This study aimed to demonstrate the practical and low-cost usability of image processing techniques in agricultural grain products by making a sample application in confectionery sunflower seeds. A desktop scanner and an open source software ImageJ were used. Local sunflower cultivars, which were obtained from Manisa, Denizli and Kayseri provinces, were used as a material. After the seeds were scanned with a scanner, their projected areas were calculated in 2D with ImageJ software. Afterward, the dimensions of the seeds were measured with a precision scale and caliper to compare with their projected areas. As a result of the statistical analysis, very high correlations could not be reached between the calculated projected areas and the other measured dimensions. Seed weight, seed length, seed width and hull weight were found to be highly or moderately correlated with the projected area values in all three seed types. Research findings show that the most important reason for not finding very high correlations was the use of local cultivars and the low level of seed purity.