Hyperspectral imagery of Pinus strobiformis infected with fungal pathogen

Hyperspectral images were taken from March till October, 2018, of southwestern white pine (Pinus strobiformis), SWWP, seedlings of ten different seed-source families. Half of the seedlings were inoculated with white pine blister rust (Cronartium ribicola). Visual assessments of vigor coincided with hyperspectral data acquisition. The aim of the experiment was to use hyperspectral data to automaticaly and objectively identify infection and degree of infection in SWWP seedlings. Moreover, we developed and evaluated a feature importance algorithm to identify the most usefull hyperspectral features for classification tasks.本数据集的高光谱图像（hyperspectral images）采集于2018年3月至10月，采集对象为10个不同种源家系的西南白松（Pinus strobiformis，简称SWWP）幼苗。其中半数幼苗被接种了松疱锈病（Cronartium ribicola）。幼苗活力的目视评估与高光谱数据采集同步进行。本实验旨在利用高光谱数据，自动化且客观地识别西南白松幼苗的感染状况及感染程度。此外，本研究开发并评估了一种特征重要性算法，以筛选出分类任务中最具应用价值的高光谱特征。