Images, particularly photomicrographs, provide qualitative information about surfaces of foods and cells. Methods to analyze the texture of images such as fractional Brownian motion (FBMM), box counting (BCM), and fractal dimension (FD) estimation from frequency domain (FDM), were used to numerically describe the surfaces of foods and the microstructure of potato cells. A FD was calculated for each image using the power-law scaling for self-similar fractals. The surface of analyzed foods had FD varying from 2.22 for chocolate to 2.44 for pumpkin shell. As reference, the FD of sandpaper having increasing grain size or roughness varied from 2.37 to 2.65. FD was also useful to numerically describe microstructural changes with time of an isolated potato cell during heating in oil and of the surface of chocolate undergoing crystallization of fats (blooming). (C) 2002 Elsevier Science Ltd. All rights reserved.