76C-21 |
Machine vision evaluation of vibration and transfer-related bloom loss in 'Misty' blueberries |
N. DEMIR, Food Science & Human Nutrition, University of Florida, P. O. Box 110370, 359 FSHN Bldg., Newell Drive, Gainesville, FL 32611-0370, C. O. Ferraz, FEAGRI, State University of Campinas, Caixa Postal 6011, Campinas, Sao Paulo, 13089-970, Brazil, S. A. Sargent, Horticultural Sciences, University of Florida, Institute of Food and Agricultural Sciences, Gainesville, FL 32611-0690, M. C. Andrade, Centro de Desenvolvimento de Tecnologia Nuclear, Belo Horizonte, MG, Brazil, and M. O. Balaban, Food Science and Human Nutrition, University of Florida, P. O. Box 110370, 359 FSHN Bldg., Newell Drive, Gainesville, FL 32611-0370. Blueberries may be bruised and lose bloom (velvet-like appearance of surface) during handling. Machine vision (MV) can rapidly, non-destructively detect changes in non-uniform surface colors. Our objectives were to correctly classify changes in bloom of blueberries subjected to various mechanical treatments using a MV system and Discriminant Function Analysis (DFA). Bucket transfer treatment: 21 hand-picked ¢ Misty ¢ blueberries (Vaccinium corymbosum) were marked with a white dot on the equator, and randomly mixed among 2 lb of berries in a plastic bucket. The berries were dropped 20 cm from the bucket onto a plastic surface. They were then subjected to 0, 2, 4, 6 and 8 transfers from one bucket to another. After each transfer the marked berries were separated, placed on a white surface with the mark pointing down, and their video picture was taken inside a light box. Three replicates were performed. Vibration treatment: 21 berries were exposed to 0, 20, 40, 60 and 80 seconds of vibration in a special assembly to simulate over-the road transport (15 Hz, 1 mm amplitude). After each treatment, berries were removed and a video picture was taken for four replications. Latex gloves were used in handling the berries. A 512-blocks color analysis was performed, 23 colors above 1% of total area were chosen. In DFA, treatments were used as grouping variables, and color spectrum as independent variables. Correct classification rates (CCR) and cross-validation analyses were performed. Average L a b values and % colors showed that there were changes in bloom during transfer and vibration treatments. DFA resulted in 100% CCR using number of transfers as grouping variable. DFA of vibration data grouped by vibration times gave 80% CCR. DFA and MV can be satisfactorily combined to detect and quantify bloom loss following mechanical treatments.
Session 76C, Fruit & Vegetable Product: Fresh Fruits and Vegetables
|