Способ и устройство для автоматического распознания некондиционных клубней картофеля

Abstract

The aim of the article is to improve the quality and productivity of sorting by developing a method and an intelligent device for automatic recognition and inspection of substandard potato tubers. The article describes a prototype of an automatic sorting machine designed to recognize external defects in potato tubers and automatically inspect them with a jet of compressed air. The recognition process consisted of three main modules: segmentation, tracking a potato moving in a frame along a conveyor belt, and classification using a trained artificial neural network. For the segmentation of potato tubers against the background of the transporting conveyor in real time, a method based on the calculation of the color threshold was used. The centroid tracking algorithm was used to track moving potato tubers. To train the artificial neural network, we created our own dataset consisting of images of marketable and defective potato tubers. A prototype of an automatic sorting machine has been developed, which is based on the concept of intelligent data analysis, according to which the images of potato tubers obtained from a video camera are processed and formed into images with subsequent recognition and signaling to the executive device of the automatic inspection system in the form of a single pulse signal when determining the tuber as substandard.