Description and classification of rock surfaces by means of laser profilometry and mathematical morphology

The evaluation of surface roughness is crucial to the hydrochemical and mechanical description of fractured rock masses. Surface roughness contains information on rock strength, deformability, permeability, etc. Recent years have witnessed a rapid development of new methods for measuring the surface of rock fracture using state-of-the-art technologies. Currently available measuring instruments, such as profilometers and confocal microscopes, provide information about hundreds of thousands of even millions measurement points which represent the investigated surface. The key problem, therefore, is to work out methods to adequately interpret such large packets of data. This study attempts a thorough analysis of this type of data using image processing and mathematical morphology methods. The paper presents the results received from morphological gradients, analyses of the results obtained from the watershed as well as the analyses of variograms. Furthermore, it proposes the application of morphological filtering for selecting the roughness component of a rock fracture. These results have been used in classifying the investigated rock. This classification was based on pattern recognition methods. By the definition of the 6D features space and the definition of learning sets, a successful classification of investigated rocks has been obtained, with up to ca. 95% correct recognitions.