Seminario de Investigación (17/01/2011)

Fecha, hora y lugar: Lunes 17 de Enero, 11:00 horas, sala de reuniones.

Expositor: Rodolfo Allendes.

Título de la charla: "Visualization of Uncertainty in Scientific Data".

Descripción:

Uncertainty is an unavoidable fact in science, whether we are speaking of measurements done on the field, or computing simulations, there will always be, albeit small, a certain amount of uncertainty in our data. And however this undeniable fact, visualization techniques more often than not avoid the representation of uncertainty, thus leading to visualizations that can be misleading and misinterpreted, and in the worst cases, to equivocated conclusions and results.

The single, perhaps most used type of uncertainty visualization is the use of error bars in one-dimensional graphs, common in science and engineering. In this thesis, I we will address the definition and implementation of new techniques for the visualization of uncertain
scientific data, with an emphasis in the visualization of two and three dimensional scalar datasets, and two dimensional flow fields.

For the visualization of scalar data, we start by developing the idea of uncertain contours in two dimensions, and together with their development, we also present a series of methods for their visualization. Later, these methods are extended, as to be also applicable for the visualization of three dimensional scalar data.

In turn, for the visualization of uncertain flow fields, we develope a series of modifications, either to specific stages of the algorithm, or by the addition of new visual properties, to two traditional texture-based methods, LIC and IBFV.

Finally, we provide through evaluation of our scalar and flow visualization techniques, concluding, for the second, that our methods are suitable for the encoding of uncertainty, and that users can map them back to their original numerical values.