Feature-preserving reduction of industrial volume data using gray level co-occurrence matrix texture analysis and mass-spring model

摘要

We propose an innovative method that reduces the size of three-dimensional (3-D) volume data while preserving important features in the data. Our method quantifies the importance of features in the 3-D data based on gray level co-occurrence matrix texture analysis and represents the volume data using a simple mass-spring model. According to the measured importance value, blocks containing important features expand while other blocks shrink. After deformation, small features are exaggerated on deformed volume space, and more likely to survive during the uniform volume reduction. Experimental results showed that our method well preserved the small features of the original volume data during the reduction without any artifact compared with the previous methods. Although an additional inverse deformation process was required for the rendering of the deformed volume data, the rendering speed of the deformed volume data was much faster than that of the original volume data.