An analysis framework for event-based sensor performance

Abstract

Event-Based Sensors (EBSs) are passive electro-optical (EO) imaging sensors which have read-out hardware that only outputs when and where temporal changes in scene brightness are detected. In the case of a static background and platform, this hardware ideally implements background clutter cancellation, leaving only moving object data to be read out. This data reduction leads to a bandwidth reduction, which is equivalent to increasing spatio-temporal resolution. This advantage can be exploited in multiple ways, using trade-offs between spatial and temporal resolution, and between spatial resolution and field-of-view. In this paper, we introduce the EBS concept and our previous experiments and analysis. We discuss important EBS properties, followed by discussion of applications where the EBS could provide significant benefit over conventional frame-based EO sensors. Finally, we present a method for analyzing EBS technology for specific applications (i.e. determine performance compared to conventional technology). This approach involves abstraction of EBS and conventional imaging technology and provides a way to determine the value of EBSs over conventional imaging technology for facilitating future EBS application development.