Texture categorization using statistical and spectral features

Texture is an important spatial feature, used for identifying objects or regions of interest in an image. Depending upon the size and spatial arrangement of texture elements i.e., texels, the texture images can be grouped under the following categories: micro, macro, periodic, aperiodic, fine, coarse, regular, random, stochastic, non-stochastic, deterministic, non-deterministic, strong and weak textures. It depends on the applications and the type of features that has to be extracted from the texture. This process is rarely an easy task. In this paper a statistical method is employed to discriminate fine textures from coarse textures and a spectral measure is used to discriminate periodic textures from aperiodic textures. The experimental evaluation of the proposed method is done using the Brodatz texture database.