88E-20 |
Machine vision quantification of color changes in blanched, treated, and dried apricots (Prunus armenica L.). |
N. DEMIR, A. Z. Odabasi, and M. O. Balaban. Food Science and Human Nutrition Department, University of Florida, P.O. Box 110370, Gainesville, FL 32611 Color changes in fruits and vegetables can be quantified using color machine vision. Discriminant Function Analysis (DFA) can classify samples by color. This rapid, non-destructive method works even when color and surface is not uniform. The objective was to measure color changes of apricots blanched and treated with different solutions and air dried, using a computer vision system, and analyze color data and average L a b values using DFA to classify treatments. Fresh apricots were dipped in different solutions for 1.5 hr (200ppm Sodium Meta Bisulfate (SMB), 1% Ascorbic acid (AA), 100 ppm 4-Hexylresorcinol (4HR), 0.75% citric acid (CA), combinations of water-honey (1:1), water-sugar (1:1); pure water (control)). Some samples were blanched (2 min at 65 ºC), cooled, dipped in sugar solution for 1.5 hr, others were kept as controls. All samples were air-dried (40% RH, 56ºC) for 10 hr. RGB colors of images captured before and after dipping, and during drying were measured. Treatments and drying times were used as grouping variables. Correct classification rates (CCR) were calculated. Cross validation analysis was performed. Average L a b values and % color changes showed that the best color preservation was obtained with blanching and sugar (1:1) dip followed by blanching only. 4HR dip resulted in colors close to controls. AA samples resulted in colors darker than controls; samples dipped in CA and SMB better preserved color. CCR for all treatments will be presented (between 75 and 100%). DFA of color data for drying time gave CCR of 89%, 95%, 99%, 95%, 99%, 98%, 96%, 97%, 95%, 98%, 91%, 95% for treatments. Cross-validation results for different treatments and drying times were calculated. This method can rapidly and quantitatively determine color changes. DFA can classify individual color data and average L a b values by different treatments.
Session 88E, Fruit & Vegetable Products: Processing
|