A 3d-2dcnn-Ca Approach for Enhanced Classification of Hickory Tree Species Using Uav Hyperspectral Data

Abstract

Hickory trees have significant economic, nutritional, and ecological values and play an essential role in human society and natural ecosystems. To maximize benefits, many base have adopted a composite planting strategy. Effective and timely monitoring of hickory forest species information is critical for accurate management and conservation. In this study, a low-altitude unmanned aerial vehicle (UAV) was utilized to acquire RGB and hyperspectral data from the canopy of two hickory bases, and a hybrid convolutional neural network structure was used to classify different tree species and homologous hickory species. The classification stability was improved by introducing a channel attention module to refine the features of the hyperspectral images. The experimental results show that in hickory species classification, compared with RGB images, hyperspectral images exhibit superior classification results, especially in the classification of highly homogeneous tree species. Compared with other classification methods, the 3D-2DCNN-CA proposed in this paper performs the best in hickory species classification, with an accuracy of 99.38% for a single hickory in the classification of different tree species and an accuracy of 93.97% for a single hickory in the classification of the same hickory species. In addition, the method achieved significant classification results at the single-tree scale. These results indicate that the way can realize fine-scale monitoring of hickory forests and provide strong support for the management of forest land and expert guidance on planting distribution.

Keywords: Hickory Forest classification, UAV hyperspectral, Convolutional neural network, Attention mechanism, Hickory genus