Event Driven Clustering Algorithm

TL;DR

The paper tackles real-time detection of small event clusters in asynchronous event-camera data. It introduces an asynchronous, event-driven clustering algorithm that leverages tempo-spatial distance and time-surface based decisions to form clusters, achieving a computational complexity of $O(n)$ independent of pixel grid size. The primary output is the set of cluster roots, each with root timestamp, root coordinates, end timestamp, total events, and contributing pixels, enabling immediate interpretation without reconstructing full clusters. Demonstrations on a 100 Hz signal generated by a lamp and captured by a neuromorphic camera illustrate robust root detection and real-time performance.

Abstract

This paper introduces a novel asynchronous, event-driven algorithm for real-time detection of small event clusters in event camera data. Like other hierarchical agglomerative clustering algorithms, the algorithm detects the event clusters based on their tempo-spatial distance. However, the algorithm leverages the special asynchronous data structure of event camera, and by a sophisticated, efficient and simple decision-making, enjoys a linear complexity of $O(n)$ where $n$ is the events amount. In addition, the run-time of the algorithm is independent with the dimensions of the pixels array.

Figures (2)

• Figure 3.1: Algorithm block diagram.
• Figure 4.1: The clusters of the periodic signal originated by the incandescent lamp. The purple X-signs mark the roots of the positive events clusters of the signal.