Event vision sensor: principles, progress and prospects

Originating from neuromorphic engineering, event-based vision sensors mimic the function of eye rod cells to enable high dynamic range and temporal differential sensing. Unlike traditional cameras, these sensors detect temporal contrast changes, offering high temporal resolution, ultra-low latency, and low power consumption through sophisticated circuit designs. This unique combination has sparked widespread academic and industrial interest. Recent technological advancements, such as backside illumination (BSI) technology, wafer stacking, and standardized industrial interfaces, have significantly enhanced EVS performance-manifested in noise reduction, resolution enhancement, and increased readout speeds. These improvements also boost compatibility with existing edge vision systems, expanding practical applications. This paper reviews the evolution of event-based vision sensors from neuromorphic engineering to current state-of-the-art technologies, exploring their development trends, working principles, and key features. It also discusses the sensors' sensitivity and the opportunities and challenges they face in infrared imaging, offering valuable insights for future research and applications.