整体变分算法在图像修补中的应用研究

对如何将整体变分模型用于图像修补进行了讨论．主要研究整体变分算法在待修补区为长条形空白区域的图像修补，并根据待修补区为长条形的特点，通过引入权值，对整体变分模型的离散化算法作了改进，使该算法利用邻域信息仅对待修补空白区域进行填充，而不改变待修补区邻域的像素值．实验表明，文中算法对窄长条状或线状空白区域的图像修补是有效的。

Image Restoration Using Total Variation Approach

The total variation algorithm applied to restore image is fulfiled without taking into account denoising. Then the main research effort focuses on the image inpainting for long but narrow rectangle blank area, and an improved image inpainting method based on the total variation algorithm is presented. It introduces different weights into the algorithm according to the features of blank area to be inpainted, and it only restores the lost information for the blank area according to the surrounding information but keeping the surrounding information unchanged. Besides, the proposed algorithm makes full use of the surrounding information during the inpainting process in order to ensure the likeness of inpainting areas. Plenty of experiments show that the proposed method is effective for the image inpainting, especially for the long and narrow blank area as well as for the linear information lost area.