Experimental Comparison between Event and Global Shutter Cameras.

Area of Science:

• Computer Vision
• Robotics
• Sensor Technology

Background:

• Event cameras offer asynchronous, low-latency data streams, contrasting with traditional frame-based cameras.
• Determining the specific application domains where event cameras outperform fast frame cameras remains an open research question.

Purpose of the Study:

• To experimentally identify application domains where event cameras surpass fast global shutter frame cameras.
• To quantify the performance differences between event and frame cameras under varying conditions.

Main Methods:

• Comparative analysis using controlled experiments (rotating disk) and a practical ballistic experiment (flying bullet).
• Evaluation metrics included sampling/detection rates, position estimation error, illuminance, and motion speed.
• Measurement of minimum pixel latency for commercial event cameras (ATIS, DVS240).

Main Results:

• Event cameras demonstrated superior bandwidth efficiency across all tested scenarios.
• Position estimation accuracy was comparable between event and frame cameras.
• Event cameras exhibited slower responses to positive contrast changes and were limited by pixel latency and sensor bandwidth in specific tasks.

Conclusions:

• Event cameras excel in bandwidth efficiency but face limitations in pixel latency and sensor bandwidth for certain applications.
• Future event camera generations may overcome current bandwidth constraints, expanding their application scope.