Metabolic syndrome and its relation with the distribution of visceral adipose tissue (VAT), has aroused great interest in the development of image processing tools for its automatic quantification. A new technique for automatic segmentation of visceral and subcutaneous adipose tissue in computerized axial tomography (CT) images, by means of assessing local anatomical invariants is presented. The assessing made by an expert of the segmented images, showed low error in the quantification of VAT tissue and a high tolerance to different anatomical configurations and other kinds of image problems.