Back to Search
Start Over
Identification of Phragmites australis and Spartina alterniflora in the Yangtze Estuary between Bayes and BP neural network using hyper-spectral data
- Source :
- SPIE Proceedings.
- Publication Year :
- 2016
- Publisher :
- SPIE, 2016.
-
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.
- Subjects :
- 0106 biological sciences
geography
geography.geographical_feature_category
biology
Artificial neural network
business.industry
Hyperspectral imaging
Estuary
Pattern recognition
04 agricultural and veterinary sciences
Spartina alterniflora
biology.organism_classification
010603 evolutionary biology
01 natural sciences
Stability (probability)
Phragmites
Bayes' theorem
040103 agronomy & agriculture
0401 agriculture, forestry, and fisheries
Artificial intelligence
business
Spectral data
Mathematics
Subjects
Details
- ISSN :
- 0277786X
- Database :
- OpenAIRE
- Journal :
- SPIE Proceedings
- Accession number :
- edsair.doi...........2e61c44518ec3d805da7a56a79cad745