摘要
线裁剪算法已被广泛应用于图像缩放中,但原始图像经线裁剪缩放后易出现失真、过度删除等畸变。为此,提出一种改进的图像适配算法。利用低通滤波器合并相邻裁剪线,应用门限技术标注不可删除点,避免下次迭代中新的裁剪线穿过高能量点,使用停止准则使得线裁剪操作进行到一定程度时,阻止其进一步迭代而产生穿过主体目标的裁剪线。实验结果表明,与前向、后向线裁剪算法相比,改进算法在不增加时间和空间复杂度的情况下,防止了线裁剪算法失真、过度删除的问题,产生较少的畸变,得到了更好的缩放效果。
All kinds of the seam carving algorithms are widely applied to image scaling. But the original image can produce distortion or excessive deleting problems after the seam carving algorithm. So an improved image adaptive algorithm is proposed. Adjacent seams can be merged by using low-pass filter. Undeletable points can be marked by applying the threshold technology to avoid that new seam pass through high energy points in the next iteration. The application of stopping criterion is able to prevent the seam pass through the main target because of the further iterations when seam carving operations are to a certain extent. Experimental results show that compared with forward and backward seam carving algorithms,this improved algorithm avoids the problems about the distortion of seam carving algorithm and excessive deletion without the increase of complexity in time and space. Less distortion is produced and scale effects are better.
出处
《计算机工程》
CAS
CSCD
北大核心
2015年第6期201-205,210,共6页
Computer Engineering
基金
国家自然科学基金资助面上项目(61372046)
陕西省自然科学基础研究计划基金资助项目(2014JM8346)
关键词
图像适配
线裁剪
能量图
门限
完整性数据
image adaptation
seam carving
energy diagram
threshold
integrity data
