There is a rapid increase of remote sensing images in the repositories. While for certain purposes, we just need the region
of interest (ROI). In this paper, SIFT is applied for feature extraction, descriptor generation and point matching with
strict criterions to locate ROI. With the robust matching pairs of points, a simple approach for computing scale, rotation
and translation separately is proposed. To solve the parameters, we introduce an improved RANSAC algorithm with two
cost functions to achieve rapid computation efficiency. The method for ROI localization and designation performs great
in a series of experiments and satisfying results can be seen.