Full text
| |
| Author(s): |
Christian Jorge Daniel Wong Cruz
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | São Carlos. |
| Institution: | Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) |
| Defense date: | 2012-09-10 |
| Examining board members: |
Rosane Minghim;
João Paulo Gois;
Haim Levkowitz
|
| Advisor: | Rosane Minghim |
| Abstract | |
The area of volume visualization encompasses a set of techniques used for representation, manipulation and display of data associated with a region of a volume, thus enabling the exploration and understanding of the interior of three-dimensional objects. However, some limitations are still encountered in this area. For example, the simultaneous exploration of more than one value in multivariate volumetric datasets. Beyond this challenge, another issue of great interest to the scientific community is the exploration of time-varying volumes. The complexity of this case lies in treatment or processing of a very large amount of data, seeking to discover properties, structures, or characteristics that may vary in time. This work proposes techniques and approaches, based on the concept of multidimensional projections, in order to support multivariate volumetric analysis of time varying data sets. The first technique proposed, called Fastmap*, enables the projection of high dimensional streaming data. The second technique presented, called Recursive Laplacian-based Neiboorhood Projection, allows the projection of data sets based on neighborhoods, maintaining the temporal coherence in the projected data, besides having the ability to project highdimensional spaces with a low level of stress. Also, we propose an approach for the analysis of specific attributes, referred to as Scatter Projection, which facilitates the exploration focused on a specific attribute and on the similarity between them. Finally, we propose an approach based on reprojection of groups using feature selection techniques for better identification of internal structures of the volume. Thus, this study contributes towards surveying and discussing limitations of the area, and then seeks ways of solving these issues, proposing techniques and approaches that enable the exploration of multidimensional volumetric time varying data sets, maintaining the temporal coherence (AU) | |
| FAPESP's process: | 09/13438-9 - Exploring Time-Varying Multidimensional Volumetric Datasets using Projections |
| Grantee: | Christian Jorge Daniel Wong Cruz |
| Support Opportunities: | Scholarships in Brazil - Master |