M.T. Suzuki (Japan)
Fractal Dimension, Hurst Analysis, Solid Texture, Volumetric Data, Voxel, Pattern Feature
Recent advancements in 3D computer graphics technology have made possible the visualization of 3D volumetric data. A large number of volumetric data is available and used for various scientific fields. It is important that software can search and classify the volumetric data in a systematic manner, especially if there is a large number of volumetric data to be processed. Various fractal based methods such as the Hurst analysis and box counting techniques have been used for extraction of pattern features from 2D image data. However, pattern feature extraction techniques for 3D volumetric data have not been sufficiently investigated. Therefore, in this paper, popular 2D circular Hurst operators are extended to 3D spherical operators for dealing with 3D volumetric data. The three dimensionally extended spherical Hurst operators can analyze pattern features directly from 3D volumetric data. Also, the Hurst Orientation Transform (HOT) method is modified for estimating the directionality of 3D volumetric data. An experimental system is implemented to test the Hurst operators. Preliminary experiments show that the system successfully extracts shape features from 3D volumetric data.
Important Links:
Go Back
