This thesis presents a method for digitization of the two-dimensional shape of granular particles by means of photo microscopy and image processing techniques implemented using a software package from Media Cybernetics, Inc: Image-Pro Plus 5.1 and the add-ins Scope-Pro 5.0, SharpStack 5.0 and 3D Constructor 5.0. With the use of these tools, it was possible to implement an efficient semi-automated routine that allows the digitization of large numbers of two-dimensional silhouettes of particles in minimum time, without endangering the quality and reliability of the shapes obtained. Different sample preparation techniques, illumination systems, deconvolution algorithms, mathematical functions, filtering techniques and programming commands are brought into play in order to transform the shape of the two-dimensional projection of particles (captured as a set of successive images acquired at different planes of focus) into a binary format (black and white). At the same time, measurements and statistical information such as grain size distribution can be analyzed from the shapes obtained for a particular granular soil. This information also includes but it is not limited to perimeter, area, diameter (minimum, maximum and mean), caliper (longest, smallest and mean), roundness, aspect ratio and fractal dimension. Results are presented for several sands collected from different places around the world. In addition, some alternatives for three-dimensional shape reconstruction such as X-ray nano tomography and serial sectioning are discussed.