基于无人机影像阴影去除的苹果树冠层氮素含量遥感反演

【目的】去除无人机多光谱遥感影像中的阴影,以提高苹果树冠层氮素含量反演模型精度。【方法】以山东省栖霞市苹果园为试验区,利用2019年6月采集的无人机多光谱影像,分别基于归一化阴影指数（normalized shaded vegetation index,NSVI）和归一化冠层阴影指数（normalized difference canopy shadow index,NDCSI）去除果树冠层多光谱影像中的阴影,提取非阴影区域果树冠层光谱信息;通过相关性分析方法,将基于原始光谱影像和基于NSVI、NDCSI去除阴影后提取的光谱数据与实测叶片氮素含量进行相关性分析,分别筛选氮素含量的敏感波段并构建光谱参量;采用偏最小二乘（partial least square,PLS）及支持向量机（support vector machine,SVM）方法构建果树冠层氮素含量反演模型并进行精度检验。【结果】绿光波段和红光波段为果树冠层氮素含量反演的敏感波段;阴影削弱了果树冠层的光谱信息,去除阴影前后,冠层多光谱各波段光谱差异较大,在红边波段及近红外波段尤为明显;基于2个阴影指数去除阴影后构建的氮素反演模型精度均有提升,最优模型为基于NDCSI去除阴影后构建的支持向量机氮素含量反演模型,该模型建模集R²和RPD分别为0.774、1.828;验证集R²和RPD分别为0.723、1.819。【结论】基于NDCSI可有效去除无人机多光谱果树冠层影像中的阴影,提高氮素含量反演精度,为果园氮素精准管理提供了有效参考。

Remote Sensing Inversion of Nitrogen Content in Apple Canopy Based on Shadow Removal in UAV Multi-Spectral Remote Sensing Images

【Objective】The shadows in UAV multi-spectral remote sensing images were removed to improve the accuracy of the nitrogen inversion model for apple canopy. 【Method】Using the UAV multi-spectral images collected in June 2019 at the apple orchard of Qixia city in Shandong province, as the experimental area, normalized shaded vegetation index (NSVI) and normalized canopy shadow index (NDCSI) were respectively used to remove shadow and to extract the spectral information of the canopy in non shadow area. The correlation analysis method was used to analyze the correlation between the spectral data, including the data obtained based on the original spectral images and the images after removing the shadow based on NSVI and NDCSI, and the measured leaf nitrogen content data, respectively, and then the sensitive wavelength of nitrogen content were screened and spectral parameters were constructed. Partial least squares (PLS) and support vector machine (SVM) methods were used to build the inversion model of nitrogen content and to carry out the precision inspection in the fruit tree canopy. 【Result】The results showed that the green band and red band were sensitive bands for the inversion of nitrogen content in fruit tree canopy based on UAV multi-spectral images. The spectral information of fruit tree canopy was weakened by shadow, and the spectral difference of canopy multispectral bands before and after shadow removal was significant, especially in red-edge band and near-infrared band. The accuracy of nitrogen inversion model based on two shadow indexes after shadow removal was improved, and the optimal model was the support vector machine nitrogen content inversion model based on NDCSI, the modeling set of this model R² and RPD was 0.774 and 1.828, the validation set R² and RPD were 0.723 and 1.819 respectively. 【Conclusion】NDCSI could effectively remove the shadow in the multi-spectral fruit tree canopy image of the UAV to improve remote sensing inversion accuracy of nitrogen content in apple canopy, so as to provide a useful reference for precise nitrogen management in orchard.