Image classification and segmentation for densely packed aggregates

Abstract

This paper presents a methodology for delineating densely packed aggregate particles based on aggregate image classification. There is no earlier work on segmentation of aggregate particles has exploited these two building blocks for making robust object delineation. The proposed method has been tested experimentally for different kinds of densely packed aggregate images, which are difficult to detect by a normal edge detector. As tested, the studied algorithm can be applied into other applications too. © Springer-Verlag Berlin Heidelberg 2007.