Knowledge on the composition and distribution of tree species is essential for forest inventories and floral diversity analyses. Remote sensing imagery with high spatial resolution combined with object-based image analysis (OBIA) might provide opportunities for efficient tree species recognition to support inventories of larger areas. Cameras carried by Unmanned Aerial Vehicles (UAVs), which, due to their relatively low flight height produce imagery with very high spatial resolution, can in particular be useful for this. For this thesis, you will work on the automatic detection of several tree species in UAV data.
Tiputini Biodiversity Station (TBS) was established by the Universidad San Francisco de Quito (USFQ) in the Ecuadorian Amazon in 1995 and ever since, many researchers have studied the vast richness in biodiversity in its surroundings. During an expedition in February 2019, with the aim of getting an insight in the great variety of tree species in the TBS area, high resolution imagery has been collected over two different plots with both an RGB and an NDVI camera using a UAV. These plots cover two different ecosystems: an area that periodically floods and an area that does not (i.e. terra firme), both covering around 60 ha. For this thesis, the collected UAV datasets of these plots will be used to explore remote sensing, OBIA, and machine learning techniques for the automatic identification of tree species. Several trees in the area have already been geo-tagged and more trees will be geo-tagged during second expedition to TBS in August 2019. This will be done by Gonzalo Francisco Rivas-Torres, researcher based at USFQ. This thesis will be carried out with joint supervision by both WUR and Dr. Rivas-Torres. During the course of the thesis, it is therefore required to have regular meetings over Skype with both supervising universities.
- Automatic detection of several tree species in high resolution UAV images
- Development and assessment of OBIA and machine learning methods
- Testing of different segmentation and classification techniques
- Cao, J.; Leng, W.; Liu, K.; Liu, L.; He, Z.; Zhu, Y. Object-Based Mangrove Species Classification Using Unmanned Aerial Vehicle Hyperspectral Images and Digital Surface Models. Remote Sensing. 2018, 10, 89.
- De Castro, A.I.; Torres-Sánchez, J.; Peña, J.M.; Jiménez-Brenes, F.M.; Csillik, O.; López-Granados, F. An Automatic Random Forest-OBIA Algorithm for Early Weed Mapping between and within Crop Rows Using UAV Imagery. Remote Sensing. 2018, 10, 285.
- Interested in working with UAV data and OBIA
- Willing to perform thesis work with joint supervision of WUR and USFQ
- Knowledge of Amazonian tree species would be great but is not required
- There is a possibility to join the expedition to TBS in August 2019. However, at your own expense. Contact me for more information.
Theme(s): Sensing & measuring