Research On The Interactive Integrated Visualization Of Multi-feature Datasets

With the progress of science and technology such as remote sensing, space detection and geographic information etc, the data size obtained or generated from the production and life is increasing. In order to find out internal relations in the massive multivariate data from various fields and make up the limitation of human cognitive ability, multivariate data visualization technology is becoming a research hotspot. However, present researches about multivariate data visualization are maily concentrated on two aspects, one is the alternation display of visualization results and the other is the superposition presentation of different visual codings. They are far from enough to mine the interrelation and hidden information from datas.For deeply mining hidden informations from multivariate datas, an interactive integrated visualization method based on multivariate datas is proposed. And the specific research work is as follows.To begin with, a view-dependent multi-resolution flow visualization algorithm is presented for the low efficiency problem of large-scale linear flow visualization. Introduce the quadtree structure in the seed point selection process and adjust the quadtree according to the viewpoint position in real time to realize the diversified distribution level of seed points and streamlines. The geometry shader is employed to produce streamlines and use the GPU powerful parallel processing ability to conduct the integral computation of streamlines in real time, in order to improve the streamline rendering efficiency.Secondly, faced with the single data analysis mean of two dimensional scalar field, an interactive visual analysis method is proposed. The quantitative analysis and feedback display of the statistical information of interested areas are realized with the Tee Chart control and double slider after the data visualization based on 2.5 dimension.And then, propose an interactive integrated visualization method of multivariate data based on data similarity. The parallel coordinate is used as the dataset selection tool and an iterative alpha color blending method is designed to realize the deeper mining and integrated presentation of multi-feature datasets; Introduce the concept of data similarity and make a formal definition, based on which to deal with colors of band overlaps to further refine the information presentation level.Finally, the visualization program framework is designed and implemented, and the proposed methods are verified with experiments and analyses to prove the effectiveness.