Mango Sorter by Machine Vision System

Abstract

This research applied machine vision system using the amount of selected colors (pixel count)to classify size, ripeness and defect of mangoes. The co-ordinates that represent the position of stem,end, front and back sides of the mango supplying detailed data to classify its shape and maturity.Every step of sorting process was controlled by computer software which was well designed andcreated on visual basic 6.0 program. The experiments were carried out with Nam Dok Mai mangoes.It was found that the software could be applied to estimate the size and shape. In comparisonbetween results from machine vision system and those from human classifying capability, 94.9percent coincidence for size and 80.3 percent coincidence for shape were obtained. For separating ofimmature mangoes, the ratio between a distance along the minor axis from stem to front of mango(W1) and a distance from stem to the end of mango (L) was used with 64.51 percent coincidence inclassification. For sorting ripe mangoes, the level of yellow area at 10 percent of total area was usedand 93.4 percent coincidence was found. For external defect inspection, the results from single-sidedinspection and those from double-sided inspection were in good agreement within 94.1 percent.

Keywords : Nam Dok Mai Mango / Sorting Machine / Machine Vision