Research On Key Techniques Of Feature-Based Texture Visualization For Complex Flow Field

Vector field visualization plays an important role in scientific research and engineer-ing. The data from numerical simulations such as CFD are getting larger and large, as the rapid development of high performance computer. It is quite a challenging issue how to analyze and understand those data. Feature-based and texture-based visualization have drawn much attention in visualization community to visually represent the flowing rule and to highlight the data interest the user. Although many progresses have achieved in these years, there are still many key problems to study, such as occlusion and clutter, the blurred image result of texture advection, how to measure the information of the flow field, and so on.In order to solve these problems, this paper studied the issues including informa-tion measurement of vortex feature, vortex extraction, high quality texture advection,3D sparse texture adaption. The main work and research achievements are as follows.1. To solve the problem of information measure and extraction of vortex, we pro-posed rotation entropy, define it in 2D consecutive flow field, and discussed the important properties of rotation entropy. Based on rotation entropy, vortex extraction algorithms are proposed in 2D/3D flow field. By analysing the linear descending distribution of rotation entropy along the radius direction of ideal vortex, we use linear fit to detect 2D/3D vor-tex feature. The experiment results show that rotation entropy can correctly represent the rotation information of 2D/3D flow field, one can correctly locate vortex features by ro-tation entropy-based vortex extraction methods, determine vortex regions, and avoid of wrong detection and missing of vortex feature.2. To solve the problem of blurred image result of texture advection, and visualize structural intensity fields, HLIC and IBFV-based high quality dynamic texture advection method HIBFV for 2D flow field was proposed. In order to improve the detail of texture, HLIC low-pass filter was designed and embedded to IBFV to realize high quality texture advection.The experiment results show that compared with LIC and IBFV, one can dy-namically represent structural intensity field with high quality by HIBFV to analyze data for users.3. To achieve the high quality texture advection and vortex detection for surface flow, we proposed a high quality texture advection method HIBFVS and rotation index based vortex detection method for curved surface. We realized high quality texture ad-vection by adding HLIC low-pass filter and streamline injection to IBFVS. We proposed the definition of rotation index, and detect vortex points and regions based on it. The experiment results show that one can represent high detail of surface flow with HIBFVS, and detect vortex feature correctly and rapidly with rotation index.4. To alleviate occlusion and clutter in 3D sparse texture visualization, we studied feature and view point-based texture adaption, proposed 3D rotation entropy and view point-based texture adaption method. With the rotation entropy-based vortex detection method and view direction and the definition of noise information alone view direction, a optimization function and rule are presented. With these rules, we designed method to gain optimized distribution of noise, realise the adaption of sparse noise. The experiment results show that more noise are distributed along the view direction.5. Feature detection and texture visualization modules are designed based on Par-aView. The modules are integrated to Para View by plugin mechanism. These modules support the reading of surface flow field data, high quality texture advection and vortex detection. The use of this system proved the availability, advancement and utility of the above results.