Archives

An Analysis of Stress in Trees with a UAV Platform with Lightweight Multispectral Sensors


Dong-MoonKim
Abstract

This study was conducted based on the need for ongoing research on the application and utilization of UAV platforms with lightweight multispectral sensors applicable to various flexibility of geographical features rather than the old aircraft-based monitoring technology in order to ensure the proper management and utilization of crops and forest resources.In this study, NIR cameras and GPS/IMU were used based on a multicopter, one of the UAV platforms, to obtain multispectral data about the subject. Data about datum points was also obtained with a GNSS receiver through an RTK survey, and ortho-photos and 3D space information were obtained through aerial photogrammetry. The biological parameters of ENDVI, a revised formula of NDVI, were used to detect the vitality of trees with plants in a stressed state.As a result, it was even possible to do an analysis for overall vegetation with ENDVI and carry out supervised classification with sample field study findings and vegetation indexes to identify six items. The classification outcomes were rearranged into three items including healthy trees, trees in a stressed state, and vacant lots and structures based on representation, adjacency and connectivity between items, and standard deviation around "grass in a stressed state," a major item in analysis. There were three boundary points between items including 0-0.43, 0.43-0.5, and 0.5-1. In addition, 86,700 ㎡, which accounted for 8.17% of the entire analysis objects, was classified to contain vegetation in a stressed state.It turned out that the application of a band-pass filter in a narrow bandwidth would provide reliable information about the analysis of plant health and stress among various types of plants. These findings can be used to analyze damages by blight and harmful insects or conduct quantitative analysis for crops and forest resources.

Volume 11 | 05-Special Issue

Pages: 2464-2470