Single-view volumetric PIV via high-resolution scanning, isotropic voxel restructuring and 3D least-squares matching (3D-LSM)

Abstract

Scanning PIV as introduced by Brücker (1995 Exp. Fluids 19 255–63, 1996a Appl. Sci. Res. 56 157–79) has been successfully applied in the last 20 years to different flow problems where the frame rate was sufficient to ensure a ‘frozen’ field condition. The limited number of parallel planes however leads typically to an under-sampling in the scan direction in depth; therefore, the spatial resolution in depth is typically considerably lower than the spatial resolution in the plane of the laser sheet (depth resolution = scan shift Δz ≫ pixel unit in object space). In addition, a partial volume averaging effect due to the thickness of the light sheet must be taken into account. Herein, the method is further developed using a high-resolution scanning in combination with a Gaussian regression technique to achieve an isotropic representation of the tracer particles in a voxel-based volume reconstruction with cuboidal voxels. This eliminates the partial volume averaging effect due to light sheet thickness and leads to comparable spatial resolution of the particle field reconstructions in x-, y- and z-axes. In addition, advantage of voxel-based processing with estimations of translation, rotation and shear/strain is taken by using a 3D least-squares matching method, well suited for reconstruction of grey-level pattern fields. The method is discussed in this paper and used to investigate the ring vortex instability at Re = 2500 within a measurement volume of roughly 75 × 75 × 50 mm3 with a spatial resolution of 100 µm/voxel (750 × 750 × 500 voxel elements). The volume has been scanned with a number of 100 light sheets and scan rates of 10 kHz. The results show the growth of the Tsai–Widnall azimuthal instabilities accompanied with a precession of the axis of the vortex ring. Prior to breakdown, secondary instabilities evolve along the core with streamwise oriented striations. The front stagnation point's streamwise distance to the core starts to decrease while the rear stagnation point distance remains constant which indicates that the front part of the ring is at first losing its mass during breakdown.

A scanning PIV method for fine-scale turbulence measurements

“…When all rays have been reconstructed, a resampling in planes of constant z (using a 10 × 10 Lanczos interpolation kernel) provides a reconstruction of the particle field on a regular cartesian grid. This extends the work of Brücker et al (2013) to consider a more generalised pinhole projection model and permit the use of a MART solver. For the stereo case, independent 2D reconstructions are performed on many different epipolar planes.…”

“…The solutions are first resampled using linear interpolation on epipolar planes with a regularly spaced grid (x, z) for arbitrary y and then resampled again to provide reconstructions on a regularly spaced Cartesian grid (x, y, z). This combines the multiple view problem discussed by Casey et al (2013) and Ponitz et al (2012) with the multiple sheet problem described by Brücker et al (2013). In addition, this method is not restricted to the use of a simple camera pinhole model.…”

“…To get around the limitations of conventional tomographic PIV, a number of authors have turned to laser scanning methods (Casey et al 2013;Ponitz et al 2012;Brücker et al 2013). The approach is to perform tomographic reconstructions of the 3D particle field as the first step, then use 3D reconstructions (discretised into cubic voxels) to perform a velocimetry.…”

“…The approach is to perform tomographic reconstructions of the 3D particle field as the first step, then use 3D reconstructions (discretised into cubic voxels) to perform a velocimetry. Brücker et al (2013) solve a tomographic reconstruction problem (although it is not expressed in such terms) for just one camera with many images taken at different laser sheet positions. Ponitz et al (2012) and Casey et al (2013) combine multiple-camera tomographic reconstructions independently calculated for each position of the (thin) laser sheet.…”

Particle image velocimetry measurement of indoor airflow field: A review of the technologies and applications

“…In general, adaptive PIV can achieve higher accuracy and spatial resolution results than adaptive correlation but consuming much more computing sources. Another advanced evaluation method potential for indoor applications is 2D or 3D Least Squares Matching (LSM) [47][48]. In contrast to correlation based methods, LSM is a gray-level tracking technique, which performs translation, deformation and rotation of the IA [47].…”