An Improved Green Vegetation Segmentation Method Based on RGB and NIR Image Pairs

Le résumé fourni par la source

Accurate extraction of green vegetation from top-down perspectives is crucial for applications such as precision agriculture, ecological monitoring, and environmental management. However, traditional methods relying solely on color indices or spectral indices often fail to accurately extract green vegetation in the presence of complex interference from soil and non-green vegetation. To address this limitation, this paper proposes a simple yet efficient method for green vegetation extraction. Leveraging RGB+NIR image pairs, the proposed approach effectively combines the strengths of spectral and color indices across diverse scenarios. By utilizing the Otsu method, it achieves robust identification of green vegetation, with the added advantage of being deployable on edge devices. NDVI-ExG achieved the highest recall (0.8710) in the ground dataset, while SAVI-ExG demonstrated the best accuracy (0.9421) and precision (0.9512) in the same dataset. In the unmanned aerial vehicle (UAV) dataset, SAVI-ExG outperformed comprehensively, achieving accuracy, precision, and recall of 0.9624, 0.9812, and 0.8753, respectively.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.