Real-Time Visualization of Large Data
Patric Ljung and Anders Ynnerman
Previous work in this area by the NVIS group has revolved around the visualization of the extremely large medical data sets whcih are becoming increasingly common as scanning technology develops. The techniques developed in this work exploit domain-specific knowledge, that of the radiologist and doctors involved in the diagnostic process, in determining how data reduction should be applied to the data under examination.
|In this way the visualization pipeline, which has only a limited bandwidth, can be made best use of, with only those parts of the data which will make a significant impact on the rendered image being loaded from disk and processed through the pipeline. These techniques have been found to result in images which, despite being based only a small percentage of the full data set, are visually indistinguishable from those based on the full data.|
While these techniques have been developed within the specific domain of medicine, they show great promise in other areas. Similar uses of domain knowledge in Engineering, Chemistry and Physics, amongst others, may provide high levels of data reduction and compression which will facilitate the examination of data from large measurement and simulation data.