Abstract

Commercial forest species discrimination is valuable for optimal management of commercial forests. Therefore, second-order image texture combinations computed from a 0.5 m WorldView-2 pan-sharpened image integrated with sparse partial least squares discriminant analysis (SPLS-DA) and partial least squares discriminant analysis (PLS-DA) were used to discriminate commercial forest species. The findings show that the SPLS-DA model, which is characterised by concurrent variable selection and reduction of data dimensionality, produced an overall classification accuracy of 86%, with an allocation disagreement of 9 and a quantity disagreement of 5. Conversely, the PLS-DA model with variable importance in projection (VIP) produced an overall classification accuracy of 81%, with an allocation disagreement of 12 and a quantity disagreement of 7. Overall, this study demonstrates the value of second-order image texture combinations in discriminating commercial forest species and presents an opportunity for improved commercial forest species delineation.

摘要

商业林物种区分对于商业林的优化管理很有价值。因此，使用结合稀疏偏最小二乘判别分析 (SPLS-DA) 和偏最小二乘判别分析 (PLS-DA) 的 0.5 m WorldView-2 泛锐化图像计算的二阶图像纹理组合来区分商业林物种。研究结果表明，SPLS-DA 模型以并发变量选择和数据维数降低为特征，产生了 86% 的总体分类准确率，分配不一致为 9，数量不一致为 5。相反，PLS-具有可变投影重要性 (VIP) 的 DA 模型产生了 81% 的总体分类准确率，分配差异为 12，数量差异为 7。总体而言，