基于尺度空间的小波纹理描述算法的壁画修复  

Murals inpainting of the wavelet texture description algorithm based on scale space

作　　者：任小康[1] 邓琳凯 

机构地区：[1]西北师范大学计算机科学与工程学院,甘肃兰州730070

出　　处：《计算机工程与科学》2014年第11期2191-2195,共5页Computer Engineering & Science

摘　　要：近年来,壁画的数字化修复技术得到了很大的发展,但是其主要算法在图像纹理方向的修复效果一般。为了解决这些问题,还原壁画的原有样貌,提出了一种简单快速的壁画修复算法。该算法在改进的Telea算法的基础上,根据图像纹理的多尺度特征性,利用小波纹理描述算法对壁画的纹理结构进行了修复处理,运用非局部梯度算子,通过计算待修复区域的像素点的非局部梯度值并依据曲率选取的全局最优方向进行扩散,达到壁画纹理修复的目的。实验结果表明了该算法的有效性。In recent years,murals digital restoration is greatly developed,but its main algorithms has no obvious effect on the restoration of image texture direction.In order to solve this problem and restore the original appearance of murals,we propose a simple and rapid mural repairing algorithm in which the Telea algorithm is improved.According to the multi-scale feature of image texture,the proposal repairs texture structure of murals by using the wavelet texture description algorithm.To repair the murals＇ texture,nonlocal gradient operator is used to calculate the nonlocal gradient value of pixels in unrepaired areas and diffusion is performed according to the global optimal direction.The experimental results demonstrate the effectiveness of the algorithm.

关 键 词：壁画修复 Telea算法 尺度空间 小波纹理描述算法 梯度算子 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]