基于全变分模型改进的图像修复算法应用  

Application of Improved Inpainting Algorithm Based on TV Model

作　　者：谢正伟[1] 王创新[1] XIE Zhengwei WANG Chuangxin(School of Electrical and Information Engineering, Changsha University of Science and Technology, Changsha 410004, China)

机构地区：[1]长沙理工大学电气与信息工程学院,湖南长沙410004

出　　处：《电子科技》2017年第1期61-64,68,共5页Electronic Science and Technology

摘　　要：传统全变分(TV)模型对非纹理破损图像有较好的修复效果,但TV算法对参数敏感,导致修复图像容易在平滑区域产生阶梯效应。文中在传统TV模型的基础上,结合非线性扩散的思想,提出了一种针对参数的自适应迭代函数图像修复算法。实验结果表明,该算法在修复过程中扩散系数和自适应迭代函数的共同作用,消除了阶梯效应;修复的图像自然过渡,边缘信息保存良好。The Total Variation （TV） has a positive effect for non-texture damaged image; Owning to the TV al- gorithm of parameter sensitivity, the result of inpainting is easy to occur staircase effect. This paper proposes an im- proved image inpainting algorithm based on TV model, with the nonlinear diffusion theory and an adaptive iterative function for the parameter; In consequence of the influence of diffusion coefficient and an adaptive iterative function, the improved algorithm eliminate the staircase effect. The consequence of image inpainting transit naturally and pro- tect the edge information.

关 键 词：TV模型 图像修复 非线性扩散 自适应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]