文献类型：期刊文献

中文题名：一种改进的DWT-FMM图像修复算法

英文题名：An Improved Image Inpainting Algorithm Based on DWT-FMM

作者：补朝阳[1];李尊[2,3]

第一作者：补朝阳

机构：[1]新乡学院化学化工学院;[2]新乡学院物理与电子工程学院;[3]武汉科技大学信息科学与工程学院

第一机构：新乡学院化学化工学院

年份：2018

卷号：35

期号：6

起止页码：44-47

中文期刊名：新乡学院学报

外文期刊名：Journal of Xinxiang University

基金：国家自然科学基金项目(U1704132);河南省高等学校重点科研项目(17A470013)

语种：中文

中文关键词：图像修复;小波变换;二阶有限差分;黄金率

外文关键词：image inpainting;wavelet transform;second finite difference;gold rate

摘要：为了提高图像修复算法的精度,避免错误信息修复及加强细节处理,在传统小波变换图像修复算法的基础上提出用二阶有限差分替代Eikonal方程的一阶有限差分形式,提高快速行进算法的精度;将黄金率引入到权重函数中,提高方向因子的比重,增强修复的方向性;采用约束最小二乘法滤波,使修复效果自然。仿真实验表明,算法在保证修复效率的基础上,提高了修复精度。
An improved image inpainting based on DWT-FMM was discussed to improve the accuracy of the algorithm, to avoid error information and enhance the details. The first finite difference was replaced by the second finite difference in Eikonal equation which was proposed to improve the precision of the fast marching method.The gold rate was introduced into the weight function to improve the proportion of the direction factor and enhance the orientation of the repair.Constrained least squares filter was employed to make the inpainting natural. Experiment results have showed that on the basis of inpainting efficiency, the proposed algorithm has improved precision and has great significance in research.