Automated image analysis for applications in reservoir characterization

The paper presents a novel software toolkit called PetroGraFX that is being created to provide fast and accurate automated assessments of petrographic images. In particular, the paper describes the image analysis algorithms that have been developed to derive porosity from the visible pore space and to segment and measure individual quartz grains in color images of thin section slides. A region growing scheme in color space is described for accurate segmentation of the pore space, as is a modified watershed transformation developed for reliable delineation of the boundaries of adjoining quartz grains.