当前位置:主页 > 社科论文 > 考古论文 >

青铜器文物碎片虚拟拼接技术研究及几何特征提取

发布时间:2018-06-08 17:43

  本文选题:青铜器碎片 + 二维拼接 ; 参考:《中北大学》2017年硕士论文


【摘要】:由于年代久远等原因,相当多的青铜器出土时存在破损情况,利用计算机辅助碎片虚拟拼接是青铜器文物修复保护和数字化处理的关键技术之一。针对青铜器具有非刚体性,且碎片的结构特征或纹理特征往往不明显,导致碎片模型匹配困难的问题,本文就青铜器碎片虚拟拼接技术展开如下研究:(1)提出青铜器碎片分类拼接思路。依据青铜器薄壁文物碎片的特点将其按照是否具有明显三维特征以及碎片大小分为二维碎片和三维碎片,并且后续拼接采用分类拼接方式,即分别在二维平面和三维空间进行匹配。(2)提出利用曲率特征虚拟拼接青铜器小碎片的方法。对青铜器小碎片分别进行轮廓提取,并在平滑处理提取结果的基础上计算其曲率;以曲率的局部极大值点作候选角点,再根据支撑域消除圆角点和伪角点;计算相邻两角点间的弦长,以弦长作为特征序列进行匹配;在平移旋转过程中利用向量模的不变性进行匹配精化处理,得到青铜器小碎片的最终拼接结果。结果表明该方法能够有效提高拼接效率和匹配效果。(3)提出结合主成分分析和快速傅里叶变换进行青铜器三维碎片虚拟拼接的方法。将轮廓曲线表示为若干特征曲线段序列,并将每段分别从世界坐标系变换到用PCA方法建立的局部坐标系下,然后投影到其中的两个坐标平面上;以该两个坐标平面上轮廓投影点的傅里叶级数和能量值为依据进行粗匹配;将基于粗匹配得到的过渡矩阵将两个待匹配碎片的轮廓线经坐标变换后投影到第三个坐标平面,通过寻找其最长公共边进行精匹配,从而得到匹配结果。(4)设计和开发了青铜器文物几何特征尺寸提取系统。该系统操作简易,能快速地提取青铜器几何特征量,包括口径、底径、高度等,为后续人工修复青铜器提供可靠数据资料。
[Abstract]:Due to the old age and so on, quite a lot of bronze wares are damaged when unearthed. The use of computer aided virtual splicing is one of the key technologies for the restoration, protection and digital processing of bronze relics. In view of the non-rigid nature of the bronze ware and the fact that the structural or texture features of the fragments are not obvious, it is difficult to match the debris models. In this paper, the virtual splicing technology of bronze fragments is studied as follows: 1) the idea of classifying and stitching bronze fragments is put forward in this paper. According to the characteristics of the thin-walled fragments of bronze ware, the fragments are divided into two dimensional fragments and three dimensional fragments according to whether they have obvious three-dimensional characteristics and the size of fragments, and the subsequent splicing methods are classified and stitched. This paper proposes a method of virtual splicing of small pieces of bronzes using curvature characteristics in two dimensional plane and three dimensional space respectively. The small pieces of bronze ware were extracted from the contours, and their curvature was calculated on the basis of the results of smoothing, the local maximum of curvature was used as the candidate corner point, and the circular corner and pseudo-corner were eliminated according to the support region. The chord length between two adjacent corners is calculated, and the string length is used as the characteristic sequence to match. The final splicing results of small pieces of bronze ware are obtained by using the invariance of vector mode in the process of translation and rotation. The results show that this method can effectively improve the mosaic efficiency and matching effect. (3) A method combining principal component analysis (PCA) and fast Fourier transform (FFT) for virtual splicing of bronze pieces is proposed. The contour curve is expressed as a series of characteristic curve segments, and each segment is transformed from the world coordinate system to the local coordinate system established by PCA method, and then projected to two coordinate planes. Based on the Fourier series and energy values of the contour projection points on the two coordinate planes, the transition matrix based on the rough matching is used to project the contours of the two pieces to the third coordinate plane after coordinate transformation. By searching for the longest common edge for fine matching, the matching result. 4) the size extraction system of the geometric feature of bronze relics is designed and developed. The system is easy to operate and can quickly extract the geometric characteristics of bronze ware, including caliber, bottom diameter, height, etc., which provides reliable data for subsequent artificial restoration of bronze ware.
【学位授予单位】:中北大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:K876.41;TP391.41

【相似文献】

相关期刊论文 前1条

1 叶恒;数码摄影与拼接技术[J];中国摄影家;2005年11期

相关会议论文 前5条

1 倪勇;黄永苍;陈勇军;陈旭;闫晓;李彬育;陈驹;;三叠大板梁拼接技术研究[A];中国钢结构协会锅炉钢结构分会第十三届年会论文集[C];2012年

2 程良骏;赵乐\,

本文编号:1996644


资料下载
论文发表

本文链接:https://www.wllwen.com/shekelunwen/kgx/1996644.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户fea7f***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com