Header

UZH-Logo

Maintenance Infos

Detection of mistletoe infected trees using UAV high spatial resolution images


Miraki, Mojdeh; Sohrabi, Hormoz; Fatehi, Parviz; Kneubühler, Mathias (2021). Detection of mistletoe infected trees using UAV high spatial resolution images. Journal of Plant Diseases and Protection:Epub ahead of print.

Abstract

Remote detection of aerial parasitic plants in forests is imperative in precision forestry, as it can help to manage tree stands and to monitor forest ecosystem health. The plain forests located in Noor and Hyrcanian forests (Iran), characterized by mixed broadleaved forests, host large populations of mistletoe (Viscum album). In this study, aiming to delineate trees infected by mistletoe, an unmanned aerial vehicle (UAV) equipped with an RGB camera was used to acquire series of images in winter (February of 2020) and summer (September of 2020). The canopy height model (CHM) was generated from UAV images then Gaussian filter was applied to the CHM. Two sites were selected and the individual tree crowns in these sites were delineated manually and automatically (using a region growing algorithm based on the filtered canopy height model). A range of UAV-based RGB vegetation indices (VIs) was generated. Individual trees were classified into two classes (i.e., infected and non-infected) using a random forest classification algorithm, and based on six image processing scenarios (i.e., three scenarios where tree crowns were delineated manually, followed by identification of infected trees in leaf-off, leaf-on, and combined leaf-off and leaf-on seasons, and three scenarios with the same identification procedure applied to automatically delineated tree crowns using a regional growing algorithm). Optimal classification results using manual and automatic crown delineation were obtained by leaf-off and combined leaf-off + leaf-on season data with the overall accuracy of 87% and 76% for site 1, respectively. Also, the overall accuracy of 80% and 69% was obtained from combined leaf-off + leaf-on season data for site 2. The study demonstrates the potential of using UAV-based RGB data for studying mistletoe infection and distribution in a complex forest ecosystem.

Abstract

Remote detection of aerial parasitic plants in forests is imperative in precision forestry, as it can help to manage tree stands and to monitor forest ecosystem health. The plain forests located in Noor and Hyrcanian forests (Iran), characterized by mixed broadleaved forests, host large populations of mistletoe (Viscum album). In this study, aiming to delineate trees infected by mistletoe, an unmanned aerial vehicle (UAV) equipped with an RGB camera was used to acquire series of images in winter (February of 2020) and summer (September of 2020). The canopy height model (CHM) was generated from UAV images then Gaussian filter was applied to the CHM. Two sites were selected and the individual tree crowns in these sites were delineated manually and automatically (using a region growing algorithm based on the filtered canopy height model). A range of UAV-based RGB vegetation indices (VIs) was generated. Individual trees were classified into two classes (i.e., infected and non-infected) using a random forest classification algorithm, and based on six image processing scenarios (i.e., three scenarios where tree crowns were delineated manually, followed by identification of infected trees in leaf-off, leaf-on, and combined leaf-off and leaf-on seasons, and three scenarios with the same identification procedure applied to automatically delineated tree crowns using a regional growing algorithm). Optimal classification results using manual and automatic crown delineation were obtained by leaf-off and combined leaf-off + leaf-on season data with the overall accuracy of 87% and 76% for site 1, respectively. Also, the overall accuracy of 80% and 69% was obtained from combined leaf-off + leaf-on season data for site 2. The study demonstrates the potential of using UAV-based RGB data for studying mistletoe infection and distribution in a complex forest ecosystem.

Statistics

Citations

Altmetrics

Downloads

1 download since deposited on 02 Sep 2021
1 download since 12 months
Detailed statistics

Additional indexing

Item Type:Journal Article, refereed, original work
Communities & Collections:07 Faculty of Science > Institute of Geography
Dewey Decimal Classification:910 Geography & travel
Scopus Subject Areas:Life Sciences > Agronomy and Crop Science
Life Sciences > Plant Science
Life Sciences > Horticulture
Uncontrolled Keywords:Horticulture, Plant Science, Agronomy and Crop Science
Language:English
Date:23 July 2021
Deposited On:02 Sep 2021 13:08
Last Modified:03 Sep 2021 20:00
Publisher:Springer
ISSN:1861-3837
OA Status:Closed
Publisher DOI:https://doi.org/10.1007/s41348-021-00502-6

Download

Closed Access: Download allowed only for UZH members