A temporally adaptive schlieren approach by fusing frame- and event-based cameras
Year: 2024
Type: article
Abstract: This Letter reports what we believe to be a novel schlieren approach with adaptive temporal resolution. The fundamental concept of this approach is to fuse an event-based camera and a low-speed frame-based camera to generate high-frame-rate videos by leveraging the strengths of both. Using a novel experimental setup, events and frames are accurately aligned in both space and time. The aligned data are then fed into a neural network to generate intermediate frames. To examine the proposed approach, tests are conducted on a plate heater, burning candle, and pulsed jet. It has been tested that this approach enables continuous visualization and recording of flows with an adaptive frame rate of up to 3.3 kFPS, breaking through the short operating times of the existing schlieren technique. The developed intermediate frame generation method also outperforms similar methods by minimizing the impact of the event camera readout latency. It achieves a maximum improvement of 3.99 dB in the peak signal-to-noise ratio.
Source: Optics Letters
DOI: 10.1364/ol.545691
Authors: Zhen Lyu, Weiwei Cai, Benlong Wang, Yingzheng Liu
Institution: Shanghai Jiao Tong University
Language: English
Topic: Advanced Vision and Imaging
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: G374944416, G3999536412
OpenAlex ID: W4405112874