Comparison of unmanned aerial vehicle (UAV) RGB imagery and point clouds in crown area estimation of individual trees within pine (Pinus eldarica) man-made forests

Subject Areas : Agriculture, rangeland, watershed and forestry

Ali Hosingholizade ¹ , Yousef Erfanifard ² , Seyed Kazem Alavipanah ³ , Homan Latifi ⁴ , Yaser Jouybari Moghaddam ⁵

1 - Department of Remote Sensing and GIS, Faculty of Geography, University of Tehran, Tehran, Iran
2 - Dept. of Remote sensing and GIS, Faculty of Geography, University of Tehran, Iran
3 - Department of Remote Sensing and GIS, Faculty of Geography, University of Tehran, Tehran, Iran
4 - Assistant Professor, Faculty of Surveying Engineering, Khajeh Nasir Tusi University of Technology, Tehran, Iran.
5 - Department of Surveying Engineering, Faculty of Engineering, Bojnurd University, Bojnurd, Iran

Received: 2022-02-14 Accepted : 2022-03-06 Published : 2023-05-22

Keywords: Marker-Controlled Watershed algorithm, Crown height model, Bojnurd Pardisan Park, Segmentation,

Abstract :

Application of unmanned aerial vehicles (UAVs) is widespread in measurement of quantitative characteristics of single trees such as crown area. However, the efficiency of different types of data collected by UAVs are less compared. Therefore, the aim of this study was comparison of UAV RGB imagery and point clouds in crown area estimation of individual pine trees within a man-made forest in Pardisan Park, Northern Khorasan province, Iran. Both datatypes were analyzed by similar segmentation method (Multiresolution segmentation on the RGB images and Marker-Controlled Watershed algorithm in the point clouds) to estimate crown area of 324 sample pine trees. The results showed that the crown area measured on the point clouds had higher accuracy and precision compared to RGB imager (Spearman correlation of 0.95 and 0.81, coefficient of determination of 0.97 and 0.59, p-value of 0.97 and 0.001, respectively). Additionally, crown area of pine trees with large crown (> 18 m2) delineated on the point clouds showed the highest accuracy in comparison to trees with small and medium crowns. In general, it was concluded that segmentation of UAV point clouds was more efficient than RGB imagery segmentation in quantification of crown area of pine trees within the study area.