Feature-based Sparse Noise Texture Sampling For3D Vector Field Visualisation

The flow visualization method is an important method for visualization oflarge-scale scientific computing and data analysis occupies a very important position.The flow field characteristics are typical and universal, but also has a complex structuralfeatures. Characteristics of visualization methods for the universality of structuralfeatures of the flow field, by extracting the main features of the flow field, so thatpeople can ignore not interested in the area, reducing the amount of data at the sametime, the extracted information timely and accurate. Existing versatility and poorscalability problem.（1）Fuzzy feature extraction method based on improved c-means algorithm tocalculate the membership of the flow field characteristics and important value.（2）In this paper, the space-filling curve dimensionality reduction method appliedto the three-dimensional vector field generated adaptive distribution of one-dimensionalpoint sequence generated by the fuzzy feature extraction feature regional map to controlthe distribution of points, effective solution points uneven distributionsparse degree ofcontrol to highlight a key regional characteristics.And put forward aweight-based ofnoise generation method.（3）With the comprehensive use of the research,this papr design the featurebasedsparse noise texture sampling for3D vector field visualization methods. The clusteranalysis calculated values to control the distribution of one-dimensional low point ofdifference sequence generated by the space-filling curve dimensionality reduction,generate densely control noise texture, texture generated by3DLIC, then accelerated bythe CUDA and GLSLimplementation. The experimental results show that this methodcan effectively relieve blocked and confusion of the complex three-dimensional flowfield visualization.