Research Of Parallel Visualization For Vector Field Based On Fast-LIC

It is a popular direction that the visualization of vector field under numerical tank in the field of scientific computing visualization.The vector field data generated by the marine scientific research experiment can be transformed into the observed graphs and images,which makes it possible to explain the scientific phenomena and objective laws in the ocean flow field through the image.It has broad and far-reaching significance.Nowadays,the vector field visualization technique can describe the direction continuity feature of the ocean flow field,and can show more details of the flow field characteristics.At the same time,it can generate multi-resolution and continuous fine images,and has a broad development space in the visual field.However,with the increasing of the vector field,the data set is on the rise,which will cause the visualization program to run for a long time.Thus it affects the visualization efficiency.In the paper,it aims to shorten the visual computing time,a parallel visualization algorithm based on Fast-LIC is proposed for the problem that the visualization computation time is too long for large-scale flow field data.In the paper,the main work is as follows:(1)Firstly,it is studied that the parallel visualization theory of vector field,and it includes the existing vector field visualization technology,and parallel computing technology,including parallel algorithm design process and performance evaluation index.At last,this paper introduces the parallel programming framework used in this paper.(2)Secondly,it is designed that a vector field visualization algorithm.And it is also designed in detail that the image generation process,including the white noise image generation,the local streamline computation of the vector field,the calculation process of the convolution kernel function,etc.A serial algorithm based on Fast-LIC is designed based on the correlation of streamline pixels.(3)Finally,a parallel analysis of the Fast-LIC image generation algorithm is carried out by PCAM design.And this paper proposes a node adaptive dynamic task allocation strategy.In order to further reduce the time of calculating the grid pixels in the vector field,combined with the characteristics of vector field pixel correlation,proposed a grid mapping task execution strategy.In the paper,parallel visualization experiment is carried out based on Windows cluster environment,and then the program execution time of serial and parallel visualization algorithm are compared by simulating the high resolution vector field data of numerical tank.The experimental results show that the parallel algorithm significantly shortens the visual computing time,and the serial image and the parallel image effect.Finally,the speedup and parallel efficiency of the parallel algorithm are analyzed,and the effectiveness of the parallel algorithm is obtained.To sum up,this paper has great significance for the study of vector field visualization under numerical tank.