The relationship between mass and visual appearance of flowers that are used as volume and background fillers in bouquets was studied. The main objective was to test whether weighing flowers could be used as a simple method for sorting, equivalent to doing so by their visual appearance, which is the combined impression made by a flower's dimensions, number of stems and leaves, and their density. Two methods of evaluating the visual appearance of flowers were used: (1) a qualitative and subjective method using an averaged decision made by a panel of human inspectors; (2) a quantitative and objective method using a computer vision system to acquire and analyse two video images of each flower, from perpendicular directions. It was found that human vision could clearly classify flowers as ''small'' or ''big'' according to their visual appearance. Classifying the flowers into two mass-classes (''light'' and ''heavy'') closely matched the two-way human vision classification (''small'' and ''big''), based on the average decision of the panel of inspectors. On the basis of the averaged panel decisions, flowers of ''light'' and ''heavy'' classes looked more uniform when viewed separately than when viewed together. Flower mass correlated well with the average image area measured by the computer vision system. Based on these results it was concluded that the flowers studied could be classified by mass to achieve better uniformity of visual appearance. (C) 1996 Silsoe Research Institute
