【作者】 吴晓莉；

【导师】 贺汉根；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2007， 博士

【摘要】 流场可视化是科学研究与工程实践迫切需要解决的研究课题,由于流场数据一般都具有复杂的拓扑结构而且兼具大小、方向和空间位置等属性,如何形象而全面地显示这些信息一直是困扰研究人员的难题。目前对于流场的可视化还没有公认普遍适用的方法。本文以国家863项目中遥科学实验信息增强研究为具体应用背景,对基于纹理和基于特征的流场可视化方法进行了研究。重点围绕二维流场基于纹理的可视化算法及其效率提高、三维流场特征信息的检测与提取、三维流线的分布策略、算法的可信度等关键问题进行了深入研究。本文主要的创新点和研究成果包括:1)针对二维流场可视化,提出了一种基于粒子纹理融合的流场可视化方法。该方法采用块纹理平流替代传统LIC方法的纹理像素平流,提高了算法的效率,得到高效的动态流场可视化结果。实验表明,该方法独立于流场数据,具有通用性。与LIC方法相比具有高速的特点;与IBFV方法相比具有生成图像对比度高、物理意义明确等特点。2)针对三维流场可视化,提出了一种三维流线生成方法。该方法分为三个步骤:首先,在临界点的快速检测基础上,根据临界点处Jacobian矩阵特征值对临界点进行分类,并对临界点与种子点模板进行匹配;其次,种子点依照优先规则排序,并从这些种子点出发在物理空间计算出流线;最后,在图像空间由预先设置的阈值对流线进行间距控制,并根据深度检测来保留离视点最近的流线,使得屏幕上的输出结果清晰。仿真结果表明,论文提出的方法可以成功地解决流场可视化的特征提取与清晰表达这两个主要问题。3)针对特征提取问题,提出了一个临界点存在性定理,并给出了相关证明,在此基础上,提出了一种临界点快速检测算法。实验表明,该方法能够提取出流场中所有的临界点。4)给出了论文所提出方法的可信度分析。从误差分析和结果验证两个方面对本文提出的基于粒子纹理融合的流场可视化方法与特征提取的三维流线生成方法的可信度进行了深入分析。在误差分析过程中,给出了双线性插值函数误差余项的估计式,并给出了证明。理论分析和实验表明,方法是可信的。在上述研究工作的基础上,设计并实现了空间晶体生长实验信息增强可视化系统,形象逼真地显示出晶体炉内温场的分布以及流体的运动变化规律,得到了用户单位的充分肯定与一致好评。更多还原

【Abstract】 Visualization of flow field is a subject that urgently needs to be solved in both scientific research and engineer practice. Since there are complex topology structures in the flow data and the data always contain various attributes such as magnitude, direction, position and so on, it is a tough problem for researchers to visualize the information comprehensively. Until now, there is no method publicly affirmed as generally applicable in this field. With the application background for Tele-Science Experiment Information Enhancement Research of National 863 Project, the texture-based and feature-based flow visualization methods are studied in the thesis. The 2D texture-based flow visualization algorithm and its performance enhancement, the features detection and extraction in 3D field, the placement strategy of 3D streamline, and the credibility of the visualization algorithm, are deeply studied in the thesis.The main innovations and research results are as following:1) For the 2D flow visualization, a particle-texture blending based visualization method is presented. In the method, block texture advection instead of pixel texture advection of the traditional LIC method is adopted to enhance the performance, and dynamic flow visualization is obtained. Experiment results indicate that the method is independent of the complexity of the flow field data and is of generality. It is superior in performance compared with LIC method. It possesses higher contrast and clearer physical meaning than that of IBFV method.2) For the 3D flow visualization, a 3D streamline generation method is presented. The method is comprised of three processes. Firstly, based on the fast detection of critical points, the critical points are classified according to the eigenvalues of their Jacobian Matirx, and the match between the critical points and the seed templates is performed. Secondly, the seeds points are ordered in priority, and streamlines are generated in physical space. Finally, in image space, the distance between streamlines is controlled by the preset threshold value, and the nearest streamline to the observer is hold according to the depth detection, thus the display clarity is ensured. Simulation results indicate that the method can solve both the feature extraction and the clarity expression problems in flow visualization.3) For the feature extraction problem, a theorem for the existence of critical points in flow fields is proposed and proved. And a fast detection algorithm of critical points is presented based on the theory. The experiment results indicate that the algorithm can extract all the critical points in the flow field. 4) The credibility analysis of the methods presented in this thesis is proposed. From the two aspects of error analysis and result validation, the credibility analyses of both particle-texture blending method and the 3D feature extraction streamline method are deeply studied. In the process of error analyses, an estimation formula of error remaining for the bilinear interpolation function is put forward and proved. From the theory analysis and test result, the credibility of the method is verified.Based on the study above, a visualization system for the information enhancement in space crystal growth experiment is designed and implemented. The temperature distribution and the flow movement in the crystal stove are vividly displayed. The performance of the system is fully affirmed and well evaluated by the user.更多还原