Transfer function design for volume rendering using K-Means clustering

This paper presents a new transfer function design approach using boundary feature extraction in medical volume data. The proposed method applies K-Means clustering to group voxels of each similar region. Then, boundaries in volume data are extracted by cluster centroids. Opacity transfer function assigns opacity to the boundaries, and the opacity values are determined by gradient magnitude. This transfer function effectively shows boundaries which are important information in volume data. Also, this approach offers ease of use to users.