RESUMO
The depth-of-field mainly affects the image quality either in scanning electron microscopy (SEM) or conventional light microscopy. The limited depth-of-field handicap of microscopy imaging can be used for obtaining "optically sectioned" specimens by moving the object along the optical axis. In this study, multiple images corresponding to different object planes were taken in order to overcome limited depth-of-field on conventional light microscope and SEM, estimation of an elevation surface and 3D reconstruction of different type volcanic ash surfaces. We used extended depth-of-field, a fusion algorithm that combines those images into one single sharp composite. Because of larger depth-of-field, we got higher-quality results even with image stacks taken by SEM with a fixed aperture in variable pressure mode. We calculated roughness descriptors, quadtree decomposition and greylevel standard deviation (sGL) and analyzed the shape of polar plots based on gradient analysis of constructed depth-maps. Furthermore, we calculated fractal dimensions of surfaces. Correlation analysis was performed to measure how these quantitative variables are related with different type ash surfaces. Roughness descriptors, quadtree decomposition, sGL and fractal dimension discriminate different types of volcanic ash surfaces.
