In order to improve the remote sensing data classification accuracy of the environment and disaster monitoring and forecasting small satellite constellation 1A (HJ-1A) Star, the authors first fused hyperspectral imager data and CCD multispectral imagery by the Gram-Schmidt fusion algorithm, and then applied dimensionality reduction to the fused hyperspectral image by using principal component analysis (PCA) and kernel principal component analysis (KPCA). Gaussian, linear and polynomial kernel functions were employed during KPCA dimensionality reduction, and the polynomial kernel function was selected with its highest accumulative contribution rate according to the evaluation results of feature extraction. Finally, the fused hyperspectral image, the PCA image and the KPCA image with the polynomial kernel function were classified using the fuzzy C-means algorithm (FCM), respectively. The experimental results show that, for the fused hyperspectral image, the feature extraction based on KPCA can increase computational efficiency and improve the classification accuracy.