Automatic grain size determination in microstructures using image processing

In microstructure analysis, the grain size determination is an important task. However, it takes a long time when it is made manually. Nowadays, automatic techniques for grain size determination have been implemented. Although these automatic techniques are documented on the ASTM standards, one of the main drawbacks is related to the quality of the digital images. There are many factors that can affect the quality of an image, such as the illumination conditions, causing noise, low contrast, bad defined boundaries, among others. When a metallographic image presents these characteristics, individual grain identification becomes a difficult task. The present work is focused on a novel methodology that enables the clear definition of the grain and the boundary regions for an accurate automatic grain size determination through some efficient image processing techniques. (C) 2012 Elsevier Ltd. All rights reserved.