[1]吴萌,王慧琴,景亚鹂.汉代建筑画像石的频率域自适应分解数字修复技术[J].西安建筑科技大学学报(自然科学版),2017,49(03):456-462.[doi:10.15986/j.1006-7930.2017.03.023]
　WU Meng,WANG Huiqin,JING Yali.Adaptive decomposed inpainting on the Han dynasty stone relief in frequency domain[J].J. Xi’an Univ. of Arch. & Tech.(Natural Science Edition),2017,49(03):456-462.[doi:10.15986/j.1006-7930.2017.03.023]
点击复制

汉代建筑画像石的频率域自适应分解数字修复技术()
分享到：

西安建筑科技大学学报（自然科学版）[ISSN:1006-7930/CN:61-1295/TU]

卷:
49
期数:
2017年03期
页码:
456-462
栏目:
出版日期:
2017-06-30

文章信息/Info

Title:
Adaptive decomposed inpainting on the Han dynasty stone relief in frequency domain
文章编号:
1006-7930(2017)03-0456-07
作者:
吴萌1; 王慧琴1; 景亚鹂2
（1. 西安建筑科技大学信控学院，陕西 西安 710055； 2. 西安碑林博物馆，陕西 西安 710001）
Author(s):
WU Meng1;  WANG Huiqin1;  JING Yali2
(1.School of Information and Control Engineering, Xi′an Univ. of Arch. & Tech., Xi′an 710055,China；2. Xi′an Beilin Museum, Xi′an 710001,China)
关键词:
画像石; 数字修复; 离散余弦变换; 轮廓波变换
Keywords:
stone relief;  digital inpainting;  DCT;  contourlet
分类号:
TU85
DOI:
10.15986/j.1006-7930.2017.03.023
文献标志码:
A
摘要:
画像石作为汉代石质建筑构件，其刻画的内容包含丰富的建筑题材，是古建研究的重要参考资料. 画像石经历两千多年各种病害侵蚀，导致其画面信息残缺不全，继续恶化将会使画像石所传承的历史信息消失殆尽. 因此急需利用现有残存信息进行画面内容复原，重建完整的汉代画像石图像. 数字修复是种非接触无损信息补全技术，可从能量泛函与纹理合成两个角度对缺失信息进行数字填充. 但是任意单一方案均难以满足汉代画像石修复既要结构信息过渡自然又要保证纹理细节的质感的需求. 频率域自适应分解数字修复技术，将汉代画像石图像在频率域DCT分解为纯结构与纯纹理两部分，分别用CDD曲率扩散技术重建结构信息；用Criminis样本块合成技术重建纹理信息；并将二者IDCT重构以获得最终修复效果. 数字修复过程中，用Contourlet轮廓波系数计算前景、背景、雕刻和病害四种不同对象的纹理能量参数. 针对不同值范围，对DCT分解位置、CDD曲率点扩散方向以及Criminisi合成样本大小等技术指标进行自适应动态调整. 该技术充分考虑了汉代画像石复杂的缺损机制，相比固定分解位置和单一的扩散方向及不变的样本尺寸，其修复过程更加灵活，仿真实验结果更加符合格式塔视觉心理学原则. 其数字修复结果作为实体修复的效果图，可避免直接修复对汉代画像石造成的二次损害.
Abstract:
As the Han Dynasty building components, the stone relief expresses the traditional architecture figures which have important information for the architects to research. More than two thousand years past, the information has been missing on and on. It is accurately needed to reconstruct the missing information. The left info includes lots of texture info and structure info. Digital image inpainting is a new technology that can restore the missing information without touching the real stone relief. But it is difficult to repair both structure info and texture info in the same time nether using neither PDE nor exemplar synthesis.DCT is used in adaptive frequency decomposed inpainting to decompose the stone relief image into pure structure part and pure texture part, then inpaint the structure info using CDD and inpaint the texture info using Criminis′s exemplar synthesis. Finally IDCF the two parts together to get the total result. When the Han dynasty relief is inpainted,Contourlet coefficient is used to design an energy parameter to calculate the texture complexity value of foreground, background, embossed and crack blocks. And then is used to choose a perfect decompose position adaptively. The energy parameter also helps to choose the order of CDD diffusing and the size of exemplar patches adaptively. The result of digital inpainting is more active than the methods which have invariant decompose position, diffusing order and exemplar size. It obeys the principle of Gestalt and can be used as an effect picture to repair the real Han dynasty stone relief. this can protect the relief from second time damage when it is repaired directly. .

参考文献/References:

[1]周学鹰. 对一幅 “汉代建筑画像石” 的重新释读[J]. 华中建筑, 2007, 21(6): 82-83. ZHOU Xueying. Research again for a piece of stone-pictures of architecture in China Han Dynasty[J]. Architectural History. 2007, 21(6): 82-83. [2]候锐.汉代画像石建筑题材研究[D].江西：江西师范大学,2013. HOU Rui. The study on the Architectural subject matter of the Han Stone [D].Jiangxi: Jiangxi Normal University. 2013. [3]于福艳.从画像石中的汉阙艺术看汉代建筑形式[J].山西建筑,2010,36 (16) :30-31. YU Fuyan. Look on Han architectural forms from Han palaces arts in stone portraits[J]. Shanxi Architecture,2010,36 (16) :30-31. [4]国家文物局. 石质文物病害分类与图示：WW/T0002-2007[S].北京：文物出版社,2008. Sdate Administration of Cultural Heritage. Classification and graphics on the deterioration of ancient stone objects：WW/T 0002-2007[S]. Beijing: Cultural Relics Publishing House, 2008. [5]BERTALMIO M, CASELLES V, HARO G, et al. Pde-based image and surface inpainting [M]. New York: Springer US, 2006: 33-61. [6]CRIMINISI A, PéREZ P, TOYAMA K. Region filling and object removal by exemplar-based image inpainting [J]. IEEE Transactions on image processing, 2004, 13(9): 1200-1212. [7]BERTALMIO M, VESE L, SAPIRO G, et al. Simultaneous structure and texture image inpainting[J]. IEEE transactions on image processing, 2003, 12(8): 882-889. [8]HESABI S, JAMZAD M, MAHDAVI-AMIRI N. Structure and texture image inpainting[C]//Signal and Image Processing, 2010 International Conference on. IEEE. Chennai, India: ICSIP, 2010: 119-124. [9]DIZDAROGLU B. An image completion method using decomposition [J]. Eurasip Journal on Advances in Signal Processing, 2011, 17(1):125-128. [10]PALLE E, JORGENSEN T, SONG M. Analysis of Fractals, Image Compression, Entropy Encoding, Karhunen-Loève Transforms [J]. Acta Applicandae Mathematicae, 2009, 108(3): 489-496. [11]MASTRIANI M, GAMBINI J. Fast cosine transform to increase speedup and efficiency of karhunen-loeve transform for lossing image compression [J]. International Journal of Information and Mathematical Sciences, 2016, 13(7):1607-1614. [12]RAJSHREE A, VENKATAPRASAD D, JOEL T, et al. Comparative Performance Analysis of Speckle Reduction Using Curvelet and Contourlet Transform forMedical Images[J]. Middle-East Journal of Scientific Research, 2016, 24(S1): 88-95. [13]于悦. 陕北画像石的地域特色和来源研究[D]. 长春：吉林大学, 2009. YU Yue. The study of regional characteristics and sources of Shanbei engraved stones [D]. Chang Chun: Jilin University. 2009. [14]CHAN T F, SHEN J. Nontexture inpainting by curvature-driven diffusions [J]. Journal of Visual Communication and Image Representation, 2001, 12(4): 436-449. [15]吴萌,王慧琴,李文怡. 多尺度唐墓室壁画病害标记及修复技术研究[J]. 计算机工程与应用，2016, 52(11): 169-174. WU Meng, WANG Huiqin, LI Wenyi. Research on multi-scale detection and image inpainting of Tang dynasty tomb murals [J]. Computer Engineering and Applications, 2016, 52(11)：169-174. [16]SUBBAN R, MUTHUKUMAR S, PASUPATHI P. Image restoration based on scene adaptive patch In-painting for tampered natural scenes [C]//Recent Advances in Intelligent Informatics. New York: Springer International Publishing, 2014: 65-72. [17]NISHIHARA A. DCT inpainting with patch shifting scheme[C]//Asia-Pacific Signal and Information Processing Association Annual Summit and Conference. Xi′an: APSIPAASC, 2011:38-47. [18]KOFFKA K. Principles of Gestalt psychology [M]. Abingdon: Routledge, 2013:27-32.

备注/Memo

备注/Memo:
收稿日期:2016-09-19 修改稿日期:2017-03-24 基金项目:陕西省教育厅专项科研计划项目（唐墓壁画的各项异性病害检测及多尺度数字修复研究） 作者简介:吴萌（1979-），女，博士，主要研究古代建筑与文物数字修复技术.E-mail：wumeng@xauat.edu.cn
更新日期/Last Update: 2017-08-27