Gradient-based classification and representation of features from volume data

The extraction and representation of information from volume data are important research avenues in computer-based visualization. The interpretation of three- or multi-dimensional data from various scanning devices is important to medical imaging, diagnosis and treatment, reliability and sustainability analyses in various industrial branches, and, in more general terms, information visualization. In this paper, we present several approaches for the classification and representation of relevant information from volume data sets. The techniques are based on the gradient vector, a property directly derived from the original volume data. We show how this property can be computed and subsequently used for classification through gradient-based one- and multi-dimensional transfer functions, as well as for the enhancement of surface features. The described techniques are illustrated through images generated using our volume rendering framework, from Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) data sets. The resulting images show how gradient-based techniques are suited for improved volume classification and the better extraction of meaningful information.