An image processing technique was used to predict the size distribution of the high speed, fine droplets at downstream of an air blast atomizer. The spray visualization setup consisted of UV lamps as light source, a stroboscope for slowing down the droplet motion, and a digital camera to capture the droplet images. The experiments were carried out at different liquid flow rates with various nozzle diameters. Two key unknown parameters (spray half angle and dispersion angle) of the air blast atomizer model in Fluent were obtained from these experiments. Using the obtained parameters and other structural parameters, the spray modeling was performed, and the Rosin-Rammler distribution was obtained and compared with those obtained from image processing technique through a diagnostic matrix. Regarding the kappa value, the agreement between predictions of the Fluent model and the image processing technique was moderate.