Home // ICNS 2021, The Seventeenth International Conference on Networking and Services // View article
Authors:
Jose F. Marin
David Mostaza-Colado
Lorena Parra
Salima Yousfi
Pedro V. Mauri
Jaime Lloret
Keywords: image processing; drone; turfgrass; green areas; vegetation indexes; wild plants.
Abstract:
Weed detection is a crucial aspect of reducing the usage of phytosanitary products. Most studies about weed detection have been performed with linear crops; few studies can be found in crops with high soil coverage. In this paper, we have evaluated the effect of drone flying height on wild species detection. We have gathered images in a golf course from 4 to 16 m above ground. A non-professional drone with a camera with 1.5 megapixels was used to gather the pictures. The images are composed of red, green, and blue bands. Images were gathered in three zones with a very high infestation, high infestation, and low infestation of Daucus carota. To evaluate the effect of flying height, we calculate the percentage of the affected area and compare the obtained rates for different height, assuming that the rate at 4 m is 100% of detection. To determine is a pixel represent the wild plant or the grass, a vegetation index is used. Our results indicate that the error in estimating the affected area is relatively low, from 8 to 10 m; in some cases, overestimation errors are detected. Nonetheless, the relative error beyond 12 m reaches up to 25% of relative error. In these cases, the error consists of an underestimation of the presence of a wild plant.
Pages: 1 to 6
Copyright: Copyright (c) IARIA, 2021
Publication date: May 30, 2021
Published in: conference
ISSN: 2308-4006
ISBN: 978-1-61208-853-2
Location: Valencia, Spain
Dates: from May 30, 2021 to June 3, 2021