Image analysis of whole grains: A noninvasive method to predict semolina yield in durum wheat

`Durum wheat grain samples (n = 327) harvested during 1998 and 1999 in eigh t different Italian agroclimatic areas and representative of the main Itali an cultivars were analyzed. Image analysis of whole grains allowed five siz e and shape descriptors (length of minor and major axes, perimeter, area, a nd ellipsoidal volume) to be measured on 100 grains for each sample. By mul tiple regression analysis, shape measures more valid to explain semolina yi eld were identified. By combining these measures with test weight or 1,000- seed weight, equations useful to predict semolina yield were developed.