Recognition of natural and artificial materials is an essential part of image understanding and computer vision. Unfortunately, the appearance of materials in real scenes is highly dependent on illumination conditions and camera position. We proposed invariant textural features, which are based on Markovian modelling of texture and which are invariant to illumination colour, local intensity variation and robust to illumination direction changes. Rotation invariance was added as well. Presented applications will include image segmentation, decoration industry, computer graphics, and medical application.