A fuzzy logic approach to edge detection in HREM images of III–V crystals

Abstract

Edge detection algorithms have been developed to partition an image field into subfields representing regions with different properties. Edges are defined by relatively notable and distinguishable image changes. Among advanced theories, fuzzy logic (FL) is highly suited to detect such edges.

High-resolution electron micrographs (HREM) of crystalline specimens reveal the nature of crystals. For studying interdiffusion phenomena in layered structures of III–V compounds, image variations, described by the term similarity, are interpreted. Contrary to natural images, a periodic array of subunits (crystal cells) dominates the image patterns in HREM applications. The crystallograhic cells are represented by values of similarity, related to known templates of a III–V compounds by a difference measure of similarity.

The idea of this fuzzy edge detection algorithm is to analyze the transition taking place between the two neighboring sides of the edge. The limited number of cells requires fitted masks of neighboring cells, and also serves as input to the set of dedicated fuzzy rules. Applying two triangular membership functions, the location of the most significant composition steps can be determined. After analyzing simulated HREM patterns, the fuzzy logic development tool has been employed to obtain the edges on experimental micrographs of MBE-grown Al/GaAs.