In this paper, a new approach of multi-class target recognition is proposed for remote sensing image analysis. A multiclass
feature model is built, which is based on sharing features among classes. In order to make the recognition process
efficient, we adopted the idea of adaptive feature selection. In each layer of the integrated feature model, the most salient
and stable feature are selected first, and then the less ones. Experiments demonstrated the approach proposed is efficient
in computation and is adaptive to scene variation.
A method of feature validity analysis is proposed in this research. More than 100 kinds of features which are commonly
used in the remote sensing image analysis have been selected. In this process, we choose level set and Otsu segmentation
methods to extract the target and use manually segmented target images as templates, based on the method, we can make
a evaluation of these features validity especially the stability in multi-resolution.