Identification of plants condition using digital images

Abstract

The paper deals with the problem of evaluation of the current state of plants by digital images using color and morphological descriptors. The above mentioned problem is widely used in automated greenhouse control systems. The paper presents a procedure for identification of the state of plants. The questions of segmentation and morphological image processing are considered. Self-organizing Kohonen networks and feed-forward networks are used as classifiers of RGB images and geometric descriptors. The use of the approach described in the paper allows remote assessing the condition of plants in greenhouses, increasing productivity and reducing the percentage of crop losses.