基于图像的旋转体文物重建系统的研究与实现

发布时间：2018-04-14 22:48

本文选题：旋转体文物 + 三维点云模型　；参考：《山东大学》2010年硕士论文

【摘要】： 博物馆是收集、保护和展示各种重要文物的场所,但展览空间有限,在实现资源共享、传播古老中国文化方面存在诸多限制。随着计算机和网络的发展,用数字化手段建成基于网络的数字化博物馆系统,可以较好地解决这些问题。而文物的三维重建技术则是数字化博物馆建设中最关键的技术之一。旋转体是文物中比较常见的一类。在出土的大量文物中,许多陶器、瓷器、铁器、铜器等都是旋转体,旋转体的三维重建技术在文物保护、展示等方面具有重大意义。因此,研究如何从现实世界直接和快速地重建旋转体的三维模型,逐渐成为计算机图形学和计算机视觉领域的研究者努力的目标。根据使用设备的不同,旋转体的三维重建主要可以分为两类方法：基于三维扫描的方法和基于二维图像的方法。前者能够精确获取旋转体的三维点云模型,但具有代价昂贵、使用方法具有侵略性、适用范围受限、数据采集不完善等缺点,并不适用于旋转体文物。后者则以使用摄像设备的方法为主,具备成本低、设备灵活、彩色纹理获取简单等特点。基于二维图像的方法也主要有两类：一类是由二维图像重建出三维点云模型,再进行表面重建,不但适用于旋转体对象,也适用于其它现实物体,但获取的点云模型通常存在噪声多、数据点稀疏不完整等问题。另一类充分利用了旋转体特有的几何结构,通过二维图像直接估计旋转体的旋转轴和母曲线,是旋转体三维重建研究的主流方法,但通常难以获取完整的纹理信息。本文在前人工作的基础上,研究设计基于多幅图像的旋转体文物重建系统。该系统使用价格较低的照相机设备取代三维扫描仪来获取三维点云模型,既可以解决三维扫描仪设备价格昂贵的问题,还能获取到完整的纹理信息。同时充分利用旋转体固有的几何特性,通过计算旋转轴和母曲线重建三维表面模型,比通用的重建方法更有针对性,能有效处理点云模型获取过程中遗留下来的噪声多、数据点稀疏不完整等问题。本文的主要工作包括以下几个方面： 1)利用旋转体结构上特有的几何特性,研究并实现基于图像的旋转体文物重建系统,并对系统中采用的关键技术进行研究,分析实验结果 2)采用SDM方法拟合母曲线。为了降低点云稀疏和噪声点多等问题对算法的影响,在进行曲线拟合之前,引入移动最小二乘法进行预处理,改善重建结果。
[Abstract]:Museums are places for collecting, protecting and displaying all kinds of important cultural relics, but the exhibition space is limited, and there are many limitations in the realization of resource sharing and the dissemination of ancient Chinese culture.With the development of computer and network, digital museum system based on network can solve these problems.The three-dimensional reconstruction of cultural relics is one of the key technologies in the construction of digital museum.Rotating objects are a common kind of cultural relics.Among a large number of cultural relics unearthed, many pottery, porcelain, iron and copper objects are rotating objects. The three-dimensional reconstruction technology of rotating objects is of great significance in the protection and display of cultural relics.Therefore, the research on how to reconstruct the 3D model of rotator directly and quickly from the real world has gradually become the goal of researchers in the field of computer graphics and computer vision.According to the different equipments used, the 3D reconstruction of rotating objects can be divided into two types: the method based on 3D scanning and the method based on 2D image.The former can accurately obtain the 3D point cloud model of the rotating body, but it is expensive, the method is aggressive, the scope of application is limited, and the data collection is not perfect, so it is not suitable for the cultural relics of rotating body.The latter has the advantages of low cost, flexible equipment and simple color texture acquisition.There are also two kinds of methods based on 2D image: one is to reconstruct 3D point cloud model from 2D image, and then to reconstruct the surface, which is not only suitable for rotating object, but also applicable to other real objects.However, the obtained point cloud model usually has many problems, such as noise, sparse data points, etc.The other is the main method of 3D reconstruction, which makes full use of the special geometric structure of rotator and estimates the rotation axis and generating curve directly by two-dimensional image. However, it is difficult to obtain the complete texture information.Based on the previous work, this paper studies and designs a rotatory object reconstruction system based on multiple images.In this system, the low cost camera equipment is used to replace the 3D scanner to obtain the 3D point cloud model, which can not only solve the problem of expensive 3D scanner equipment, but also obtain the complete texture information.At the same time, making full use of the inherent geometric characteristics of the rotating body, the 3D surface model can be reconstructed by calculating the rotation axis and the parent curve, which is more targeted than the general reconstruction method, and can effectively deal with the noise left over in the acquisition process of the point cloud model.Data points sparse incomplete and other problems.The main work of this paper includes the following aspects:1) based on the unique geometric characteristics of the rotator structure, the image based cultural relic reconstruction system is studied and implemented. The key technologies used in the system are studied, and the experimental results are analyzed.2) SDM method was used to fit the parent curve.In order to reduce the influence of point cloud sparsity and noise points on the algorithm, the moving least square method is introduced to preprocess the algorithm before curve fitting to improve the reconstruction results.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP311.52

【引证文献】