Jptical-electronic system for apple sorting line: development and implementation

Abstract

This article covers the relevance of developing a technical tool for automatic sorting of apples, particularly, its main element – an optical-electronic system. The conducted research showed that in order to sort harvested apples into classes, it is important not only to determine the size of the apples, but also the presence of defects on them. The existing fruit sorting machines do not implement abovementioned functions, which leads to additional manual labor costs, reduces in productivity and quality. The article provides a classification of fruit quality recognition systems and substantiates the structural and technological scheme of the optical-electronic system. The system consists of an illuminated optical module and a video camera, an electronic control unit and a fruit conveyor. In the course of the research there were substantiated a single-threaded type of fruit flow with forced rotation and a structural and technological scheme of the optical-electronic system. Also, based on a color segmentation algorithm, a tracking algorithm and deep learning of an artificial neural network that provide detection of features, determination of the diameter and recognition of fruit defects from mechanical damage, pests and diseases and sorting them into three commercial classes, there were developed an algorithm of the operation of the system together with the software. The developed system was implemented in LSP-4 apple sorting and packaging machine, which successfully passed acceptance tests and production testing at Ostromechevo OJSC. LSP-4 machine acceptance tests revealed that it provides an accuracy of sorting by size – up to 75.4%, by the presence of defects – up to 73.1% and by labor productivity – up to 1.8 ton/hour.