| 刊名 | Journal of Landscape Research |
| 作者 | LI Gen, WANG Fan, WU Jing |
| 作者单位 | Anhui Xinhua University, Hefei 230088, Anhui, China |
| DOI | 10.16785/j.issn 1943-989x.2026.3-4.002 |
| 年份 | 2026 |
| 刊期 | 4 |
| 页码 | 5-8,15 |
| 关键词 | Multispectral remote sensing, UAV, Deep learning,Diseased tree identification, Semantic segmentation |
| 摘要 | To address the problems of low efficiency and poor precision in traditional forest disease monitoring, an automatic identification method for diseased trees was proposed by combining UAV multispectral remote sensing with deep learning. A vertical take-off fixed-wing UAV equipped with a Rededge-MX multispectral camera was used to acquire high-resolution multispectral images of forest areas. After preprocessing steps including radiometric calibration and aerial triangulation, the SCANet semantic segmentation model was introduced to achieve automated extraction of discolored diseased trees. The results showed that the proposed method achieved a precision of 0.909 3, a recall of 0.808 2, and a miss rate of only 0.191 8 for diseased tree identification, significantly outperforming traditional methods such as support vector machine (SVM) and BP neural network, as well as mainstream deep learning models including ResNet and DenseNet. This method enables efficient, high-precision, and automated monitoring of diseased trees in forest areas, providing technical support for the prevention and control of forest pests and diseases. |
