摘要：In computer graphics, non-photorealistic rendering(NPR) is an important branch. NPR technology is to achieve a variety of artistic effects through the computer, such as oil painting, cartoon, watercolor and other effects. The purpose of this paper is to automatically convert the 3D model into two-dimensional Chinese ink painting effect on graphics processing unit(GPU), and has improved the traditional algorithm that has some lacks of rendering effect. The algorithm is divided into two major steps: contour line rendering, and interior rendering. For contour line rendering, on the basis of the traditional extraction of contours, this paper adds self-bold, particle diffusion and other steps. For interior rendering, on the basis of the traditional layered rendering of diffuse lighting, two methods are added. The first method is based on the mean filtering, and the filter kernel is stratified by the principle of percentage-closer soft shadows(PCSS). The second method is Noise texture mapping, to obtain an ink diffusion effect.

摘要：The building information model/modeling （BIM） technology is currently applied in a broad range of applications and research for facility management （FM）, while the BIM-based mobile FM is difficult owing to various factors and environments. For example, the mobile applications usually require frequent cross-equipment compatibility. This paper proposes a reasonable BIM-based FM cross-platform framework and develops a mobile application on the basis of an existing BlM-based FM system. The developed mobile application is applied in a case study of a metro station project in Guangzhou to verify its effectiveness in FM practice. It helps maintenance staff in viewing BIMs, accessing related information, and updating maintenance records in a unique platform. The test results demonstrate that the proposed BIM-based cross-platform framework meet the FM application requirements and supports the extension of FM functions.

摘要：This paper proposes a new method to simplify mesh in 3D terrain. The 3D terrain is presented by digital elevation model. First, Laplace operator is introduced to calculate sharp degree of mesh point, which indicates the variation trend of the terrain. Through setting a critical value of sharp degree, feature points are selected. Second, critical mesh points are extracted by an recursive process, and constitute the simplified mesh. Third, the algorithm of linear-square interpolation is employed to restore the characteris- tics of the terrain. Last, the terrain is rendered with color and texture. The experimental results demonstrate that this method can compress data by 16% and the error is lower than 10%.

摘要：In the process of seismic data interpretation, the extraction of a horizon or a fault is generally needed. In this paper we present a fast extraction method. First select some feature points interactively, then reconstruct the surface according to the selected feature points by using B-spline interpolation curve or surface. In order to solve the error-drawing problem appeared in the process of interactive rendering, which is caused by the change of sampling interval as the view point changes, we combine shear-warp and splatting algorithm to render the surface. The rendering of seismic data and specific surface in our work are achieved on GPU platform using CUDA programming language, which make it able to meet the requirements of real-time rendering.

摘要：针对无任何预知信息下的扫描点云数据配准问题,通过引入新的匹配点对度量准则和改进最近点迭代算法,提出一种扫描点云数据的自动配准方法.该方法分为初始配准和精细配准2个阶段.初始配准阶段中,在考虑孤立点的曲率相似度的基础上,通过引入一种新的点的邻域曲率相似度度量准则,构造出一个有效的一一对应的初始匹配点对数组;然后根据刚体变换的特点和不变量引入距离约束和超线段约束,对初始匹配点对进行过滤;最后利用最终得到的匹配点对的几何信息计算初始配准参数.在精细配准阶段,构造了参与最近点迭代算法的有效初始点集,并改进了最近点的计算过程.数值实例结果表明,文中方法初始配准效果良好,二次配准效果更加准确,达到了不同视角扫描点云数据配准的要求.