摘要
从理论上分析了细缝裁减方法的思路,找到细缝裁减过后出现图像降质的原因,并提出了改进方案。主要思路是在细缝被裁减掉之后,将去除的细缝像素点的灰度能量加权累加到邻域像素点。越接近细缝的邻域像素点增加的灰度能量越大,远离细缝的像素点增加的灰度能量越小。通过与原始细缝裁减方法进行实验对照发现,新算法可以有效避免细缝的局部聚集问题,使细缝尽可能均匀分布,从而有效降低图像重要区域的变形带来的主观视觉降质。
We analyze why the seam carving brings about image degradation and have an idea of avoiding neighboring seams removed.In our improved approach,the grave level energy of removed seam is transferred to the pixels of its neighborhood.It decreases the probability of neighboring pixels being chosen as seams.It makes seams distribute as evenly as possible.The experimental results show that our approach bring less subjective image degradation.
基金
国家自然科学基金资助项目(61040052)
北京市优秀人才资助项目(2009D005015000010)
关键词
细缝裁减
能量函数
灰度能量转移
seam carving
energy function
gray energy transfer
