摘要
针对传统插值算法在图像整体缩放中边缘区域易产生锯齿或模糊的现象,提出一种新型的图像缩放算法。将相邻像素形成的矩形区域进行三角剖分,根据待构造曲线段的位置关系及插值控制点的距离大小构造辅助的控制顶点,再利用相应的Bezier曲线插值图像构造的辅助控制点,最后以产生的Bezier曲线作为边界曲线,利用Coons曲面构造方法构造出退化的三角形曲面片进行边缘细节的强化。提出的方法可以很好地保持图像边缘细节,尤其在倾斜的边缘和尖角等处保持了更多的细节特征,使放大的图像有更好的效果。
Traditional algorithms can result in the blurring or zigzag on edges in the image zooming. A new method for image zooming is presented which focuses on the triangulation of the pixels in the rectangle mesh. The accessorial control points associated with the undetermined curves can be presented by the position and distance of the interrelated pixels. The Bezier curves, which interpolate the accessorial control points, are used to construct the Coons surface via boundary. The new method can preserve and intensify the edge characters of the image. Especially, the edge characters in the gradient or sharp-angled area can be preserved better.
出处
《量子电子学报》
CAS
CSCD
北大核心
2012年第5期523-529,共7页
Chinese Journal of Quantum Electronics
基金
国家自然科学基金(60910005)
安徽省自然科学研究项目(KJ2009A123
KJ2011B086)
