Texture is one of the most important features for classifying and recognizing objects and scenes, and can be characterized by local variations in pixel values that are repeated regularly or randomly throughout the object or image. Several methods for classifying images using texture features have been proposed in the literature. However, there is no generic method or formal approach that is useful for a wide variety of images. Texture refers to a visual pattern that has some homogeneous properties that are not simply the result of color or intensity. Unlike other characteristics (brightness, color), texture cannot be defined on a single pixel, but rather across a region or set of pixels. The three main approaches used in image classification to describe textures are statistical, structural and spectral, which are presented in this paper.