Universal outlier detection for PIV data

By Jerry Westerweel, Fulvio Scarano. 2005

1,439 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computational Mechanics, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/4d6601b6336f4c0aede32c06307ed0e792966914

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computational Mechanics

In Statistics and Probability

Citations in the corpus

Evolution of the streamwise vortices generated between leading edge tubercles (2016, methodology): “Spurious velocity vectors or outliers were detected using a custom-written Matlab code which implemented the universal median technique described by Westerweel & Scarano (2005).”
Digital particle image velocimetry (DPIV) robust phase correlation (2009, methodology): “The erroneous vectors were determined using the Universal Outlier Detection [30] with a threshold of 3.”
Importance of the nozzle-exit boundary-layer state in subsonic turbulent jets (2018, methodology): “After each pass, a universal outlier detection (UOD) was applied on a 3× 3 grid to eliminate corrupted data and to enhance the particle-motion calculation (Westerweel & Scarano 2005).”
Experimental characterization of the agitation generated by bubbles rising at high Reynolds number (2009, background): “Average bubble velocity 〈V 〉 normalized by the velocity V0 of a single rising bubble as a function of the gas volume fraction α: filled symbols, present work; , Zenit et al. (2001); , Larue de Tournemine (2001); , Garnier et al.”
Three-dimensional instantaneous structure of a shock wave/turbulent boundary layer interaction (2009, methodology): “Post-processing involved identifying erroneous vectors using the universal median test (see Westerweel & Scarano 2005) with a maximum median deviation set at 2.5.”
Development and evaluation of gappy-POD as a data reconstruction technique for noisy PIV measurements in gas turbine combustors (2016, methodology): “Four different GPOD methods are implemented in this paper, three adapted from literature and one novel method that utilizes ideas from PIV post-processing (Everson and Sirovich, 1995; Gunes et al, 2006; Raben et al, 2012; Westerweel and Scarano, 2005).”
Effects of micro-ramps on a shock wave/turbulent boundary layer interaction (2009, methodology): “Spurious vectors were removed using the universal median test [36] and replaced by interpolation.”
On the use of helium-filled soap bubbles for large-scale tomographic PIV in wind tunnel experiments (2015, methodology): “The resulting vector field is post-processed with the universal outlier detection (Westerweel and Scarano 2005) and filtered with a second-order polynomial in the time domain with a kernel of seven snapshots.”
Three-dimensional vorticity patterns of cylinder wakes (2009, methodology): “After each interrogation the resulting data is validated by the universal median test (Westerweel and Scarano 2005), with an outlier detection rate of approximately 3%.”
Tomographic particle image velocimetry (2006, methodology): “Data validation based on a signal-to-noise ratio threshold of 1.2 and on the normalized median test with maximum threshold of 2 (Westerweel and Scarano 2005) returns 4% spurious vectors.”
Momentum and scalar transport at the turbulent/non-turbulent interface of a jet (2009, positive): “The fraction of spurious vectors in the final result is less than 2 %, which could be reliably detected by means of a median test (Westerweel 1994; Westerweel & Scarano 2005) and were replaced by linear interpolation.”
The accuracy of tomographic particle image velocimetry for measurements of a turbulent boundary layer (2011, methodology): “Fortunately, method for optimising the reconstruction of these sparely seeded volumes have been created (Worth and Nickels 2008; Atkinson and Soria 2009) that can reduce this time to as little as 5 minutes ( Atkinson and Soria 2009 ).”
Particle imaging techniques for volumetric three-component (3D3C) velocity measurements in microfluidics (2011, methodology): “An outlier detection (Westerweel and Scarano 2005) was used to remove strong outliers.”
Dense velocity reconstruction from tomographic PTV with material derivatives (2016, positive): “After each iteration, the vector fields are subject to universal outlier detection to remove spurious vectors (Westerweel and Scarano 2005).”
PIV uncertainty quantification by image matching (2013, methodology): “Recognizing and eliminating such incorrect vectors is a mandatory step to obtain unbiased velocity statistics and the procedure is referred to as data validation (Westerweel 1994, Westerweel and Scarano 2005).”