Integrating UAV Photogrammetry and Object-Based Image Analysis for Fine-Scale Vegetation Classification

Abstract

Classification of vegetation cover is a fundamental aspect of environmental monitoring by remote sensing, offering crucial information on ecosystem structure, land use, and resource management. This research merges unmanned aerial vehicle (UAV) photogrammetry and object-based image analysis (OBIA) to develop a semiautomatic framework for vegetation mapping at hyper-spatial resolution. Using a DJI Mavic Pro UAV, RGB high-resolution images were captured; these were then processed for orthomosaics and Digital Surface Models (DSM). Vegetation was first extracted using the GCC index while vertical structure was calculated as a normalized DSM. Image segmentation, feature engineering, and rule-based classification created a hierarchical classification based on thresholding height. Results from the study demonstrate that the study area is covered with about 45% vegetation, most grassland, and low shrubs below 1 meter, and mid-canopy and high canopy plants are fewer in number. The results prove that OBIA outperforms traditional pixel-based methods of image analysis for very high-resolution images obtained from UAV systems. It also enhances its precision as well as its ecologic significance. There are definite challenges in terms of RGB-only sensor issues as well as processing requirements, although the proposed UAV-OBIA methodology is an excellent viable alternative for in-depth monitoring of plant life in data-poor regions.