A fast all-in-one method for automated post-processing of PIV data.

Area of Science:

• Fluid dynamics
• Data processing
• Image analysis

Background:

• Particle Image Velocimetry (PIV) data processing involves distinct validation, outlier replacement, and smoothing stages.
• Existing methods can be inefficient and struggle with clustered outliers or extensive missing data.

Purpose of the Study:

• To introduce a novel, unified post-processing technique for PIV data.
• To develop a fast, unsupervised method that integrates data validation, outlier replacement, and smoothing.

Main Methods:

• The proposed method, Discrete Cosine Transform-Penalized Least Squares (DCT-PLS), utilizes penalized least squares (PLS) and generalized cross-validation.
• It incorporates the discrete cosine transform (DCT) for efficient, multi-dimensional data smoothing.

Main Results:

• DCT-PLS demonstrated superior efficiency compared to conventional methods like the normalized median test.
• The method effectively handles clustered outliers and large amounts of missing data in PIV velocity fields.
• Its multi-dimensional capability makes it suitable for volumetric PIV data.

Conclusions:

• DCT-PLS offers a robust and efficient all-in-one solution for PIV data post-processing.
• The technique is versatile, applicable to both simulated and experimental data across various dimensions.