Optical grading of satsuma mandarin using linear discriminant function.
Morita K. | Shiga T. | Taharazako S.
Surface color and defect are the major factors in the present visual grading of fruits and those are to be estimated on the status of appearance. The theory of discriminant function is applied to calculate the value of statistical grading index according to the properties of spectral reflectance at the wavelengths of 680, 520 and 740 nm for detecting surface color and defect. Satsuma mandarin are attempted to be re-grade into four groups that have already been graded by human beings in the packinghouse. The UCS values (Lightness, Chroma and Hue) for the fruit surface are measured on both the blossom and stem ends, and visual grading of acceptable group is significant at the values of Chroma and Hue for each grade, but the spoiled grade is not significant at all the UCS values. Therefore, these three grades are noted to be capable of being graded by Hue and Chroma values but it is impossible for the spoiled grade to be sorted out from the other grades by the UCS values. The light reflectance at the wavelength of 680 nm is measured for detecting surface color and the light reflectance at 520 and 740 nm is measured for detecting surface defects at both ends. These twelve grading values from a fruit are applied to calculate the linear discriminant function for mechanical grading. After re-grading with this method, the measuring and grading errors of all factors (color and defect indices) of each grade become smaller as compared with the results that have been obtained from the visual grading at the packinghouse. This shows that this method can be used for automatic grading.Show more [+] Less [-]