Development of a multiscale interactive visualization algorithm for gigapixel video streams

Abstract

The paper presents a number of algorithms for the interactive visualization of events with varying degrees of details over time and space in large streams and data files. The basis of the idea is the partitioning of the gigapixel video streams into multiscale videopyramids consisting of the video streams of lower resolution and allowing to broadcast gigapixel data visualizations to thin clients, such as Web browsers. In order to solve this problem, new algorithms for temporal and spatial synchronization of multiple video streams in modern web browsers with support for HTML5 are developed. A series of experiments on the load distribution on multiprocessor systems for encoding videopyramids were performed. Some estimation of the optimal gigapixel video streams partitioning for parallel data transfer over the network and simultaneous imaging in modern web browsers are obtained. The developed software based on these algorithms was used to create the visualization of Venus flight (June 5, 2012) filmed by the U.S. Satellite Solar Dynamics Observatory (SDO). As a result, a multiscale videopyramid available for viewing in a web browser with the ability to zoom in and to change the view area is obtained.