Optimum image processing for morphological study of granular films

Abstract
The binarization of transmission electron micrographs of thin granular films is an essential step for the statistical study of the film morphology in the framework of scaling laws and fractal theory. An image processing is developed, leading to an unambiguous optimum binarization threshold, without any loss of information. This treatment, efficient even on films with unimodal gray level histograms, is tested on different granular metallic or metal-dielectric films