Parallel Topological Analysis Of Two-dimensional Flow Field Based On Physical Feature

With the rapid development of science and technology, scientific visualization has been increasingly involved in a wide area whose objects become increasingly complex, the amount of data to be processed becomes larger and larger and real-time requirement is increasing. Vector field visualization as the most challenging important part of scientific visualization, the traditional planar vector field visualization methods have been unable to meet requirement. In this case, the visualization of feature as a new visualization technique has been proposed and has gained rapid development. This "selective" visualization can filter out irrelevant data, greatly reducing the amount of information to be processed, while guarantee the accuracy of amount. Flow field topology analysis is an important method of feature visualization.The advantage of parallel cost and the application meeting the requirement of real-time performance promote the development of parallel computing. Parallel processing techniques provides an important platform for real-time visualization of massive data. In this paper, using widely accepted MPI technology based on information and domain decomposition algorithm, a large complex problem is divided into small problems who can using parallel computing.It makes use of cluster using master-slave parallel programming techniques to improve the speed of operation to meet real-time requirement.The combination of topology analysis and parallel technology provides a new way to solve large-scale flow field data in real-time visualization. The algorithm consists of five parts:sampling data interpolation encryption, localization and classification of critical points, feature calculation, based on user setting thresholds for the feature regions of critical point, according to the user's selection to paint feature topology and flow diagram. Sampling data interpolation encryption algorithm, the calculation of the critical point position and classification, calculation of curl and divergence's eigenvalue and generating the flow lines use parallel computing. The existing traditional topological structure methods of the flow field are based on geometric feature extraction. It is difficult to meet remote areas visualization in real-time. This paper uses curl and divergence filter concept, with the integration method to calculate the physical characteristics of flow field.It finds the critical point on the interpolation encryption data and uses Jacobian determinant to classify the critical point and store.According to the displaying graphics' type of the user's selection and the setting threshold to find the feature regions of the critical point on the obtained characteristics image data. The flow line use the fourth order Runge-Kutta method to obtain. Flow line density control is based on the type of Feature Area's critical points to control. Different critical point has different method of selecting seed point. Different number of the selected seed point and location has different flow lines density. Based on the demonstration of the relationship between critical theory and physical Features, it uses parallel algorithm making combine of critical theory and flow field of two-dimensional visualization method based on physical feature, abandoning complexity based on geometric computation and the fault of difficulty of extracting saddle Feature data based on the physical Features.This paper puts forward a fast clear method of showing the Features of flow visualization. This algorithm can extract a flow feature image with clear structure topology and complete feature. The algorithm can effectively realize the selective visual performance of planar discrete flow field by maintaining the physical characteristics unchanged and adjusting the parameters of selection, and then describe the overall topology structure quickly and accurately from the physical and Macro point of view, gaining the purpose of topological simplification and description compression.Experimental results show that this method can accurately and quickly find the critical function area.It is a new method extracting topological structure of the flow field.