Abstract: Seam carving algorithm is an image resizing algorithm based on image content.Resizing an image with this method may deform the image on salient parts.To improve the seam carving algorithm,this paper uses both the salient map produced by the graph-based visual saliency model and the image gradient map to compute the importance of pixels on human visual sense.Experiment results show that the proposed approach gets a better result compared with seam carving algorithm.