Use of machine vision system for evaluation of biocontrol potential of two antagonistic yeasts in combination with silicon against blue mold of apple fruit

Plant diseases severity is presently evaluated using manual tools such as ruler, digital ruler or by scaling. Extending the great potential of computer science in accurate and rapid determination of disease severity, this study was conducted to evaluate biological control potential of two yeasts, Pichia guilliermondii (A6) and Candida membranifaciens (A4) in combination with silicon (Si) at 0.1%, 0.3% and 0.5% for controlling Penicillium expansum P1 and P2 at 20 ˚C. The results showed that antagonistic yeasts controlled blue mold of apple significantly. Combination of the yeasts with Si improved the control of blue mold in comparison with yeasts alone. Statistical information obtained from the infected area in RGB channels showed that blue average (R2=0.70) was the most important and green standard deviation (R2=0.43) was the least important factors in recognition of disease severity. Linear discrimination analysis using blue average, green average, blue standard deviation, red standard deviation, green kurtosis and green standard deviation separated the disease severity statistical groups with the accuracy of 80%. The overall results of this study suggest that the color could be a powerful descriptor which could be applied for determination of disease severity in apple blue mold.