融合事件相机–帧相机的高精度PTV–PIV技术研究

High spatiotemporal resolution PTV–PIV techniques via fusion of event- and frame- based cameras

  • 摘要: 为了提升传统帧相机–PTV/PIV技术在大粒子位移(数十个像素)流场测量中的粒子追踪精度和可靠性,引入基于动态视觉机制的事件相机,发挥其对不同速度粒子的时间自适应采样等优势,设计了一套融合事件相机–帧相机的高精度PTV–PIV技术。首先开发了事件数据增强的PTV算法,通过事件数据引导帧相机图像的粒子跟踪,以获得更高精度的粒子速度矢量;随后开发了事件数据增强的PTV–PIV融合算法,根据PTV粒子速度矢量修正帧相机图像PIV测量的流场结果,获得散度更低、精度更高的速度场。利用该技术对雷诺数为4400的自由射流进行了实验测量,粒子密度为0.00570.0127粒子/像素。结果显示,当粒子跨帧位移较大时,事件数据增强的PTV算法的大位移速度矢量(大于10个像素)数量远超帧相机图像的PTV矢量数量,速度矢量正确率大于40%,速度矢量的数量和精度远大于直接从帧相机粒子图像中获得的粒子追踪结果;相较于事件相机或帧相机图像的PIV速度场,事件数据增强的PTV–PIV融合算法修正的速度场离群速度矢量数量更少,平均速度散度不超过0.06 s−1,具有更高的流场重构精度。高精度PTV–PIV技术特别适用于分析粒子跨帧位移较大的流动,有效提高测速的稳健性和可靠性,也为高精度双曝光模式PIV技术提供了优化方向,在推动实验流体研究发展方面具有巨大的潜力。

     

    Abstract: To enhance the accuracy and reliability of PTV/PIV techniques based on traditional frame-based cameras for particles with inter-frame displacements reaching tens of pixels, a high-precision PTV–PIV technique fusing event- and frame- based cameras was designed by introducing an event-based camera with dynamic vision mechanism, leveraging its advantages of high-response time-adaptive sampling of particles at varying velocities. An event-data enhanced PTV algorithm was developed, which can guide particle tracking in frame-based camera images using event-based data to achieve higher-precision particle velocity vectors. An event-data enhanced PTV–PIV fusion algorithm was developed, which can correct the flow field results from PIV measurements in frame-based camera images based on PTV velocity vectors, yielding velocity fields with lower divergence and higher accuracy. This technique was applied to experimental measurements of a free jet with a Reynolds number of 4400, featuring particle densities of 0.0057 and 0.0127 particles per pixel. Results indicate that when particles exhibit significant inter-frame displacement, the event-data enhanced PTV algorithm generates substantially more high-displacement velocity vectors (exceeding 10 pixels) than those oblained by frame-based camera PTV algorithm. Particle matching accuracy consistently exceeds 40%, with both the quantity and precision of velocity vectors far surpassing particle tracking results derived directly from frame camera. Compared to the PIV velocity field from event- and frame- based camera images, the PTV–PIV fusion algorithm produced a velocity field with fewer outlier velocity vectors and an average velocity divergence not exceeding 0.06 s−1, demonstrating higher accuracy in flow field measurement. This high-precision PTV–PIV technique is suitable for analyzing flows with significant inter-frame particle displacements, enhancing the robustness and reliability of PIV. It provides an optimization direction for the high-precision dual-exposure PIV and holds great potential for advancing experimental fluid mechanics.

     

/

返回文章
返回