Text this: Classification of hyperspectral data for land cover mapping : is there any significant improvement?