研究目的
Investigating the application of hyperspectral image analysis in distinguishing among different plant species by combining spectral information and spatial location of pixels for classification.
研究成果
The proposed SI-GCK method yields accurate classification maps and SI can be used as spectral features of plants for classification. The results are very encouraging in plant classification, but further validations are needed.
研究不足
Further validations such as other spatial features and composite kernel learning should be researched in future developments.
1：Experimental Design and Method Selection:
The proposed SI-GCK method uses Spectral Index (SI) to represent plant spectral and semi-supervised graph-based composite kernel (GCK) method for classification.
2：Sample Selection and Data Sources:
Two real hyperspectral images, Indian Pines data set and Salinas Valley data set collected by AVIRIS, are used.
3：List of Experimental Equipment and Materials:
Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) for data collection.
4：Experimental Procedures and Operational Workflow:
Extraction of hyperspectral image spectral feature by computing spectral indices; integration of spectral features and spatial location information using graph-based composite kernel approach; classification by means of Nystr?m Method Formulation.
5：Data Analysis Methods:
Three metrics, overall accuracy (OA), average accuracy (AA) and kappa coefficient of agreement (KA), are used to assess the classification accuracies.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
