Detecting texture periodicity from the co-occurrence matrix
Pattern Recognition Letters
The amplitude varying rate statistical approach for texture classification
Pattern Recognition Letters
Texture Discrimination Based Upon an Assumed Stochastic Texture Model
IEEE Transactions on Pattern Analysis and Machine Intelligence
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.10 |
Statistical approaches to texture use features that describe the distribution of intensities or local features but ignore their spatial interdependence. Structural approaches concentrate on the spatial interaction of elementary regions, local features, or intensities. In most of the structural methods, it is assumed that the examined texture pattern is more or less regular. Such methods are inappropriate for dealing with the gradual transition from regularity to randomness which is typical for the spectrum of natural textures. In this paper, an attempt is made to bridge the gap between statistical and structural texture analyzers. The contrast curve of a texture pattern is defined as the first moment of a co-occurrence matrix plotted as a function of the intersample distance. The curve is used as a texture descriptor. An algorithm is presented that generates the contrast curves of both regular and random textures in the framework of a simple, uniform model. The model has four parameters that are easy to interpret. The generated curves are fitted to the experimental ones, and certain combinations of the parameters of the best fitting curves are introduced to measure texture regularity.