A NOVEL METHOD BASED ON THE OTSU THRESHOLD FOR INSTANTANEOUS ELIMINATION OF LIGHT REFLECTION IN PIV IMAGES
Abstract The base of particle image velocimetry (PIV) is the maximization of the correlation between the distribution of particle images in an interrogation window or a volume separated by an instant of time. In real images, the unwanted reflection of light on fixed walls or moving objects can directly interfere with the correlation, deteriorating the PIV quality. In this work, a new method for automatically generating instantaneous masks based on the Otsu threshold for instantaneous elimination of light reflection in PIV images is proposed. This method separates the saturated image caused by the unwanted scattering of light from the tracer particles images through the Otsu threshold combined with the Gauss filter and Wiener adaptive local filter. This new method, called Otsu-Gauss-Wiener (OGW), was first tested using synthetic PIV images. In these tests, the authors analyzed the reflection caused by an object regarding different sizes, shapes, and intensities to evaluate the performance of the proposed method. Later, the OGW method was tested in PIV experimental cases with real adversities, for example, scattering of light on a fixed wall in a channel with periodic hills (Case B – 4th PIV Challenge), strong reflection in a centrifugal impeller (Case C – 1st PIV Challenge) and light scattering caused by an out-of-plane motion of the diaphragm of a pulsatile pediatric ventricular assist device. The results shown that the method can remove the reflections by static and moving objects using an automatic mask generated for each instantaneous image.