scispace - formally typeset
Search or ask a question
Posted ContentDOI

Event-based Vision Processors for Neuro-Vision Applications

22 May 2023-
TL;DR: A comprehensive overview of the remarkable progress achieved in the field of neuromorphic vision processors over the past few decades can be found in this paper , where two distinct types of vision processors, frame-based and event-based, are discussed.
Abstract: The concept of neuromorphic vision chips involves integrating an image sensor and brain-inspired parallel processors on a single silicon die, mimicking the human visual system with the image sensor acting as the retina and the processors as the brain's visual cortex. Compared to traditional image processing systems, neuromorphic vision chips can overcome serial data transmission and processing bottlenecks, performing image processing at high speeds of over 1000 frames per second. This makes them ideal for edge-computing applications that require quick response times and reduced communications bandwidth. This article presents a comprehensive overview of the remarkable progress achieved in the field of neuromorphic vision processors over the past few decades. Our primary focus is on two distinct types of vision processors: frame-based (FB) and event-based (EB) vision processors. These two categories differ in multiple aspects, including system architecture, image sensing techniques, image coding, image processing algorithms, and design methodologies.