基于全局图像结构信息的Seam Carving算法

Seam Carving algorithm based on global image structure

提出了一种基于全局图像结构信息的Seam Carving算法,它根据像素的重要性修改图像尺寸和比例。通过从图像提取特定方向的边缘结构信息,再利用每个像素的梯度信息,从全局和局部两方面定义新的像素能量计算函数,以此来阻止seam通路与特定方向图像边缘的交叉,避免边缘像素的不一致位移,以此保持图像的边缘结构。实验结果证明,算法减少了处理后图像的结构形变,有效改进了Seam Carving算法的处理效果。

This paper presented a Seam Carving algorithm based on global image structure,which changed the sizes and scales of images by the importance of each pixel.It defined a new pixel energy function from both global and local information by extracting special image edges and computing each pixel gradients,which prevented seams from crossing image edges,avoided the pixels on edges moving differently and preserved image edge structures.Experiments confirm that this algorithm reduces the structure distortion and improves the visual effect of images by Seam Carving process effectively.