This paper presents a new approach for analyzing the microstructure of SiCp-reinforced aluminum matrix composites from digital images. Various samples of aluminum matrix composite were fabricated by hot pressing the powder mixtures with certain volume and size combinations of pure At and SiC particles. Microstructures of the samples were analyzed by computer-based image processing methods. Since the conventional methods are not suitable for separating phases of such complex microstructures, some new algorithms have been developed for the improved recognition of the particles in the metal matrix composites. One of the most determining attribute of composites structure is anisotropy, but the measurement of this parameter is very difficult and slow. For this reason, the anisotropy of metal matrix composites was investigated by the newly developed algorithm which has an optimized speed.