The periodic nature of the fabric images allows using fast Fourier transform techniques in image processing for its characterization. Due to the repetition of patterns inside the images of textile, is possible to find a form relatively easy of description in their energy spectrum. A recent work outlines a group of geometric descriptors for the Fourier spectrum of the images; looking for this efficiency to rotation, translation and scale change invariance [1]. These descriptors showed to be very effective to represent a textile fabric and can be used to characterize the quasi periodic textures in real time and in situ non destructive techniques. Samples of textile textures are tested to this technique of parametric representation with the purpose of analyzing their robustness and reproducibility. Finally, a set of textile fabrics is subjected to this model with the aim of evaluating the possibilities to carry out classification, verification and pattern recognition. PACS: 42.30.Kq