Streamlines Analysis And Visualization Using Cluster Method

This paper proposes a method with the use of cluster analysis and visualization methods for the flow visualization of flow lines, this method includes cluster analysis of the flow field, self-tuning spectral cluster, simplify of features and illuminated streamlines.For many problems caused by the layout of seed points, this paper proposes a cluster method to the streamlines data which divides those streamlines with the same geometric characteristics into a set of categories.This paper applies the K-means, hierarchical cluster and spectral cluster to classify streamline effectively, which plays an important role in visualization.Cluster analysis is a method without knowing features of the data and the number of the features in advance, therefore it is unable to weigh the merits of the classification results. Based on Jacobi iteration, this paper translates optimal cluster process into a matrix diagonalization process, besides uses the corresponding coefficient to measure the cost of calculation, and operational results with the minimum cost are considered as optimal cluster results.For visualization after cluster, this paper proposes two different display modes to simplify streamlines including the centerline display and Bundle cluster display. These two modes effectively reduce the number of streamlines while retain the main characteristic of the flow field. Illumination effect is provided to flow lines using OpenGL line texture rendering.Finally, the implement of the proposed method as a plugin in Paraview achieves the combination of cluster analysis and visualization in user interaction, and obtains good analyze results.