改进的图像缝雕刻算法  

Improved Image Seam Carving Method

作　　者：聂栋栋[1] 马利庄[2] 

机构地区：[1]燕山大学理学院,秦皇岛066004 [2]上海交通大学计算机科学与工程系,上海200240

出　　处：《计算机辅助设计与图形学学报》2011年第11期1890-1895,共6页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60873136;61073089);河北省自然科学基金(A2011203053);秦皇岛市科学技术研究与发展计划项目(201001A038)

摘　　要：针对现有算法中对低频前景区域的处理效果欠佳的问题,提出一种改进的图像缝雕刻算法.它利用像素梯度信息对像素能量进行定义,然后提出最大能量最小函数来确定最佳的处理线路,再通过删除或添加单像素宽的线路,得到适合大小的目标图像.实验结果表明,该算法不仅能够保护图像中前景对象的显示效果,使之免受图像尺寸整体变化的影响,而且能够减少由于线路定位误差带来的前景对象扭曲.An improved image seam carving method has been proposed to solve the poor processing problem for the low-frequency foreground regions in existing methods. It defines the pixel energy with image gradient information, proposes a maximum energy minimization function to determine the best seams to be handled, and resizes images to the target size by adding or deleting one-pixel-width seams. Experiment results show that the proposed method can not only protect the foreground objects in the images from the influence caused by the overall image size change, but also reduces their distortion caused by positioning error of seams.

关 键 词：图像适应 图像缩放 图像形变 缝雕刻算法 动态规划 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]