摘要
基于TV(total variation)模型的修复算法有较好的恢复效果,但对参数的选取比较敏感,且运算量大。文章提出了基于TV模型的改进的自适应算法,可根据破损区域外部参考像素对待修补点的相关度,通过设置不同的参数和权值,将不同形状的待修复区域所用的不同算法统一表示,使其应用范围更广、速度更快;此外,在迭代过程中,设置不同的参数以解决参数选取的敏感问题,从而达到更好的修复效果。实验表明,该算法能高效、稳定地处理破损区域的图像信息。
The inpainting algorithm based on total variation（TV） model performs well,but it is time-consuming and sensitive to parameter selection.This paper proposes an improved adaptive image inpainting algorithm based on TV model.According to the correlation between inpainting points and reference pixels of surrounding information of the damaged area,different weights and parameters are introduced.It is a unified expression of different algorithms for different features of the damaged areas,which is time-saving and can be widely used.In view of the iterative process,different parameters are set to solve the sensitive problem and get better inpainting effect.The experimental results show that the improved algorithm can treat image information on the damaged area effectively and steadily.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2010年第12期1916-1920,共5页
Journal of Hefei University of Technology：Natural Science
基金
教育部科学技术研究重大资助项目(309017)
关键词
整体变分
图像修复
统一表达式
自适应
total variation（TV）
image inpainting
unified expression
adaptation
