改进的TV模型图像修复算法  

Improved algorithm based on TV model for image inpainting

作　　者：林云莉[1] 赵俊红[1] 朱学峰[1] 胡永健[2] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640 [2]华南理工大学电子与信息学院,广东广州510640

出　　处：《计算机工程与设计》2010年第4期776-779,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60772115;60572140)

摘　　要：分析了基于整体变分(total variation,TV)模型的图像修复算法,TV模型修复算法只使用各向异性扩散,TV模型各向异性扩散仅向图像边缘方向扩散,容易在平滑区域引入阶梯效应。提出了一种改进的图像修复算法,该算法同时结合了各向同性和各向异性扩散,利用区域频率差异实现了在不同的区域使用不同的迭代方程,有效避免了原始算法引入的阶梯效应,同时在平滑区域提高了迭代效率。Matlab环境下的仿真结果表明,改进算法的修复效果和峰值信噪比的计算结果均明显优于原始算法。The image inpainting algorithm based on total variation（TV） model is analyzed.The inpainting algorithm of TV model only use the anisotropic diffusion.The anisotropic diffusion of TV model just diffuses along the image edge and introduces step effect easy in smooth regional.Now an improved image inpainting algorithm is proposed.Improved algorithm combines isotropic with anisotropic diffusion and uses the regional frequency difference to realize different iterative equation on the different regional.The step effect in original algorithm is effective avoided and raise iterative efficiency on smooth regional.The simulation results in Matlab environment demonstrate that improved algorithm＇s inpainting effect and the result of peak signal to noise ratio are both better than original algorithm.

关 键 词：图像修复 整体变分模型 各向同性 各向异性 阶梯效应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]