Japanese blue honeysuckle (Lonicera caerulea L. var. emphyllocalyx Nakai) is a unique form of edible honeysuckle that has exceptionally tasty berries. Visual characteristics of the berry such as the color of the skin and the presence of defects are the most decisive factors in determining its quality. An image analysis based methodology for classifying the berries under uncontrolled outside lighting conditions was developed. A color sheet with hue value around 29° was determined as the background to support the berries whose hue values were found near 212°. With the thresholding level computed by Otsus algorithm in the red channel, berries were segmented from the background successfully. Three parameters, average and standard deviation of hue component and average of saturation component, were chosen as the best descriptions for each berry according to the Fishers least significant differences test. Three canonical functions and corresponding group centroids of each function obtained by discriminant analysis were able to classify the berries aimed for fresh market, processing, and waste at success rates of 95.1%, 85.1%, and 94.3%, respectively.
No takes yet. Share an insight, caveat, or question.
Fu et al. (2011) studied this question.