A Smart Compression Scheme for GPU-Accelerated Volume Rendering of Time-Varying Data

The visualization of large-scale time-varying data can provide scientists a more in-depth understanding of the inherent physical phenomena behind the massive data. However, because of non-uniform data access speed and memory capacity bottlenecks, the interactive rendering ability for large-scale time-varying data is still a major challenge. Data compression can alleviate these two bottlenecks. But just simply applying the data compression strategy to the visualization pipeline, the interaction problem can not be effectively solved because a lot of redundant data still existed in volume data. In this paper, a smart compression scheme based on the information theory is present to accelerate large-scale time-varying volume rendering. A formula of entropy was proposed, which can be used to automatically calculate the data importance to help scientists analyze and extract feature from the massive data. Then lossy data compression and data transfer is directly operated on these feature data, the remaining non-critical data was discarded in the process, and GPU ray-casting volume render is used for fast rendering. The experiment results shown that our smart compression scheme can reduce the amount of data as much as possible while maintaining the characteristics of the data, and therefore greatly improved the time-varying volume rendering speed even when dealing with the large scale time-varying data.