Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Spike Imaging Velocimetry: Dense Motion Estimation of Fluids Using Spike Cameras

Yunzhong Zhang, Bo Xiong, You Zhou, Changqing Su, Zhen Cheng, Zhaofei Yu, Xun Cao, Tiejun Huang

TL;DR

This work tackles noninvasive, high-temporal-resolution flow measurement in turbulent regimes where traditional PIV struggles. It introduces Spike Imaging Velocimetry (SIV), a spike-camera–driven dense motion estimation framework equipped with three modules—Detail-Preserving Hierarchical Transform (DPHT), Graph Encoder (GE), and Multi-scale Velocity Refinement (MSVR)—and the spike-based PSSD dataset. On PSSD, SIV achieves state-of-the-art End-Point Error compared with both image-based and spike-based baselines, with ablations validating the contribution of each module. The study demonstrates the potential of spike cameras for high-temporal-resolution fluid velocity measurement and provides open-source data and implementation to advance PIV research.

Abstract

The need for accurate and non-intrusive flow measurement methods has led to the widespread adoption of Particle Image Velocimetry (PIV), a powerful diagnostic tool in fluid motion estimation. This study investigates the tremendous potential of spike cameras (a type of ultra-high-speed, high-dynamic-range camera) in PIV. We propose a deep learning framework, Spike Imaging Velocimetry (SIV), designed specifically for highly turbulent and intricate flow fields. To aggregate motion features from the spike stream while minimizing information loss, we incorporate a Detail-Preserving Hierarchical Transform (DPHT) module. Additionally, we introduce a Graph Encoder (GE) to extract contextual features from highly complex fluid flows. Furthermore, we present a spike-based PIV dataset, Particle Scenes with Spike and Displacement (PSSD), which provides labeled data for three challenging fluid dynamics scenarios. Our proposed method achieves superior performance compared to existing baseline methods on PSSD. The datasets and our implementation of SIV are open-sourced in the supplementary materials.

Spike Imaging Velocimetry: Dense Motion Estimation of Fluids Using Spike Cameras

TL;DR

This work tackles noninvasive, high-temporal-resolution flow measurement in turbulent regimes where traditional PIV struggles. It introduces Spike Imaging Velocimetry (SIV), a spike-camera–driven dense motion estimation framework equipped with three modules—Detail-Preserving Hierarchical Transform (DPHT), Graph Encoder (GE), and Multi-scale Velocity Refinement (MSVR)—and the spike-based PSSD dataset. On PSSD, SIV achieves state-of-the-art End-Point Error compared with both image-based and spike-based baselines, with ablations validating the contribution of each module. The study demonstrates the potential of spike cameras for high-temporal-resolution fluid velocity measurement and provides open-source data and implementation to advance PIV research.

Abstract

The need for accurate and non-intrusive flow measurement methods has led to the widespread adoption of Particle Image Velocimetry (PIV), a powerful diagnostic tool in fluid motion estimation. This study investigates the tremendous potential of spike cameras (a type of ultra-high-speed, high-dynamic-range camera) in PIV. We propose a deep learning framework, Spike Imaging Velocimetry (SIV), designed specifically for highly turbulent and intricate flow fields. To aggregate motion features from the spike stream while minimizing information loss, we incorporate a Detail-Preserving Hierarchical Transform (DPHT) module. Additionally, we introduce a Graph Encoder (GE) to extract contextual features from highly complex fluid flows. Furthermore, we present a spike-based PIV dataset, Particle Scenes with Spike and Displacement (PSSD), which provides labeled data for three challenging fluid dynamics scenarios. Our proposed method achieves superior performance compared to existing baseline methods on PSSD. The datasets and our implementation of SIV are open-sourced in the supplementary materials.

Paper Structure

This paper contains 20 sections, 13 equations, 5 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Spike-based Optical Flow Estimation
  4. Image-based PIV Algorithms
  5. Event-based PIV Algorithms
  6. Preliminary
  7. The Working Principle of the Spike Camera
  8. Data Generation
  9. Approaches
  10. Overall Architecture
  11. Detail-Preserving Hierarchical Transform
  12. Graph Encoder
  13. Multi-scale Iterative Optimizer
  14. Multi-scale Velocity Refinement
  15. Loss Function
  16. ...and 5 more sections

Figures (5)

  • Figure 1: A schematic of fluid velocity field estimation based on spike cameras, and a comparison of the complex velocity field estimated using our proposed SIV network with PIV-NetS pivnets and RAFT-PIV raftpiv.
  • Figure 2: Spike Imaging Velocimetry (SIV) network.
  • Figure 3: (a) Detail-Preserving Hierarchical Transform (DPHT) Module. (b) Graph Encoder (GE).
  • Figure 4: Multi-scale Velocity Refinement Module.
  • Figure 5: Visualization examples of different algorithms in Steady Turbulence (Problem 1), High-speed Flow (Problem 2), and HDR Scenes (Problem 3) are shown. For each problem, the red box in the top optical flow map and the green box in the bottom EPE error map indicate the same region of the flow field.