[Paper Review] Event-based Imaging Velocimetry -- An Assessment of Event-based Cameras for the Measurement of Fluid Flows

Christian Willert, Joachim Klinner|arXiv (Cornell University)|Feb 22, 2022

Advanced Memory and Neural Computing32 references54 citations

TL;DR

This paper proposes Event-Based Imaging Velocimetry (EBIV), a novel approach for fluid flow measurement using event-based cameras that capture microsecond-resolution intensity changes instead of frames. By leveraging continuous laser illumination and processing event streams via motion compensation and correlation-based algorithms, EBIV enables high temporal resolution, low-light-sensitive 2D-2C flow velocimetry at particle densities previously unachievable with EBV, demonstrating feasibility in water and air flows with real-time processing potential.

ABSTRACT

Contrary to conventional frame-based imaging, event-based vision (EBV) or dynamic vision sensing (DVS) asynchronously records binary signals of intensity changes for given pixels with microsecond resolution. The present work explores the possibilities of harnessing the potentials of event-based vision for fluid flow measurement. The described implementations of event-based imaging velocimetry (EBIV) rely on the imaging small particles that are illuminated by a laser light sheet which is similar to classical two-dimensional, two-component (2d-2c) PIV with the difference that a continuously operating laser-light sheet is used without modulation of the laser or camera. The moving particles generate continuous time-stamped events on the detector that are later used to infer their velocity using patch-wise processing schemes. Two flow estimation algorithms are proposed; one uses a "motion compensation" that maximizes the local contrast, the other is based on a sum-of-correlations approach. The underlying motion detection schemes along with the complete absence of background signal allows straightforward retrieval of the events associated with individual particles thereby allowing the reconstruction of individual particle tracks. Alternatively, the event data can be processed with conventional PIV algorithms using images reconstructed from the event data stream. The concepts are demonstrated on simple flows in water and air.

Motivation & Objective

• . To assess the feasibility of event-based vision (EBV) for planar 2D-2C fluid flow measurement.
• To evaluate EBIV performance at high seeding particle densities, exceeding those reported in prior EBV-based flow studies.
• To compare two event-based motion estimation algorithms—motion compensation and sum-of-correlations—for velocity field reconstruction.
• To demonstrate EBIV applicability in both water and air flows using continuous laser illumination without pulsed light.
• To explore the potential of EBIV for time-resolved 2D and 3D particle tracking and dense velocity field estimation.

Proposed method

• . Event-based cameras (SilkyEvCam EvC3A and Prophesee HD) capture asynchronous, time-stamped events from intensity changes in a laser-illuminated flow field.
• A continuous-wave (CW) laser illuminates micrometer-sized particles, generating a continuous stream of positive and negative events on the sensor.
• Two motion estimation algorithms are implemented: one maximizing local contrast for motion compensation, the other using sum-of-correlations across image patches.
• Event data is processed to reconstruct particle tracks and velocity fields, with alternative reconstruction of intensity images for compatibility with conventional PIV algorithms.
• The method enables dense velocity field estimation without frame-based acquisition, leveraging the high temporal resolution (1 µs) and dynamic range (>120 dB) of EBV sensors.
• GPU-accelerated processing enables real-time analysis with latency on the order of milliseconds, supporting real-time flow diagnostics.

Experimental results

Research questions

• RQ1 . Can event-based cameras achieve reliable 2D-2C fluid flow velocity measurements at high particle image densities using continuous laser illumination?
• RQ2 How do motion compensation and sum-of-correlations algorithms compare in accuracy and robustness for EBIV velocity estimation?
• RQ3 What are the limitations of EBIV in terms of velocity range, temporal resolution, and sensitivity to background noise or flicker?
• RQ4 Can individual particle tracks be reconstructed from event data with sufficient accuracy for particle tracking velocimetry (PTV) applications?
• RQ5 To what extent can EBIV replace or complement conventional PIV in high-speed, low-light, or surface-proximate flow measurements?

Key findings

• . EBIV successfully measures 2D-2C fluid flow velocities in both water and air at particle densities significantly higher than previously reported with event-based imaging.
• The motion compensation algorithm achieved higher accuracy in velocity estimation by maximizing local contrast, outperforming the sum-of-correlations method in complex flow regions.
• Event-based cameras with 1280×720 resolution and 4.86 µm pixels demonstrated effective particle tracking and velocity field reconstruction with temporal resolution down to 250 µs (equivalent to 4 kHz frame rate).
• The absence of background signal in event data enables direct and unambiguous identification of particle events, simplifying particle tracking and reducing noise-related errors.
• EBIV enables real-time processing with millisecond-level latency, making it suitable for online flow diagnostics and control applications.
• The technique is particularly advantageous in proximity to surfaces, where conventional PIV suffers from excessive light scatter, as static surfaces do not trigger events.