Texture analysis is important in image processing in many application, medical image analysis, material science, satellite imagery and so on. No formal definition of texture exists; basically this property describes characteristics such as smoothness, coarseness and regularity, i.e. relations between adjacent pixels. In texture analysis one of the most difficult aspect is to define a set of features that adequately describes the characteristics of the texture. In this presentation the problem of identifying different texture is faced by applying to a suitable transformation of the data the discrete level set formulation of a modified Mumford and Shah energy functional. The proposed texture segmentation procedure has been applied to real images; in particular X-SAR data and metallographic data. The characteristics of these data require ad hoc image preprocessing and suitable choices of the texture properties to be considered for the segmentation.