Research And Implementation Of Feature-based Streamline Selection Method For Flow Field

Streamline Visualization is an important method for visualizing vector field visualization with characteristics of intuitiveness and interactive analysis,and is widely used in engineering practices.Streamline visualization can be formulated as the problem of seed placement or streamline selection.Seed placement depends on the number and position of seeds.But because the type and location of features in flow field is difficult to be predicted,seed placement cannot ensure the integrity of the flow field features.Compared with seed placement,the advantage of the streamline selection is that initial streamlines can cover all regions of flow field,as long as an appropriate selection method being implemented,most features will be extracted accurately.However,the problems of missing features and visual cluster exist in most streamline selection algorithms.To solve these problems,we present the streamline selection method for 2D and 3D.The main work of this paper can be summarized as follows:1.The paper presents a feature type-based streamline cluster selection method for 2D flow field.Firstly,the streamline feature type is defined based on the type of critical points in 2D flow filed.Given a large number of randomly or uniformly seeded streamlines,the proposed approach judges the feature type for each streamline by using methods of winding angle and information entropy.All streamlines are then clustered according to their feature types and positions,ensuring the integrity of field features in streamline selection.A streamline similarity measure based on the dynamic time warping algorithm and the mean of closest point distances is presented to avoid selecting redundant streamlines.Finally,the paper designs a strategy for streamline selection.Results show that the developed algorithm can reflect the key features of a flow field more effectively and greatly improve the readability of streamline visualizations.2.The paper proposes a streamline selection method of predominantly to extract the vortex features for 3D flow field.Firstly,a lot of streamlines are randomly seeded to cover all regions of flow field.By utilizing the information entropy,we successfully extract all important streamlines which can reflect the variation of flow field.Then the vortex core lines are extracted thorough the method of rotation entropy.Based on vortex core lines,vortex streamlines are easily selected from the important streamlines.Excluding the vortex streamlines,the rest of the important streamlines are clustered by the position of point that has the maximum information entropy,separating streamline subsets that belong to different feature regions.A streamline simplified method is designed to ease the problem of overlapping.Finally,streamlines are selected from the previous separated feature streamline subsets.In addition,we design a parallel acceleration method for our algorithm on CUDA.Experimental results show that our method can effectively capture feature structures in 3D flow field and improve the streamline visualization significantly.3.On the basis of the work above,a feature-based streamline selection visualization software has been designed and implemented.The software consists of data preprocessing module,entropy calculation module,CPU and GPU collaborative control module,streamline growing module and streamline rendering module.We use multiple sets of data to verify the correctness and validity of the proposed methods.