基于缝雕刻的图像重定向算法  

Image retargeting algorithm based on seam carving

作　　者：聂栋栋[1] 马勤勇[2] 

机构地区：[1]燕山大学理学院,河北秦皇岛066004 [2]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《燕山大学学报》2012年第4期353-357,共5页Journal of Yanshan University

基　　金：河北省自然科学基金资助项目(A2011203053);秦皇岛市科学技术研究与发展计划项目(201001A038)

摘　　要：图像重定向算法利用人对图像内容的感知差异进行非均匀的图像缩放,以适应新的显示需求。本文提出了一种基于缝雕刻的图像重定向算法。它利用每条seam上像素的累加能量和最大能量构造混合能量模型,使算法既能够利用人眼对梯度信息的敏感性,又可以减少复杂纹理背景对seam提取的干扰。实验结果显示,相比其他几种算法,本文算法可以更好地保护图像中视觉关注对象的边缘轮廓结构,处理后的图像不仅视觉效果更好,与原图的相似性距离也更小。Image retargeting algorithm is a non-uniform image resizing algorithm utilizing the differences of human perception to image content in order to satisfy the new demand of image displaying. A new image retargeting algorithm is proposed in this paper based on seam carving, in which a hybrid energy function is constructed by using the accumulative energy and the maximum energy in a seam, so as that proposed algorithm can both take advantage of the human sensitivity on image gradients and reduce the inter- ference of complex texture background in seam extraction. The experiments show that compared with some other seam carving al- gorithm, proposed algorithm can better protect the edges and contours of the objects with visual attention in the images, the image results of proposed algorithm not only have better visual effect, but also have less similar distance to original images.

关 键 词：图像重定向 缝雕刻算法 混合能量函数 图像缩放 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]