基于稀疏图像序列的复杂雕塑形体3D建模方法的研究

发布时间：2019-02-20 21:27

【摘要】： 几何建模在CAD、产品设计以及远程展示、医疗医学等领域有着广泛应用。简便、快速地实现雕塑形体几何建模一直是人们期望解决的难题。基于计算机视觉的建模方法是近年研究热点，，但大多都停留在理论上，真正简便而又实用的系统并不多见。本文充分利用图像中两种较为可靠、稳定的信息：遮挡轮廓和阴影，研究雕塑形体的3D重构问题，提出了三种新的雕塑形体几何建模方法。首先，从分析遮挡轮廓约束入手，在传统体相交方法基础上，通过曲面变形技术实现约束条件的充分表达，得到一种简便的雕塑曲面重构方法。和其他方法相比，由于条件利用充分，可以由较少图像数目得到较好结果，而且建模结果是便于使用的样条表达形式。其次，针对遮挡轮廓提取难题以及上一方法中计算量较大问题，设计了一套简单的成像系统，不仅简化了遮挡轮廓提取，而且还利用剪影、剪影生成点信息改进了上面提出的方法，在效率、精度方面都有所提高。最后，在同样的成像系统下，研究了物体表面自遮挡阴影对凹陷区的形状限制，得到了相关的约束条件，并由此实现了一种简单的凹陷区域重构方法，使简便的被动式视觉方法有可能重构非凸的复杂形体。
[Abstract]:Geometric modeling is widely used in CAD, product design, remote display, medical medicine and so on. Simple and fast geometric modeling of sculptural body has always been a difficult problem that people expect to solve. Modeling methods based on computer vision have been a hot topic in recent years, but most of them remain in theory, and there are few real simple and practical systems. In this paper, we make full use of two reliable and stable information in the image: shading contour and shadow, study the 3D reconstruction of sculpture, and propose three new geometric modeling methods of sculpture. Firstly, based on the analysis of occlusion contour constraints and the traditional method of body intersection, a simple method of sculptural surface reconstruction is obtained by fully expressing the constraint conditions by means of surface deformation technology. Compared with other methods, due to the sufficient use of the conditions, better results can be obtained from fewer images, and the modeling result is a user-friendly spline representation. Secondly, a simple imaging system is designed to solve the difficult problem of occlusion contour extraction and the large amount of calculation in the previous method. It not only simplifies the extraction of occlusion contour, but also uses silhouette. The method proposed above is improved, and the efficiency and precision are improved. Finally, under the same imaging system, the shape limitation of the self-shading shadow on the surface of the object is studied, and the relevant constraints are obtained, and a simple method of reconstruction of the hollow region is realized. It is possible to reconstruct nonconvex complex objects by a simple passive visual method.
【学位授予单位】：汕头大学
【学位级别】：硕士
【学位授予年份】：2003
【分类号】：TP391.9

【相似文献】