GPU-Accelerated Focus and Context based Visualization of Volume Data

Abstract

Volumetric datasets are increasing at incredible rates in terms of number and size which generate number of challenges like maintaining performance and extracting meaningful information. The large size of data set bounds the amount of data that can be looked simultaneously, requiring management of overall context of the data as the user looks at or zooms into a peculiar area of interest. Volume rendering is a technique used to display a 2D projection of 3D volume dataset. Ray Casting and Isosurface are commonly used algorithms for volume rendering. Isosurface rendering falls under non-direct volume (surface) rendering whereas Ray Casting is a direct volume rendering technique. The Visualization Toolkit (VTK) is an open source library which supports a wide variety of visualization algorithms for volume rendering of medical datasets. For volume rendering, both Isosurface as well as Ray Casting techniques have been used. Focus and context based visualization is an approach which is used for viewing a part of interest without loosing the rest of the context. Distorton and non-distortion approaches are used to achieve the focus effect. We have implemented a focus and context framework that employs zoom operation and various standards visualizing approaches to magnify the features of concern, while compressing (at low resolution) the remaining volume parts without clipping them away entirely. To achieve interacive visualization, GPU based volume visualization technique has been used in providing a solution to above problem (i.e. the real time requirement)