Abstract
The genetic algorithm (GA) had been already effectively applied by the present authors to the temporal particle pairing problem of the two-dimensional PTV (Particle Tracking Velocimetry) system software. The algorithm has been further refined and is now applied to the spatial particle pairing problem of the stereoscopic three-dimensional PTV system software. Every individual particle in the left-camera image should be correctly paired with the most probable partner in the right-camera image. The basic methodology of this spatial particle pairing is a classical epipolar line analysis and the object (fitness) function of the present genetic algorithm also uses the concept of this type of analysis. As a result of the performance check using synthetic PIV images, the new algorithm is capable of increasing the number of correct particle paring by a factor of 8 to 22% with respect to the classical epipolar line analysis method.
