激波特征可视化是流场可视化的重要内容.目前主流的激波特征提取方法是先采用正则马赫教检测激波,再设计过滤算法去除噪声.已有方法在计算正则马赫数时,并未区分压力梯度和密度梯度；在过滤噪声时,有效性依赖于数据集本身,适应性和准确性差.当流场中存在强度不同的多激波特征时,往往在过滤噪声的同时也滤掉了“弱激波”,且即便对于单激波特征,也常出现激波面不连通甚至断裂现象.论述了基于压力梯度计算正则马赫数的必要性；利用激波物理特性,结合光线投射算法优势,提出了一种基于两级采样的多激波特征可视化方法;并针对拓扑复杂的3D非结构化网格数据,在GPU上设计实现了算法.实验表明,该方法能自动识别并过滤噪声,即便对包含多激波特征的复杂流场,也具有很好的适应性和准确性,过滤效果明显优于已有方法,且对较大规模非结构化网格数据,绘制性能可满足实时交互.%Shock feature visualization plays an important role in flow visualization. To perform shock extraction, existing methods usually carry out shock detection with the normal Mach first and then filter the noise. When computing the normal Mach, they do not distinguish between the pressure gradient and the density gradient. Moreover, their noise filter has poor adaptability and accuracy with the availability depending on the test data. It usually makes the weak shock filtered along with the noise especially when there are multi-shock features with different strengths in flows. Besides this, the shock surface may be unconnected or split even for the single-shock flows. It is necessary to perform shock detection with the pressure gradient. On the basis of the shock attributes, a novel visualization method is proposed for multi-shock features using two-level sampling on the framework of recasting. The work is performed on GPU for the 3D unstructured-grid data with the complicated topology. The experimental results show that our method can automatically filter the noise and its adaptability and accuracy are much better than those of the existing methods even for the multi-shock flows. Meanwhile, a real-time performance is achieved for the large unstructured-grid data.
展开▼
