COMPARING THREE IMAGE PROCESSING ALGORITHMS TO ESTIMATE THE GRAIN-SIZE DISTRIBUTION OF POROUS ROCKS FROM BINARY 2D IMAGES AND SENSITIVITY ANALYSIS OF THE GRAIN OVERLAPPING DEGREE

The grain-size distribution (GSD) of porous rocks is important in order to better understand their hydrodynamic behavior. Clear and precise GSD data can be used to computationally reconstruct rock structure for further analysis. In this study, three main algorithms for image analysis have been examined to estimate the GSD of clastic rocks. The main challenge in GSD determination from images is in detecting overlapping grains and measuring their size separately. In this study, three previously developed image processing algorithms are implemented on two-dimensional (2D) binary images of rocks in order to compare the obtained GSD from each of the methods, i.e., the mean intercept length method, erosion and dilation method, and watershed segmentation method. Grains can be visually overlapped for several geological reasons, such as severe compaction, diagenesis, or cementation. When the overlapping degree of grains is severely increased, the image processing algorithms fail to detect the true grain size. A sensitivity analysis has been done on several synthetic random packed rock samples to evaluate the field of applicability and the accuracy of the aforementioned methods via different grain overlapping degrees. Finally, a comparison between the discussed methods is presented, which helps researchers choose the appropriate algorithm that fits their rock samples.