RECOGNITION OF PARTIALLY OVERLAPPED PARTICLE IMAGES USING THE KOHONENNEURAL-NETWORK

A neural network is proposed for the recognition of partially overlapp ed particle images in the analysis of Particle Tracking Velocimetry (P TV) frames. The Kohonen neural network is an approximation to an optim um classifier. In this work it allows single particle images to be dis tinguished from overlapped particle images by shape analysis: it class ifies 99.1% of the spots correctly (in test images). If a spot has an almost circular shape, the barycenter co-ordinates are extracted. If t he spot shape is far from being circular, it is believed to be a parti cle overlap, and a procedure to find more centroids is activated. The particle recognizer based on the Kohonen neural network is tested on b oth multi-exposed and single-exposure images at high particle density, and compared to a particle recognizer that did not consider the parti al overlap. The management of overlapped particles causes the neural n etwork to produce a big improvement in the number of barycenters that can be extracted from these images. The practical consequence is that the seeding density in PTV can be increased, so as to improve the spat ial resolution of the technique in the velocity field calculation.