Colour-based segmentation of carry-on baggage images

Electronic detection devices such as X-ray machines and CAT scanning devices allow screening of carry-on baggage in order to detect threat objects (handguns, knives, ammunition, etc.). Obtained images are colour coded to assist material discrimination. Image processing techniques are used to support pre-board screeners in a decision-making process. A segmentation method for extracting regions of interest from baggage images is described. The method consists of partitioning the image into homogenous regions using the split and merge algorithm. The following homogeneity predicates are used: the intensity (for regions which are surely achromatic), the hue (for surely chromatic regions) and the combination of hue and intensity (for regions which are neither surely chromatic nor achromatic). The experimental results of the segmentation of baggage images are given.