Advanced Model-based Algorithms for X-Ray Portal Screening
F3-A1 (Phase 1)

Download the 2013 Project Report

We present novel analysis, modeling, and reconstruction methods for X-ray based portal screening. One part of this work continues development of a new tomographic screening configuration based on limited-angle linear tomography. In contrast to conventional methods, this new configuration aims to produce a tomographic, 3D reconstruction from time sequential 2D views of moving objects. Another thrust is focused on understanding and exploitation of multi-energy X-ray CT. Part of this multi-energy work aims at a first-principles understanding of the benefits and limitations of multi-energy sensing. We demonstrate results suggesting that the use of more than two energies can be beneficial for X-ray based discrimination tasks and further that the conventionally used Compton and photoelectric basis functions may not be optimal. Another part of the work focuses on improving existing dual-energy X-ray methods. We present application of learning based methods to the problem of discrimination of explosives from non-explosives through the optimization of multi-energy sensing.


The aim of this project is to design x-ray CT image formation methods with potential application for luggage screening in airports.
- F3-A Project Team
Project Leader
  • W. Clem Karl
    Boston University

Students Currently Involved in Project
  • Limor Eger
    Boston University
  • Zachary Sun
    Boston University