The analysis of sedimentary rock microscopy images allows the characterization of grain morphology in sedimentary petrography. Properties such as grain shape, size, pore tortuosity, and others constitute the basis for such analysis. The estimation of those properties demands the segmentation of the grains (or pores) from the background followed by the mathematical analysis of the segmented objects, a challenging process due to the number of objects present in the scene and to difficulties such as low contrast and the 3D nature of the data. We have developed graph-based image processing methods for the segmentation and analysis of sedimentary rock microscopy images using Optimum-Path Forest operators. The developed techniques have achieved results superior to state-of-the-art commercial softwares used in the petroleum industry.