Non-wood forest information extraction based on ALOS data

Non-wood forest is a kind of important forest resource. This paper focused on the information extraction of non-wood forest based on Advanced Land Observation Satellite (ALOS) data. Band characteristics were analyzed to get understanding of this data wholly by information content, correlation coefficient and Optimum Index Factor (OIF). A new set of data with eight bands were obtained by the fusion of Normalized Difference Vegetation Index (NDVI), the first three components of Principal Component Analysis (PCA1, PCA2, PCA3) and the four bands of ALOS data. Various kinds of vegetations, especially non-wood forest was analyzed through the Spectral Feature Model (SFM) and Maximum Likelihood (ML) with association of topographical map and field investigation data. Results show that NDVI and PCA can improve the extraction accuracy of non-wood forest. In addition, SFM reduces the phenomenon of mixed classification and improves the information extraction accuracy of non-wood forest, which will provide reference for the classification of vegetation.