Veranstaltungsprogramm

Eine Übersicht aller Sessions/Sitzungen dieser Tagung.
Bitte wählen Sie einen Ort oder ein Datum aus, um nur die betreffenden Sitzungen anzuzeigen. Wählen Sie eine Sitzung aus, um zur Detailanzeige zu gelangen.

 
Sitzungsübersicht
Sitzung
B18: Joint GI_Forum and EARSeL UAS Summit I
Zeit:
Donnerstag, 04.07.2019:
11:30 - 13:00

Chair der Sitzung: Alexander Almer
Ort: B - Blauer Hörsaal
HS402

Zusammenfassung der Sitzung

additional presentations:

On the Potentiality of UAV Multispectral Imagery to Detect Oak Wilt Disease Using Convolutional Neural Networks

Hwa-Seon Lee1, Won-Woo Seo2, Kyu-Sung Lee3

Organisations: 1: Inha University, Korea, Republic of (South Korea); 2: Inha University, Korea, Republic of (South Korea); 3: Inha University, Korea, Republic of (South Korea)

Presenting author: Hwa-Seon Lee

Topics: Machine learning, deep learning, EO applications, societal benefits, SDGs

Keywords: UAV, multi-spectral, oak wilt disease, CNN


Mapping of Rumex obtusifolius in Native Grassland using Unmanned Aerial Vehicle: From Object-based Image Analysis to Deep Learning

Olee Hoi Ying Lam1, Bethany Melville1, Marcel Dogotari1, Moritz Prüm1, Hemang Narendra Vithlani1, Corinna Roers2, Rolf Becker1, Frank Zimmer1

Organisations: 1: Hochschule Rhein-Waal, Faculty of Communication and Environment, Germany; 2: Naturschutzzentrum im Kreis Kleve e.V., Germany

Presnting author: Olee Hoi Ying Lam

Topics: Machine learning, deep learning, Computer vision, knowledge organising systems

Keywords: weed mapping, image segmentation, neural network, UAV, RGB-imagery


Zeige Hilfe zu 'Vergrößern oder verkleinern Sie den Text der Zusammenfassung' an
Präsentationen

UAV-based Tree Height Estimation in Dense Tropical Rainforests Areas in Ecuador and Brazil

Stefan Reder, Lilli Waßermann, Jan-Peter Mund

HNE Eberswalde, Deutschland

The aim of this study was to develop an easy applicable, cost-efficient workflow for tree height estimation in remote, inaccessible rainforest areas in Ecuador and Brazil. Structure from Motion was combined with a digital terrain model (DTM) from the Shuttle Radar Topography Mission (SRTM) to complement relief information to the photogrammetric point clouds (PPC) that represents the upper canopy layers. Based on by ground points extracted form a generated 3D model, a vertical shift of the model was applied to adjust the ellipsoid level of the PPT. Digital surface models (DSM) of 22 research plots (75m x 75m) were normalized to canopy height (CHM) to allow the estimation of relative tree heights in all research plots without using ground control points (GCP). The calculated tree height values indicate the applicability of the proposed workflow even in dense tropical forest canopies. This approach allows the classification of canopy structures for identifying forest succession and for other ecological forest monitoring purposes. Our results emphasize the potential of 3D models for tree height estimation derived from PPTs based on UAVs imagery in rainforests research.



Aerial and Terrestrial Photogrammetric Point Cloud Fusion for Intensive Forest Monitoring

Stuart Krause1,2,3

1Thünen Institute of Forest Ecosystems; 2Faculty of Forest and Environment, Eberswalde University of Sustainable Development; 3Department of Geography, University of Bonn

Remote sensing methods for forest monitoring are rapidly evolving due to recent advancements in UAVs and photogrammetry. Photogrammetric point clouds provide the possibility to achieve the non-destructive derivation of individual tree parameters at a low cost. The fusion of aerial and terrestrial photogrammetry for creating full tree point clouds is of utility for forest research, as tree volume can be assessed more economically and efficiently than traditional methods. This however is a challenge to implement due to difficulties with co-registration and issues of occlusion. This study explores the possibility to use spherical targets typically used for Terrestrial Laser Scanning in order to accomplish the co-registration of UAV-based and terrestrial photogrammetric datasets. Results show a full tree point cloud derived from UAV oblique imagery with terrestrial imagery. Despite issues of noise produced from the sky in terrestrial imagery, the study shows a promising methodology for aerial and terrestrial point cloud fusion.



 
Impressum · Kontaktadresse:
Datenschutzerklärung · Veranstaltung: GI_Week 2019
Conference Software - ConfTool Pro 2.6.128+TC
© 2001 - 2019 by Dr. H. Weinreich, Hamburg, Germany