibition, directionally selective long-range
diffusive kernels, and renormalizing global gain control.
Experimental results from a fabricated pixel prototype in 1.2 m
CMOS technology demonstrate the robustness of the BCS model in
selecting image contours in a cluttered and noisy
background.
|
