Pips image analysis to support cultivar identification of Vitis vinifera L

Identification of cropped cultivars, as well as wild relatives of any world spread agronomic crops, is a big challenge both for economical implications and for the safeguard of special products against fraudulent mystifications. This is more true for products that have achieved the European trademark attribution, in which the used raw material play a key role. Vitis vinifera L. is one of the most worldwide hedonistic crops with an huge multitude of trade name, in which the fundamental characteristic is the grape variety, the land of cultivation and the processing. With this aim, we have tried to develop a system of statistical identification, based on morphocolorimetric features measured using image analysis techniques, able to identify and classify some of the grapevine varieties of a living collection, developed and maintained in Sardinia island by Agricultural Research Agency of Sardinia (AGRIS), that moreover preserves ex situ the autochthonous grapevine varieties. Pips images were acquired at the Germplasm Bank of Sardinia (BG-SAR) using a flatbed scanner, and successively analysed at the Stazione Consorziale Sperimentale di Granicoltura per la Sicilia (SSG) using a specifically developed macro based on the image analysis software KS 400. The morpho-colorimetric data and elliptic Fourier descriptors obtained, were elaborated by Linear Discriminant Analysis to build a statistical classifier able to identify the cultivars and landraces through pip traits. Some attempts were done to compare the cropped varieties with some wild populations of Vitis sylvestris Blume of Sardinia island and some archaeological remains of pips founded in Sardinia too.