In-Pixel Time-Gating: Handling Activity in a PFM Event-Based Image Sensor with AER Readout
The spiking camera, which encodes light intensities into an asynchronous pulse stream, holds great promise for event-based and high dynamic range (HDR) applications. However, standard address event representation (AER) circuits can saturate due to intense pixel firing activity, creating greedy paths...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universidad de Sevilla (US) |
| Repositorio: | idUS. Depósito de Investigación de la Universidad de Sevilla |
| OAI Identifier: | oai:idus.us.es:11441/182453 |
| Acceso en línea: | https://hdl.handle.net/11441/182453 https://doi.org/10.1109/JSEN.2025.3638650 |
| Access Level: | acceso abierto |
| Palabra clave: | Address event representation (AER) Asynchronous operation Event-based vision sensor Image reconstruction Image sensor Spikes Spiking vision sensor |
| Sumario: | The spiking camera, which encodes light intensities into an asynchronous pulse stream, holds great promise for event-based and high dynamic range (HDR) applications. However, standard address event representation (AER) circuits can saturate due to intense pixel firing activity, creating greedy paths in arbitration trees that give exclusive attention to certain regions. This article addresses that challenge through a novel in-pixel spike filter, which blocks new spikes from being transmitted while the readout circuits process previously stored requests, effectively distributing the attention of the arbiter tree. To evaluate the technique, we designed and fabricated a 96 × 64 spiking image sensor in a standard 180-nm CMOS technology. Experimental results confirm that this time-gating mechanism ensures a sparser readout of the spike data, managing the activity load in the arbiters while preserving the original information of the spike data. Besides, the window indirectly extends the sensor’s DR and throughput by allowing higher spiking rates under high activity and short integration times without saturating the readout channel. Also, it provides a tunable tradeoff between data compression and image quality, opening new methods for event-driven applications requiring efficient data usage. |
|---|