DETECTION OF POTATO TUBERS AGAINST THE BACKGROUND OF SOIL

The paper considers the problem of segmentation of an image containing potato tubers against the background of the soil in order to isolate image elements belonging to the tubers. Blurred color channels were used as features for classifying elements. The elements were classified using a random forest and the linear discriminant analysis (LDA) procedure. The results of an experimental study are presented, indicating the effectiveness of the proposed approaches.