Visualization Techniques for the Comparative Analysis of Multivariate Data of Fiber Reinforced Polymers

Abstract

The comparative analysis of data from multiple candidate material systems is essential for the improvement and development of advanced composite materials. Since the characteristics (e.g., length, orientation) of individual objects (e.g., pores, fibers) are crucial for the properties and the behavior of materials, the exploration of these is of utmost importance. Currently, the comparison of these multivariate, numerical data requires significant mental efforts, since the experts have to perform a sequential comparison of many slightly different charts. This work aims to simplify the workflow for comparing many datasets by providing experts with an interactive visualization framework. By combining different representations at various levels of detail, it assists users to get a holistic view on the similarities of the objects and their characteristics. An overview visualization allows to determine the similarity of the objects of several samples, additional detail visualizations give the user detailed information about groups of similar structures. We demonstrate the applicability of our tool on two use cases, a fiber material subjected to an in situ test, and a comparative study of multiple pore datasets.