POD-based background removal for Particle Image Velocimetry

We propose a novel image preprocessing method for Particle Image Velocimetry (PIV), to remove background noise sources such as time dependent light reflection, light nonuniformities or camera dark noise. The method is based on the Proper Orthogonal Decomposition (POD) of the video sequence. In particular, it is shown that the PIV particle pattern is recovered by filtering out few of the first POD modes of the video, which are representative of typical background noise sources in PIV. After describing the theoretical framework of the proposed POD filter, the method is tested on synthetic and experimental images, and compared with well-known pre-processing techniques in terms of image amelioration, computational cost, and improvements in the PIV interrogation. The results show that the proposed method is insensitive to background noise intensity, gradients, and temporal oscillations contrary to existing methods. The computational cost is orders of magnitude lower than popular image recontrasting techniques such as CLAHE or min/max.
Reference
POD-based background removal for Particle Image Velocimetry
M.A. Mendez, M. Raiola, A. Masullo, S. Discetti, A. Ianiro, R. Theunissen, J.-M.Buchlin
Experimental Thermal and Fluid Science, Volume 80, January 2017, Pages 181–192.

Download the paper:
http://seis.bris.ac.uk/~aexrt/PIVPODPreprocessing/paper.html

For more information:
http://seis.bris.ac.uk/~aexrt/PIVPODPreprocessing/index.html