基于破损分类的西藏壁画图像修复关键技术研究

项目名称： 基于破损分类的西藏壁画图像修复关键技术研究

项目编号： No.61461047

项目类型： 地区科学基金项目

立项/批准年度： 2015

项目学科： 无线电电子学、电信技术

项目作者： 姜军

作者单位： 西藏大学

项目金额： 43万元

中文摘要： 通过对现有破损西藏壁画调研、数字化采集及破损特性分析，将破损西藏壁画图像初分为三大类：裂隙及字迹污染等造成的小区域破损、颜色脱落等导致的大区域破损和雨水冲刷造成的非完全破损。针对不同类的破损西藏壁画，展开适合于西藏壁画的图像修复关键技术研究，主要研究内容包括：（1）裂隙及字迹污染等西藏壁画图像的破损区域检测算法研究；（2）适合于西藏壁画小区域破损的图像修复算法研究；（3）能有效保持结构连通的大区域破损西藏壁画图像修复关键技术研究；（4）适用于非完全破损西藏壁画的图像修复算法研究。本项目研究是数字图像修复技术在西藏壁画数字化保护与修复这一实际应用的有益探索。

中文关键词： 西藏壁画；图像修复；破损分类；样本块；相似性度量

英文摘要： This project studies the current situation of Tibet's murals and analyzes the damage characterization of the digital imaging of Tibet's murals. The damage Tibet's murals are divided into three types: samll missing regions with crack and writing, half-missing regions with rain wash, and larger missing regions with color loss. To address the different damage types of Tibet's murals, the key techniques of image completion for broken Tibet's murals are studied. The main research content includes the following four aspects. (1) the damage detection algorithm is researched for the broken Tibet's murals image including the crack and writing; (2) image inpainting method is studied for broken Tibet's murals with samll missing regions; (3) key techniques of image completion for broken Tibet's murals with larger missing regions are studied to maintain the structure connectivity; and (4) image completion methods for broken Tibet's murals with half-missing regions are researched. The research of this project is beneficial exploration for the digital image completion technology in the practical applications of the digitalized protection and virtual restoration of Tibet's murals.

英文关键词： Tibet's murals;digital image inpainting;damage classification;exemplar block;similarity measurement