FPGA based Architecture for Robust Optical Flow Computation

Abstract

Motion Computation, also called Optical Flow, consists of measuring the motion of an entity attending to the modulus (how fast) and the phase (direction) of its movement. There are a plethora of different models and algorithms whereas none of them cover all problems associated to the real world. Our contribution presents a novel customizable architecture, including resource usage and performance, of a neuromorphic robust optical flow adjustable platform, although its main drawback is the elevated computational complexity. Bioinspiration and robustness properties of the final system provides an extended set of applications fields.