Research On Multivariate Volume Data Visualization Based On Parallel Coordinate Prime Dimension

Recently, with the development of computational fluid dynamics, medical imaging, meteorological simulation and other science fields, the complexity of scientific data is growing rapidly, and how to show the high dimensional data become one of the major issues in scientific visualization. As a branch of high dimensional data, the frequency of multivariate volume data has been higher and higher. The analysis and representation of multiple attributes and the relationships among multivariate volume data can help users understand the complex data characteristics quickly. Therefore, multivariate volume data visualization as well as intervention and guiding in the interactive process has great significance. Due to the multiple attributes, multivariate volume data is difficult to analyze and express. Although scholars has introduced parallel coordinates as an assistant to show data, multivariate volume data visualization still exists some problems such as not-flexibility design method and tedious regulation. Therefore, this paper mainly carried out the following work.Firstly, a detailed analysis of the current multivariate volume data research direction has been carried out. In view of the operation complexity and low interaction in the process of multivariate volume data rendering, the concept of Parallel coordinate Prime Dimension(PPD) is introduced which combines the parallel coordinate’s data analysis function and the interactive characteristics of transfer function curves.Secondly, an improved intensity-gradient histograms has been proposed for feature extraction of the PPD. The interface of transfer function uses the feature as background guiding users to design the transfer function.Thirdly, an interactive parallel coordinates method which allows adaptive range selection is proposed based on the traditional parallel coordinates. The method map the spatial position of sampling points which satisfies the range of each dimension into PPD, extract the data range and generate transfer function curves which change along with the user interaction in real time. Therefore, the problem that traditional parallel coordinate can not set the value of each data’s opacity and color can be solved.Finally, the interactive linkage design ideas is added into the volume data visualization pipeline. An frame of multivariate volume data interactive visualization has been proposed and realized which uses the GPU ray-casting algorithm in real-time rendering. And then the proposed method is compared and analyzed to prove the effectiveness of the method.