Flotation froth can be considered as a complex surface consisting of geometric elements characterized by different colors, surface reflectance, dimensions and "texture" according to the process behavior. Froth surface characteristics can thus be considered an indicator of the process. This paper presents the algorithms and the procedures developed by the authors to set up a digital flotation surface modeling (DSFM) strategy to utilize the identification of those elements of a feature vector (consisting also of other froth bubble numerical parameters such as color, morphological and morphometrical attributes) to classify froths with respect to conventional process characteristics such as recovery and grade. (C) 2002 Elsevier Science B.V. All rights reserved.