Structural X-ray Image Segmentation for Threat Detection by Attribute Relational Graph Matching

This paper addresses part of the problem dealing with the automatic threat detection for accompanied baggage based on multi-energy X-ray imagery for station security. Segmentation is the first significant stage to extract interested objects in the images for detailed analysis and recognition at following stages. In order to obtain the integrated objects for subsequent analysis and recognition, we propose a structural segmentation method based on ARG matching. The proposed segmentation algorithms are a series of graph-matching algorithms based on models under a kind of similarity measure fuzzy similarity distance (FSD) that represents the similarity of the attributed relation between the vertex neighborhood and a certain model. Finally, the number of layer attribute for each region is obtained, and the integrated objects can be extracted using relational attributes and space information. The results show a good average integrity of objects segmented from experimental images