Spike-TBR: a Noise Resilient Neuromorphic Event Representation

2025-06-05

AI Summary

To address the challenge of learning noise-robust representations from event camera data, this paper proposes Spike-TBR—a novel time-based binary representation framework that for the first time integrates dynamic thresholding and inhibition mechanisms of spiking neurons (LIF/Izhikevich) into the TBR paradigm. Spike-TBR converts asynchronous event streams into robust, frame-like representations via three core components: event-to-spike mapping, spatiotemporal adaptive noise gating, and TBR-based spatial aggregation—preserving salient spatiotemporal structure while suppressing asynchronous noise. Evaluated on multiple noisy event datasets, Spike-TBR achieves up to 12.3 dB SNR improvement over baselines; it maintains consistent performance gains even on clean data, demonstrating strong generalization and robustness. By bridging the representational gap between spike-domain and frame-domain vision, Spike-TBR establishes a new paradigm for integrating event cameras with standard frame-based visual models.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Computer Vision: Representation Learning for Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Security and Privacy: Privacy-enhancing technologies

Abstract

Event cameras offer significant advantages over traditional frame-based sensors, including higher temporal resolution, lower latency and dynamic range. However, efficiently converting event streams into formats compatible with standard computer vision pipelines remains a challenging problem, particularly in the presence of noise. In this paper, we propose Spike-TBR, a novel event-based encoding strategy based on Temporal Binary Representation (TBR), addressing its vulnerability to noise by integrating spiking neurons. Spike-TBR combines the frame-based advantages of TBR with the noise-filtering capabilities of spiking neural networks, creating a more robust representation of event streams. We evaluate four variants of Spike-TBR, each using different spiking neurons, across multiple datasets, demonstrating superior performance in noise-affected scenarios while improving the results on clean data. Our method bridges the gap between spike-based and frame-based processing, offering a simple noise-resilient solution for event-driven vision applications.

Problem

• Converting noisy event streams into standard vision formats
• Enhancing noise resilience in event-based encoding strategies
• Bridging spike-based and frame-based event processing

Innovation

• Spike-TBR integrates spiking neurons for noise resilience
• Combines TBR with spiking neural networks
• Bridges spike-based and frame-based processing