Abstract
Summary
Unmanned aerial vehicles (UAV) equipped with cameras have become a powerful technology to collect high resolution remote sensing data from agricultural crops. When equipped with multispectral cameras, light invisible for the human eye may be captured and used to characterize the physiological status of the vegetation. UAV imagery may be divided into three steps (1) spectral characterization of the targets of interest, (2) flight and image acquisition and (3) image processing and interpretation. The overall aims of this study were to improve knowledge in all three steps associated with UAV-based remote sensing for practical use in agriculture and to contribute to the incipient research on UAV based remote sensing for agricultural applications. Three case studies were performed to (1) Characterize the spectral signatures of sugar beet (Beta vulgaris L.) and creeping thistle (Cirsium arvensis L.) and to investigate the possibilities of spectral based discrimination of the species. (2) To perform image based classification of crop-weed in sugar beet crops and health status in orange tree orchards and (3) to describe crop cover heterogeneity from images acquired with UAV. This thesis provides evidence, through real applications, of the potential for UAV based remote sensing for improved weed and disease detection in agriculture. It showed that it was possible to discriminate between sugar beet and thistle based on their field spectral signature captured with a field portable spectroradiometer and raised the feasibility of discriminating the two types of plants with a reasonably high success rate using multispectral images centered on four to six narrow bands. Moreover, UAV remote sensing arises as a potential tool for HLB virus monitoring in citrus trees, although further work is needed to improve the feature selection for the classification models.
Unmanned aerial vehicles (UAV) equipped with cameras have become a powerful technology to collect high resolution remote sensing data from agricultural crops. When equipped with multispectral cameras, light invisible for the human eye may be captured and used to characterize the physiological status of the vegetation. UAV imagery may be divided into three steps (1) spectral characterization of the targets of interest, (2) flight and image acquisition and (3) image processing and interpretation. The overall aims of this study were to improve knowledge in all three steps associated with UAV-based remote sensing for practical use in agriculture and to contribute to the incipient research on UAV based remote sensing for agricultural applications. Three case studies were performed to (1) Characterize the spectral signatures of sugar beet (Beta vulgaris L.) and creeping thistle (Cirsium arvensis L.) and to investigate the possibilities of spectral based discrimination of the species. (2) To perform image based classification of crop-weed in sugar beet crops and health status in orange tree orchards and (3) to describe crop cover heterogeneity from images acquired with UAV. This thesis provides evidence, through real applications, of the potential for UAV based remote sensing for improved weed and disease detection in agriculture. It showed that it was possible to discriminate between sugar beet and thistle based on their field spectral signature captured with a field portable spectroradiometer and raised the feasibility of discriminating the two types of plants with a reasonably high success rate using multispectral images centered on four to six narrow bands. Moreover, UAV remote sensing arises as a potential tool for HLB virus monitoring in citrus trees, although further work is needed to improve the feature selection for the classification models.
| Original language | English |
|---|
| Publisher | Department of Plant and Environmental Sciences, Faculty of Science, University of Copenhagen |
|---|---|
| Publication status | Published - 2014 |