Particle Image Velocimetry System with Self-Organized Feature Map Algorithm

Abstract

Self-organized feature map algorithm and the classical particle tracking technique have been adopted together to analyze the single-exposure double-frame particle images for flow measurement. Similar to the normal correlation technique in particle image velocimetry, the whole region is divided into many small interrogation spots. Instead of applying the correlation algorithm to each of these spots to obtain their rigid translation, the self-organized feature map algorithm is used to compress the information such that every spot is represented by three coded equivalent particles. After tracking these three particles, a linear distributed velocity function can be obtained at every spot. The spot can contain not only translation, but also rotation, shear, and expansion while there is only rigid translation in the spot assumed in the commonly used correlation method. In addition to the theoretical explanation, the suggested method has been verified by a number of digital flow fields which have randomly distributed synthetic particles.