基于Seam Carving的图像放大算法研究与改进

Research and Implementation of Image Magnification Algorithm Based on Seam Carving

针对Seam Carving图像放大算法中,能量值前k小的像素带可能出现共用点的情况,提出了一种改进的图像放大自适应算法,通过判断图像中出现共用点的像素带的情况及计算其共用率的值,经过一定的策略确定待新增的像素带位置,该算法使得图像中视觉关注非重要信息区域像素带的增加能得到较均匀的分布;实现了人工干预选择视觉关注重要区域的方法,避免了自适应算法本身因误判视觉关注重要区域而带来关注对象发生变形扭曲的情况.实验证明,改进的算法能得到更好的放大效果.

An improved image magnification adaptive algorithm was proposed on the basis of seam carving algorithm which may appears the situation that the first k small pixels share the same points. It finds the newly adding position of pixels by some strategies through judging the sharing situation and its sharing rate, which makes the addition of new pixels in non-important area of visual attention more reasonable. The manual intervention to select the important area of visual attention is implemented to avoid perpetual object becoming warped and deformed by wrong judgment itself. Experiments show that the improved algorithms get better effects.