Using Visualization and Animation to Convey Motion in Experimental Data


Forward Forward

Abstract

Computer visualization is a valuable technique for understanding massive amounts of data. By providing faithful and efficient depiction of the underlying phenomena, computer visualization makes it possible for scientists to take advantage of faster parallel computers and advanced experimental techniques. This thesis uses static visualization and computer animation to provide graphical explanations for a pattern forming system produced from a combustion experiment. Like any other system, pattern forming systems must be properly formulated and modeled in order to be visualized. The original data is recorded on videotapes, preprocessed into individual raster frames and stored in a directory. Tracking is then conducted on the frames, and quantified information for individual objects is generated as the starting point of the visualizations. Two criteria are developed as a basis of data modeling: a minimum area invariant and the principal representation of image object.

The static visualizations model the data to take advantage of the ParaGraph visualization tool which is designed for monitoring the performance of parallel programs. The data is processed so that it has state information in the form of ParaGraph input format (PICL). Static visualizations are useful for summarizing state information over time or space, but they lack the sense of motion. Computer animation gives the impression of motion by adding time control to visualizations. The animation programs are based on the POLKA animation library. The static visualization and computer animation techniques are conducted on three modes of the flame system---hopping, rotating, and radial extinction. Each of these modes again has several submodes. The visualizations provide insights which were not available from direct viewing of the tape.