An efficient fusion approach based on two dimensional subspace feature for SAR images recognition

In this paper, we present an efficient fusion approach to identify synthetic aperture radar(SAR) images. Firstly, we consider the problem of two dimension linear discriminant analysis (2DLDA) and propose a new two dimensional method named as two dimensional subclass discriminant analysis (2DSDA), which can deal with multimodally distributed data. Secondly, we present a two dimensional features fusion method, which can reduce the feature dimension and the correlation. This approach is tested on MSTAR Public Release SAR data using standard and extended operation conditions. And experiments show that this fusion approach is feasible and effective.