摘要
传统的图像修复算法速度慢,对大面积的破损区域修复效果较差.本文针对这一缺点,首先对待修复图像进行小波分解,使得图像的破损区域在低频部分留下的空洞大为缩小,然后利用基于快速行进法(FMM)的图像修复算法修复低频部分的破损区域,再利用低频信息来预测相应的高频信息,最后进行小波重构,并对受损部分进行自然化处理,得到修复的图像.仿真实验结果表明,本文提出的算法速度快,修复结果基本恢复了原有的视觉效果.
To overcome the shortcoming of conventional methods,being slow and incompetent for large area repairing task ,a new image repairing algorithm is proposed in this paper. First,the image to be repaired is decomposed by means of wavelet,which makes the region to be repaired dwindle to a great extent. Then, the low frequency part of the region is made to be repaired with the help of image repairing algorithm based on Fast Marching Method （FMM）, and the relevant high frequency information is predicted with the help of low frequency information. Finally,the image is reconstructed with wavelet,and the repaired region is subjected to natural treatment. Thus, the repaired image is obtained. Results show that this algorithm can be operated at a high speed and can restore the Original visual effect on the whole.
出处
《内蒙古工业大学学报（自然科学版）》
2008年第4期259-263,共5页
Journal of Inner Mongolia University of Technology：Natural Science Edition
基金
内蒙古工业大学重点研究项目
编号:ZD200601
内蒙古自治区自然科学基金项目资助
编号:200711020801
关键词
图像修复
小波分解
小波重构
快速行进法
image repainting
wavelet decomposition
wavelet reconstruction
fast marching method
