Improved seam carving for content-aware image retargeting

Abstract

When resizing image, effective means should not only consider geometric constraints, but also handle according to the image content. It is the fact that the image content in the middle is not more attractive than at the edge necessarily, the image resizing means on the basis of content is turning into a new hotspot in the image processing domain. Image resizing is also known as image retargeting. Firstly, this means treats the areas that attract human eyes' attention as more important regions, but the areas that not catch the people's eyes as unimportant districts, then to the greatest extent maintaining the important regions but changing unimportant districts in order to fit target image size. Therefore, this technology is referred to as content-aware image resizing. Among all the means relating to content-aware image resizing, seam carving algorithm's status is beyond all doubt, its advancement and innovation is better than other ways. In this article, seam-carving means was used as a starting point, after analysing the typical seam-carving algorithm, this paper presents a improved algorithm based on seam-carving. In our experiment, it is obvious to see that our new improved way enhances the resizing effects, avoids the distortion and makes better results.