Seam Carving image scaling method with visual significant graph

Zheng-hong GUO,Jun-hua ZHANG,Xiao-peng GUO,Li-ye MEI
DOI: https://doi.org/10.7540/j.ynu.20170223
2018-01-01
Abstract:Seam Carving algorithm may result in the distortion of the visual subject area and the missing of the important content of the image when scaling an image with no obvious contrast between the subject area and the background color or with larger subject area.In order to improve the lack of Seam Carving algorithm,we use a combination of image gradient graph and saliency map to improve the gradient energy graph of the image.The results of experiments show that,compare with the Seam Carving algorithm,the proposed method can better maintain the important content of the image,preserves a better overall visual effect,and obtains a larger average energy value of the image pixel and a better image scaling quality in scaling the image.
What problem does this paper attempt to address?