一种基于形态成分分析的唐卡图像修复算法

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。

An Algorithm for Tangka Image Inpainting Based on Morphological Component Analysis

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.