Segmentation of x-ray images using Probabilistic Relaxation Labeling

摘要

Segmentation is a process of separating objects of interest from their background or from other objects in an image. Without a suitable segmentation scheme, it is very difficult to detect contraband in X-ray images. In this paper, a Probabilistic Relaxation Labeling (PRL) segmentation scheme is presented and compared with other segmentation methods. PRL segmentation is an interative algorithm that labels each pixel in an image by cooperative use of two information sources: the pixel probability and the degree of certainty of its probability supported by the neighboring pixels. The practical implementation and results of the PRL segmentation on X-ray baggage images are also discussed and compared with other segmentation methods. 13 refs., 12 figs.