Studies on Automation of Works in Orchard (Part 1)

Detection of Apple by Pattern Recognition

Abstract

This paper reports detection of an apple using pattern recognition technology to automate the harvesting operation. The technique uses algorithm to recognize an apple, in which the original image of apple tree's canopy was initially taken into a personal computer, then transformed to a binary image by filtration and edge extraction. Calculation of variance was done on the image outline for correlatipon to a round shape. The varieties of apple used for this trial were “FUJI” and “ORIN”. Photos taken in the orchard were made with the direction of sunlight and with backlight. From the results, the algorithm programmed for detecting an individual fruit from the image of an apple tree was concluded to be proper regardless of apple varieties and light conditions.