中文摘要：

3D非结构化网格格心格式数据是近年流场数值模拟结果的常见形式,目前的可视化方法无法直接绘制此类数据,通常采用外推技术将其转换为格点格式数据后再进行绘制,导致数据精度损失,严重影响绘制质量.在多遍光线投射算法框架下,设计一种非结构化网格格心格式数据直接采样计算方法（避免外推）,使采样过程中的所有计算任务基于原始数据完成,以提高采样计算精度.具体为：设计了基于胞心值和单元梯度的采样点流场数据重构方法;采用基于面通量的格林公式计算单元梯度;考虑流场中物理量的相互关联,首次在流场可视化中引入Roe平均方法,用相邻单元胞心值构造面通量.分析和实验表明,该方法能明显提高采样计算精度,产生高质量的体绘制图像,使用户更准确地洞察和分析流场特性.

英文摘要：

3D unstructured-grid cell-centered data are commonly produced by the recent numerical simulations. For visualization, existing approaches usually pre-extrapolate cell-centered data into cell-vertexed data, which depress the rendering accuracy and the image quality. This paper proposes to do direct sampling for these cell-centered data avoiding extrapolation on the framework of multi-pass raycasting. During sampling, the whole computing work is done using the original data leading to a high rendering accuracy. The field at a sample is reconstructed by the cell-centered data and the cell-gradient. A ceil-gradient is well estimated by the Green-Gauss theorem with the aid of face-flux construction. Considering the relationship among the flow variables, this paper constructs the face-flux by the Roe-average method using the two cell-centered data values of the face-adjacencies. The analysis and experiments demonstrate that the approach gains high-accuracy sampling and a high-quality image leading to powerful insight into the characteristic of the flow fields.