一种新的基于修正LOT模型的图像放大算法

摘要

Image zooming schemes are becoming increasingly popular due to the wide use of digital imaging devices and the increasing sensor-capturing capabilities in spatial resolution. Based on duality strategy,in this paper,propose a Chambolle dual algorithm based on modified LOT model for image zooming. Motivated by the connections between the dual variable and the unit normal vector,in the first step,instead of smoothing the normal vector directly,compute the unit normal vector by means of dual formulation. In the second step,re-construct the zooming image using the Chambolle dual algorithm. In particular,the two-step models take full advantages of both the ROF model and the LLT model so that the proposed image zooming methods can retrieve relatively smooth edges while alleviating the jagged effects. Experimental results show the good visualization and better SNR of the zooming algorithm by comparing it with bilinear interpola-tion image zooming method and LOT based image zooming method.%由于图像采集设备以及成像手段的日益更新,在图像处理领域中图像放大技术已经得到越来越多的关注和重视.基于对偶策略,文中提出了一种基于修正LOT模型的Chambolle对偶图像放大算法. 由于对偶变量可以等价为图像的法向量,提出在模型的第一步,直接计算对偶变量来代替光滑化法向量. 第二步,采用Chambolle对偶算法重构放大图像.两步模型由于结合了ROF模型与LLT模型,该模型可以更好地保持图像边缘的信息,同时在一定程度上也避免了"锯齿效应冶的产生,使图像边缘尽可能保持平滑. 数值实验表明,通过与双线性插值、LOT模型图像放大方法进行比较,文中算法取得了理想的视觉效果与更高的信噪比.