Exploiting the Hessian matrix for content-based retrieval of volume-data features

We propose an algorithm for content-based retrieval of representative subsets of volume data. Our technique is based on thresholding of the eigenvalues of the Hessian matrix. We compare our approach to feature detection based on the gradient magnitude and observe that our method allows the representation of volumes by a smaller amount of voxels. Practical applications of our method include fast volume display due to object-space oriented techniques, generation of preview data sets for web-based repositories, and the related progressive visualization over the network. For these applications, the size of the representative subset can be estimated automatically with respect to the bottleneck of the visualization system or a network bandwidth.