When
Where
Title
Optical Characterization, Phenomenology, and Laser-Aided Applications of Event-Based Sensors
Abstract
Humanity’s desire for information and the necessary infrastructure to record and analyze it has been increasing exponentially in recent history. As our ability to collect this data has outpaced our ability to process and store it, we have begun looking to biomimetic design in hopes of identifying methods for more efficiently collecting pertinent information and throwing away the rest. Biologically inspired vision (i.e. event-based sensors) offers a more efficient alternative to conventional electro-optical sensing under high-speed operation; boasting a high-dynamic range that is comparable with the human eye, a sparse data bandwidth that allows for low-power operation, and an asynchronous readout which allows for exceptional temporal resolution. As with all new paradigms, this new class of sensors demands the development of new optical characterization techniques which can also be compatible with legacy methods and metrics. This work presents a new framework for evaluating the measurement of these nonlinear shift-invariant systems as well as a laboratory characterization method which permits direct comparison with traditional frame-based metrics for spatial resolution. Having shown how to optically characterize these systems, we then present an application which takes advantage of some of their greatest strengths (high temporal resolution, log-compression, and sparse-data bandwidth) while also shoring up several of their significant weaknesses. This application, event-based active imaging, utilizes an amplitude modulated continuous wave laser source to illuminate a scene at a set frequency, thus providing the event-based sensor with a reliable source of events that it can be synchronized to and use to further reject clutter and events due to unwanted fluctuations in ambient illumination. Further, we show that this technique can be optimized through an investigation of the influence of various spectral and temporal filters as well as the choice of modulation waveform. Finally, this work discusses our discovery of an emergent phenomenon that is unique to event-based sensors in the form of temporal super-resolution by way of nonuniform sampling with asynchronous pixel-clusters. This work sheds light on an entirely new avenue for applications of event-based sensors and showcases a powerful new analysis tool and its potential use for free-space optical communications, UAV detection/tracking, and compressive sensing schemes.
Please email Jini at jini@optics.arizona.edu or David at davidjbloom@arizona.edu for a Zoom link.