Bio-inspired robust optical flow processor system for VLSI implementation

Specific areas in the brain are dedicated to process information coming from the retina, in order to detect and estimate the direction and speed of objects in the visual scene. This is the starting point to design and build models for motion estimation, taking cortical biology and physiology as inspiration, as evidence of the existence of spatial and temporal operators working in parallel in biological visual systems. Presented is an amended architecture of a neuromorphic robust optical flow based on reconfigurable hardware with properties of the cortical motion pathway. It delivers a useful framework for building complex bio-inspired real-time systems with high computational complexity. This architecture is customisable and adaptable, while emulating several neuromorphic properties, such as use of several information channels of small bitwidth. Obtained results show accuracy in several noisy environments.