Co-occurrence Matrices for Volumetric Data

A.S. Kurani, D.-H. Xu, J. Furst, and D.S. Raicu (USA)

Keywords

Imaging and image processing, cooccurrence matrices, volumetric data, volumetric texture

Abstract

In this paper, we investigate a new approach to the co occurrence matrix currently used to extract textural features: co-occurrence matrices for volumetric data. While traditional texture metrics have concentrated on 2D texture, 3D imaging modalities are becoming more and more prevalent, providing the possibility of examining texture as a volumetric phenomenon. Just as computer graphics have used 3D textures as a more realistic alternative to 2D texture mapping, we expect that texture derived from volumetric data will have better discriminating power than 2D texture derived from slice data. An experimental study has been conducted in which the results for textural features derived from 2D are compared to those results derived from using co occurrence matrices for volumetric data. Our preliminary experimental results indicate that the volumetric texture features have better discriminating power than 2D texture derived from slice data.

Important Links:



Go Back