An Improved Asynchronous Corner Detection and Corner Event Tracker for Event Cameras

Abstract

AbstractIn order to improve the efficiency of event-based corner detection and matching in high-speed UAV navigation and control task with event cameras, and to reduce computing costs in operation, this paper proposed an improved asynchronous corner detection algorithm and tracker for event cameras. The corner detection algorithm dynamically stores a fixed number of events on the surface of the detection of a continuous value of 0 arc, as a corner detection. The corner detection realizes the simplified description of the event angle, calculates the connection of the corner events by this descriptor, and matches them together in the form of particles. Through the evaluation on public Event Camera Datasets, our improved corner detection and tracking algorithm is more efficient than the existing algorithms, the performance under different motion is more stable.KeywordsAsynchronous corner trackingCorner detectionEvent cameras