In this paper, we propose a new classification algorithm for high spatial resolution remote sensing data, which makes
use of both the spatial information and spectral information of the remote sensing data, and is also suitable for the
specific data format that high special resolution remote sensing data has. Because the classification is based on the
remote sensing raw data, image fusion process of the raw data can be avoided, which will lower the computational cost.
Moreover, the classification algorithm framework we propose can be extended for other segmentation algorithms.