Abstract: In recent years, the ranking based band selection has been very successful in remote sensing image classification. Hyperspectral imagery often contains hundreds of images, hence, dimensionality reduction should be applied to overcome the difficulty of Hughes phenomenon. In this study, two approaches for efficiency of band selection and robustness are applied. Classification is done using ranking based Fast Density Peek Clustering (FDPC) algorithm. For FDPC algorithm, first the ranking score of each band is computed by weighing the normalized local density and the intracluster distance rather than equally taking them into account. Secondly, an exponential-based learning rule is employed to adjust the cutoff threshold for a different number of selected bands where it is fixed in the FDPC. Finally, the selected features are processed by MPCA (Multilinear Principal Component Analysis) technique to reduce the data redundancy and increasing robustness. From the experimental analysis it is observed that the proposed ranking based classification is more than efficient and robust when compared with existing band selection techniques.