An adaptive sampling and windowing interrogation method in PIV
An adaptive sampling and windowing interrogation method in PIV
复制标题
DOI：
10.1088/0957-0233/18/1/034
复制
发表时间：
2006
影响因子：
2.4
通讯作者：
R. Theunissen;F. Scarano;M. Riethmuller
中科院分区：
文献类型：
--
作者：
R. Theunissen;F. Scarano;M. Riethmuller
本研究提出了一种基于互相关的PIV图像询问算法，该算法根据图像特性和流动条件调整询问窗口的数量和大小。所提出的方法释放的约束，均匀的采样率（笛卡尔网格）和空间分辨率（均匀窗口大小）通常采用PIV询问。特别是在非最佳实验条件下，其中流动播种是不均匀的，这导致鲁棒性（每个窗口的颗粒太少）或测量精度（太大或间隔粗糙的询问窗口）的损失。研究了两个标准，即对图像中局部信号内容的适应和对局部流动条件的适应。描述了递归询问方法内的自适应准则的实现。询问窗口的位置和大小局部地适应于图像信号（即，接种密度）。此外，局部窗口间距（通常由重叠因子设置）与速度场的空间变化有关。该方法的可行性说明了两个实验的情况下，一个统一的审讯方法的局限性出现清楚：激波边界层相互作用和飞机涡尾。实例表明，空间采样率可以适应于实际的流动特征，并且可以布置询问窗口大小，以便遵循播种颗粒图像和流速波动的空间分布。与均匀询问技术相比，空间分辨率被局部增强，而在种子不良的区域中，分析的鲁棒性水平（信噪比）几乎保持恒定。
This study proposes a cross-correlation based PIV image interrogation algorithm that adapts the number of interrogation windows and their size to the image properties and to the flow conditions. The proposed methodology releases the constraint of uniform sampling rate (Cartesian mesh) and spatial resolution (uniform window size) commonly adopted in PIV interrogation. Especially in non-optimal experimental conditions where the flow seeding is inhomogeneous, this leads either to loss of robustness (too few particles per window) or measurement precision (too large or coarsely spaced interrogation windows). Two criteria are investigated, namely adaptation to the local signal content in the image and adaptation to local flow conditions. The implementation of the adaptive criteria within a recursive interrogation method is described. The location and size of the interrogation windows are locally adapted to the image signal (i.e., seeding density). Also the local window spacing (commonly set by the overlap factor) is put in relation with the spatial variation of the velocity field. The viability of the method is illustrated over two experimental cases where the limitation of a uniform interrogation approach appears clearly: a shock-wave–boundary layer interaction and an aircraft vortex wake. The examples show that the spatial sampling rate can be adapted to the actual flow features and that the interrogation window size can be arranged so as to follow the spatial distribution of seeding particle images and flow velocity fluctuations. In comparison with the uniform interrogation technique, the spatial resolution is locally enhanced while in poorly seeded regions the level of robustness of the analysis (signal-to-noise ratio) is kept almost constant.