一种柔化锯齿的超分辨率图像重建方法

基于锯齿会导致图像边缘长度增加的认识,通过减小边缘长度来抑制甚至消除图像锯齿,从而达到边缘锯齿被柔化的效果.受图像分割算法几何切的启发,首先提出一种新的图像边缘长度先验模型,该先验模型扩展了几何切的概念,使用全邻域系统,得到一种更具有实际意义的边缘长度的定义.然后把这个先验项连同图像似然项作为超分辨率图像重建的目标函数,并且通过最速下降法来极小化这个目标函数,从而达到减小边缘长度的目的.从实验结果可以看出该边缘长度先验模型具有一定的收敛性,并且边缘锯齿得到显著地柔化,图像变得更加地清晰,产生了良好的视觉效果.

A Method of Softening the Jagged in Image Super-resolution Reconstruction

Based on the knowledge that the jagged will increase the length of edge,this paper suppressed or even eliminated the jagged by reducing the length of the image′s edge,so as to achieve the effect of softened jagged edges.Inspired by an image segmentation algorithm,the geometric cuts,this paper proposed a new priori model of the edge length firstly,which extended the concept of geometric cuts,used the whole neighborhood system,and got a more meaningful definition of the edge′s length.Then the priori term together with likelihood term formed the objective function of the image super-resolution reconstruction,and minimized by the steepest descent method,in order to achieve the purpose of reducing the length of the edges.Finally the results presented in this paper can show the convergence of the priori model,jagged edges are softened significantly,the images become more clearly,and have good visual effect.