Health status detection of oil palm tree using an unmanned aerial vehicle multispectral image based on picterra platform

Name: Hong Lay

LCSH: Burapha University.Faculty of Geoinformatics

Classification :.DDC: 629.13339

LCSH: Oil palm

LCSH: Drone aircraft

Abstract: Oil palm plantations are a significant export crop for Cambodia, providing employment opportunities and economic growth. An Unmanned Aerial Vehicle (UAV) was used to capture two plots of oil palm area for this research. Oil palm trees were extracted from high-resolution images using the Picterra platform. Furthermore, oil palm trees are counted both automatically and manually, with the effect demonstrating a high overall accuracy. In addition, also used the multispectral image to assess the health of oil palm trees based on the Parrot Sequoia camera. The camera has occurred in three bands like Green, Red, Red Edge, and Near-Infrared. Thereby, the health of an oil palm tree is determined using vegetation indices such as NDRE, GNDVI, and NDVI. On the other hand, maximum, low, mean, and standard deviation in vegetation and chlorophyll content were contrasted with the vegetation indices. The NDVI indices are superior to the NDRE and GNDVI indices. There are two objectives of the research as the following; 1) to detect and count oil palm trees of very high-resolution images from UAV with Picterra platform and 2) to evaluate and compare oil palm trees health by Using NDRE, GNDVI, and NDVI indices in vegetation and chlorophyll content. Oil palm trees were detected and counted using UAV-based high-resolution imagery, and their health was assessed using multispectral images. According to the Picterra platform, the output of counting is Plot-1 has 3456 oil palm trees, and Plot-2 has 3477 oil palm trees. The accuracy of oil palm detection using the F-score of Plot-1 is 100%, and Plot-2 is 98.97%. In this research, Picterra is a high-performance platform that can use to retrieve objects from UAV imagery. The results of the health assessment of oil palm trees reveal from Normalized Difference Red Edge that Plot-1 has three classes: low chlorophyll (0.14– 0.29) of 22.92%, medium chlorophyll (0.29–0.33) of 48.64%, and high chlorophyll (0.33–0.44) of 28.44%, and Plot-2 has three classes: low chlorophyll (0.13–0.26) of 22.93%, medium chlorophyll (0.26–0.31) of 48.27%, and high chlorophyll (0.31– 0.40) of 28.80%. Plot-1 has three classes: unhealthy (0.41–0.65) of 10.22%, moderately healthy (0.65–0.71) of 43.30%, and very healthy (0.71–079) of 46.48%, while Plot-2 has two classes: moderately healthy (0.46–0.69) of 40.91% and very healthy (0.69– 0.78) of 59.09%, according to Green Normalized Difference Vegetation Index calculations. Whereas Normalized Difference Vegetation Index shows that Plot-1 has three classes: unhealthy (0.33–0.71) of 4.34%, moderately healthy (0.71–0.81) of 37.12%, and very healthy (0.81–0.88) of 58.54%, while Plot-2 has two classes: moderately healthy (0.54–0.81) of 22.56% and very healthy (0.81–0.88) of 77.44%. In this thesis, the vegetation index is extracted from multispectral images of the UAV platform, and the oil palm tree is classified. The results have been published in an international academic conference and applied in Cambodia's MRICOP Company. Therefore, the Picterra platform is helpful for object extraction and geospatial analysis since the F-score has resulted in high accuracy assessment.

Burapha University. Library

Address: CHONBURI

Email: buulibrary@buu.ac.th

Name: Zhenfeng Shao

Role: Principal advisor

Name: Kitsanai Charoenjit

Role: Co-advisor

Created: 2021

Modified: 2022-06-02

Issued: 2022-06-02

วิทยานิพนธ์/Thesis

application/pdf

CallNumber: Th 629.13339 Ho772H

eng

DegreeName: Master of Science

Level: Master's Degree

Descipline: Geoinformatics

Grantor: Burapha University

©copyrights Burapha University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	62910208.pdf	8.15 MB	2	2024-08-09 15:04:52

ใช้เวลา

0.027271 วินาที