Related Experiment Video
Updated: Jun 7, 2026
07:34
A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions
Published on: March 25, 2014
S2E: Spatio-temporal filtering of spike streams for motion-selective event generation
Lingxiao Zheng1, Yajing Zheng2, Rui Zhao1
1State Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing, China.
Summary
This study introduces a novel spikes-to-events (S2E) pipeline for neuromorphic vision. It effectively denoises spike data and synthesizes high-quality events, improving motion contour clarity and reducing artifacts for single-sensor systems.
Area of Science:
- Neuromorphic Engineering
- Computer Vision
- Sensor Fusion
Background:
- Spike cameras capture luminance with micro-second precision.
- Event cameras detect motion-induced intensity changes.
- Existing spikes-to-events (S2E) methods suffer from noise amplification and miss slow variations.
Purpose of the Study:
- To develop an integrated S2E pipeline for emulating both foveal texture and peripheral motion pathways using a single spike sensor.
- To overcome limitations of existing S2E methods by reducing reconstruction noise and capturing slow intensity variations.
Main Methods:
- A double-stage intensity estimation using a Markov-random-field-based spatio-temporal filter to denoise spike streams.
- Spike-driven log-differencing for synthesizing high-quality events at microsecond resolution.
Main Results:
- The proposed S2E pipeline generates event streams with clearer motion contours.
- Substantially lower background artifacts compared to raw event cameras and previous S2E approaches.
- Demonstrated effectiveness on PKU-Vidar-DVS, motVidarReal2020, and momVidarReal2021 datasets.
Conclusions:
- The developed framework offers a practical solution for single-sensor neuromorphic vision systems.
- Unifies foveal intensity reconstruction and peripheral motion detection on a single spike-camera platform.
- Enables more robust and versatile neuromorphic sensing capabilities.
