Remote sensing image registration using SIFT and vegetation index analysis

Aiming at the high accuracy and speed requirements of images registration for multiband data or hyperspectral data, a new method which combines scale invariant feature transform (SIFT) with vegetation index analysis is put forward. Firstly, feature points extracted by SIFT algorithm are classified into two sets - points on vegetation area and points on non-vegetation area, which is based on vegetation index; then the two sets of feature points are matched separately using spectral angle distance as the similarity measure. Transformation parameters are obtained by least square method after mismatched points are removed. Experimental results show that the proposed method achieves higher speed as well as good registration accuracy.