Premium
COLOR VISION SYSTEM FOR RIPENESS INSPECTION OF OIL PALM ELAEIS GUINEENSIS
Author(s) -
ABDULLAH MOHD Z.,
GUAN LIM C.,
MOHAMED ABDUL M. D.,
NOOR MOHD A. M.
Publication year - 2002
Publication title -
journal of food processing and preservation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.511
H-Index - 48
eISSN - 1745-4549
pISSN - 0145-8892
DOI - 10.1111/j.1745-4549.2002.tb00481.x
Subject(s) - ripeness , hue , machine vision , artificial intelligence , palm oil , palm , computer vision , elaeis guineensis , computer science , agricultural engineering , color space , mathematics , image processing , pattern recognition (psychology) , food science , engineering , chemistry , ripening , image (mathematics) , physics , quantum mechanics
In checking harvesting discipline and quality control for oil palm fruits, color has presumably been an important guide to whether the oil content has reached a maximum where the fruit bunch is ready for cutting. However, establishing a single and harmonious standard base on color is a very contentious issue in the oil palm industry because of the subjective nature of the human vision of color. This was further complicated due to the lack of information on fruit color upon which to base a definite ripeness criterion. We demonstrated in this paper that this problem can be solved using machine vision technology. Methods used were to treat color in HSI (Hue, Saturation and Intensity) color space and applied multivariate discriminant analysis. These have proven to be highly effective for color evaluation and image processing. The vision system was trained to classify oil palms into four quality grades according to PORIM (Palm Oil Research Institute of Malaysia) inspection standards. These are the unripe, the underripe, the optimally ripe and the overripe classes. Depending upon the quality feature evaluated, misclassification by the vision system varied from 5 to 12% but averaged at about 8%. Machine vision disagreement ranged from 2 to 19%.