Abstract: Scaling and seam carving are commonly used algorithm in image resizing, but scaling is not
suitable for the case that the resize ratio is not uniformity, seam carving will also has a bad effect on
important area in some cases. In this paper we proposed a novel algorithm which is based on important
region maintain. Firstly, we make a corner detection based on difference of Gaussian, then Graham-scan
algorithm is applied to do an important region detection, after that we will add different weights
between important region and unimportant region. The experimental result shows our method is better
than other image resizing algorithms in maintaining the important area.

Key words: seam carving, important area, corner detection, image resizing, convex hull