A two-step method for preprocessing volume data

摘要

A method to preprocess volume data for rendering and analyzing is proposed here. We segment the data by constructing a threshold super-surface, which is sensitive to noise. So first we reduce the noise using level sets. The data is decomposed into upper level sets and lower level sets, and the small connected components of each level set are removed. In the second step, the segmentation problem is modeled as a partial difference equation (PDE). The super-surface is obtained by solving this PDE. Finally, the performance of our method is demonstrated with an experiment.