New methods of image analysis are discussed and tested in the contribution by mean of computer experiment. All methods are designed for the study of multilevel images, where the information is stored in the grey scale of the image. The first approach is based on standard bilevel morphological methods and an attempt is made to extend their range of validity. In the second approach, the multilevel image is conversed into a set of bilevel images and these images are processed simultaneously. As the third approach giving the best results, a genuine multilevel image analysis based on gradient methods is suggested. All these techniques are implemented as computer programs and can be applied to the data derived from experiments with continuous metal films and surfaces. (C) 2003 Elsevier Science B.V. All rights reserved.