19 September 2016 Identification of Phragmites australis and Spartina alterniflora in the Yangtze Estuary between Bayes and BP neural network using hyper-spectral data
Author Affiliations +
Abstract
The aim of this work was to identify the coastal wetland plants between Bayes and BP neural network using hyperspectral data in order to optimize the classification method. For this purpose, we chose two dominant plants (invasive S. alterniflora and native P. australis) in the Yangtze Estuary, the leaf spectral reflectance of P. australis and S. alterniflora were measured by ASD field spectral machine. We tested the Bayes method and BP neural network for the identification of these two species. Results showed that three different bands (i.e., 555 nm,711 nm and 920 nm) could be identified as the sensitive bands for the input parameters for the two methods. Bayes method and BP neural network prediction model both performed well (Bayes prediction for 88.57% accuracy, BP neural network model prediction for about 80% accuracy), but Bayes theorem method could give higher accuracy and stability.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Pudong Liu, Jiayuan Zhou, Runhe Shi, Chao Zhang, Chaoshun Liu, Zhibin Sun, Wei Gao, "Identification of Phragmites australis and Spartina alterniflora in the Yangtze Estuary between Bayes and BP neural network using hyper-spectral data", Proc. SPIE 9975, Remote Sensing and Modeling of Ecosystems for Sustainability XIII, 99750H (19 September 2016); doi: 10.1117/12.2236772; https://doi.org/10.1117/12.2236772
PROCEEDINGS
8 PAGES


SHARE
Back to Top