Abstract In this work, we present an approach to reconstruct high-resolution flow velocity or scalar fields from sparse particle-based measurements such as particle tracking velocimetry, thermographic phosphors or pressure-sensitive particles. The proposed approach can be applied to any of those fields; without leading its generality, it is hereby assessed for flow velocity measurements. Particles allow probing physical quantities at multiple time instants in randomly located points in the investigated region. In previous works, it has been shown that high-resolution time-averaged fields can be estimated by an ensemble average of the particles contained into spatial bins whose size can be reduced almost ad libitum. In this work, high-resolution ensemble particle modes are estimated from the ensemble average of particles, weighted with Proper Orthogonal Decomposition time coefficients which are estimated from low-resolution spatially-averaged fields. These modes represent a self-tunable compressed-sensing library for the reconstruction of high-resolution fields. High-resolution instantaneous fields are then obtained from a linear combination of these modes times their respective time coefficients. This data-enhanced particle approach is assessed employing two DNS datasets: the wake of a cylinder and a fluidic pinball. It is shown here that it is possible to reconstruct phenomena whose characteristic wavelength is smaller than the mean particle spacing whenever such events are correlated with any other flow phenomenon with a wavelength large enough to be sampled. The proposed approach is also applied to experimental wind-tunnel data, again showing excellent performances in presence of realistic measurement noise conditions.

中文翻译：

---

从稀疏数据到高分辨率场：集合粒子模式作为高分辨率流动表征的基础

摘要 在这项工作中，我们提出了一种从基于稀疏粒子的测量（例如粒子跟踪测速、热成像荧光粉或压敏粒子）重建高分辨率流速或标量场的方法。所提出的方法可以应用于这些领域中的任何一个；在不引出其一般性的情况下，特此对流速测量进行评估。粒子允许在被调查区域中随机定位的点的多个时刻探测物理量。在以前的工作中，已经表明可以通过包含在空间箱中的粒子的集合平均值来估计高分辨率时间平均场，空间箱的尺寸几乎可以随意减小。在这项工作中，高分辨率的集合粒子模式是从粒子的集合平均值中估计出来的，用从低分辨率空间平均场估计的适当正交分解时间系数加权。这些模式代表了一个用于重建高分辨率场的自可调压缩传感库。然后从这些模式乘以它们各自的时间系数的线性组合获得高分辨率瞬时场。这种数据增强的粒子方法使用两个 DNS 数据集进行评估：圆柱尾流和流体弹球。这里表明，只要特征波长小于平均粒子间距的现象与具有足够大的波长以进行采样的任何其他流动现象相关，就可以重建这些现象。所提出的方法也适用于实验风洞数据，