DATA ENGINEERING FOR PROSPECTIVE VEGETATION INDICES OF LEAF DIAGNOSTICS BASED ON HSL COLOR FORMATION MODEL

Abstract

The work is devoted to the implementation of traditional technologies of visual monitoring of plants in the technology of precision agriculture, namely the improvement of remote monitoring with the help of UAVs in relation to marker vegetation indices. Classic vegetation indices such as NDVI are used to solve a limited range of problems and are used primarily to adjust the amount of nitrogen fertilizers during differentiated treatment of field areas. Such indices are poorly adapted to identify the causes of stress. For stresses of a technological nature, in particular, on winter rapeseed crops, marker indices are used, which are difficult to adjust to identify abnormal coloration of affected plants. In addition, such indices are sensitive to changes in lighting and require atmospheric correction measures. The purpose of the work is the formation of a new approach to the automation of visual diagnostics of plants, which is based on the adaptation of machine vision technologies to the existing technologies of noncontact expert assessment of plants. A hypothesis was put forward about the possibility of creating vegetation indices based on an alternative model of HSL coloration, which would be more resistant to changes in illumination.