A methodology of textural analyses based on image analysis is proposed and tested based on study of fault rock samples from the Tatra Mts., Poland. The procedure encompasses: (1) SEM-BSE imagery of thin sections; (2) image classification using the maximum likelihood method, performed with GIS software; (3) statistical analysis and fractal dimension (self-similarity) analysis. The results of this method are comparable to those obtained with methods involving specialized software. The proposed analytical procedure particularly improves qualitative observations with quantitative data on grain shape and size distribution. The potential of the method is shown, as an auxiliary tool in determining the nature of deformation processes: the role of high-temperature dynamic recrystallization processes is recorded using grain shape indicators, whilst the switch from ductile to brittle conditions is reflected by the grain size distribution pattern.