Visualization and interpretation of 3-D scientific data sets
Lambertus Hesselink, Jim Helman, Kevin Wu, Ronald J. Adrian · 1988
Recent advances in (optical) diagnostic and numerical techniques are providing data sets of ever increasing complexity and size. The need for interpretation and visualization of these data is becoming commensurately more urgent. Present approaches in digital image processing and pattern recognition are mainly aimed at 2-D imagery. Extensions to higher order dimensions need to be carefully considered and implementation of new approaches is not a simple task. We present several approaches currently under development at Stanford University. An implementation of Clarks’s algorithm is used to visualize a stack of cross sectional images obtained by illuminating a transparent object with a sheet of laser light. The object is either convected through the stationary sheet - and by using a Galilean transformation a time sequence of images is transformed into a spatially stacked array -, or the sheet is rapidly translated normal to its plane to scan the three-dimensional object. The resulting data set is rendered visible using the SURF graphics display software developed at Stanford. This package allows interactive removal of obscuring outer surfaces to reveal interior structure. Interpretation of three-dimensional data sets is a complicated task, and we will present one approach for a fluid flow involving image decomposition into elementary structures such as nodes, saddles and foci. Graph theory is used to represent the elementary structures and their connections.