Machine vision for non destructive evaluation of fruit quality

Abstract

The aim of this paper is to show preliminary results in developing image analysis techniques to measure some parameters that define the quality of different fruits and vegetables: size, colour, external defects and stem location. Because of the high efficiency of the segmentation procedure, estimation of size is solved for the fruit varieties in which experiments were performed. The partial coloured areas that appear on some varieties of peaches are also properly located. Greenish to red colour co-ordinates estimated from the vision system in tomatoes are well correlated with respect to standard colorimeter values. Damaged area is properly detected in oranges. Exact location of the stem is achieved in oranges and peaches.