一种基于形态成分分析的唐卡图像修复算法  

An Algorithm for Tangka Image Inpainting Based on Morphological Component Analysis

作　　者：刘仲民[1] 胡文瑾[1,2] 李战明[1] 

机构地区：[1]兰州理工大学电气工程与信息工程学院,兰州730050 [2]西北民族大学数学与计算机科学学院,兰州730030

出　　处：《计算机工程》2014年第1期250-253,267,共5页Computer Engineering

基　　金：国家自然科学基金资助项目(61162021);中央高校基本科研业务费专项基金资助项目(zyz2012078);甘肃省工业过程先进控制重点实验室基金资助项目(XJK0901)

摘　　要：基于形态成分分析的图像修复算法,通过增加全变分的方式,使得有毛糙边缘的分段光滑图像恢复效果较好,但易产生阶梯效应。针对该问题,将p-Laplace算子引入到基于形态成分分析的图像修复算法中,既保证图像在边缘的良好扩散能力,又避免在图像平滑区易产生虚假边缘的缺陷,同时对噪声有更好的抑制作用。实验结果表明,该算法对于唐卡图像中出现的折痕或划痕、斑块状破损有较好的修复能力。In the context of inpainting based on Morphological Component Analysis（MCA）, the imposition of a total variation penalty is useful, particularly well in recovering piecewise smooth objects, but easy to produce staircase. A new image inpainting method based on morphological component analysis is proposed. The proposed algorithm utilizes p-Laplace operator in the information spread not only along the edge direction, but also in gradient direction, which not only preserves edge, but also avoids staircase in the smooth area, and at the same time, the result is also less sensitive to noise. Experimental results for Tangka image which contains scratch and block loss show that the proposed method achieves better inpainting effect.

关 键 词：图像修复 唐卡图像 形态成分分析 全变分模型 P-LAPLACE算子 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]