破损藏式古唐卡数字修复中的自适应修复模型研究

项目名称： 破损藏式古唐卡数字修复中的自适应修复模型研究

项目编号： No.61461048

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： 仁青诺布

作者单位： 西藏大学

项目金额： 40万元

中文摘要： 藏式古唐卡具有极高的历史、科学和艺术价值，然而，目前针对藏式古唐卡的数字化修复工作存在着修复精度不高、耗时长等缺点。为了克服这些问题，本项目拟构建针对藏式古唐卡不同损伤区域的自适应破损区域检测算法，以准确地确定破损区域的边缘；拟构建自适应置信度计算方法和针对彩色图像的等照线计算方法，以更准确地计算边缘点的优先级；拟构建模板块自适应选择方法、最佳样本块自适应搜索方法和块效应自适应消除方法，以更高的精度和更短的耗时自适应地修复藏式古唐卡。本项目的研究成果，可以对数字图像修复、数字图像处理等方面的研究提供重要的理论参考价值，同时，也为抢救藏式古唐卡这样的历史文化遗产提供新思路，对传承和保护藏民族的优秀传统文化具有深远意义。

中文关键词： 图像修复；受损图像修复；图像颜色修复；藏式古唐卡；自适应修复模型

英文摘要： The ancient Tibetan thangka has very high historical, scientific and artistic values, however, the current digital inpainting methods for Tibetan thangka have some weaknesses, such as low accuracy and high time-consuming. In order to overcome these problems, this project will establish an adaptive damaged area detection algorithm in order to determine the edges of the different damaged area types of ancient Tibetan thangka accurately. At the same time, an adaptive isophote calculation method for color images will be constructed, and an adaptive confidence calculation method also will be built, thus, the priority of the edge pixel can be calculated more accurately. Meanwhile, the methods of adaptive template block selection, adaptive optimal sample block searching and adaptive blocking effect reduction will be assembled as adaptive inpainting model which can inpaint ancient Tibetan thangka with high accuracy and low time-consuming. The research results can provide important theoretical reference values to digital image inpainting and digital image processing, and also can provide new ideas to rescue the historical and cultural heritage such as the Tibetan thangka. Therefore, this project has far-reaching significance to protect and inherit the excellent Tibetan traditional culture.

英文关键词： image inpainting;damaged image inpainting;image color inpainting;ancient Tibetan thangka;adaptive inpainting model