A graph based segmentation strategy for baggage scanner images

Abstract

Image processing and image analysis are often required in real life scenarios. Segmentation is one of the key concepts used and for which has not yet found a general solution that can be applied for every stage. In this paper a graph based segmentation strategy is proposed aimed to images resulting from baggage scanners used by the General Customs of the Republic of Cuba. This strategy is a bottom up one that combines the Minimum Spanning Tree and the mixing regions approaches. It defines a new standard for the two-component merge that considers both global and local features of the image. The numerical experiments show the effectiveness of the strategy for custom scanner images and how it can be easily adapted to other image types such as natural images. © 2014 Springer International Publishing.