3D gray level co-occurrence matrices for volumetric texture classification

In this work, we stress the classification of volumetric textured images, i. e. three-dimensional (3D)texture. 2D classification is a field frequently treated in our days, passing through the statistical methods, the spectral transformations towards the parametric methods. The approach containing the matrix of two-dimensional and three-dimensional co-occurrence preoccupies many researchers in the field of characterization and classification of volumetric textures. The methods of forming 3D images become increasingly widespread as they provide the possibility of examining texture like volumetric phenomenon. Texture classification derived from volumetric data will have a better distinctive power than 2D texture derived from the data of the section. An experimental study was then undertaken in which the results for the devices of texture derived from 2D are compared with those obtained results using matrices of co-occurrence for volumetric data. The preliminary experimental results indicate that the devices of texture have a better distinctive power than 2D texture derived from the data of the section. This device is more robust in presence of noise than the method containing 2D texture derived from the data of the section.