A PC-based machine vision system for obtaining three-dimensional parti
cle co-ordinates using a single camera is presented. Based on photogra
mmetric imaging, this system uniquely combines the advantages of stere
o and orthogonal views to determine the accurate locations of the part
icles. A 3D cross-correlation algorithm has been implemented to follow
the particles from frame to frame and compute the velocity vectors. E
xperiments have been carried out to obtain the velocity profiles in a
cubical test section with an offset inlet and outlet. The results obta
ined from the experimental investigation were compared with numerical
simulation data obtained from FLUENT and STAR-CD. (C) 1997 Elsevier Sc
ience Ltd.