Using Image Processing Methods to Improve the Explosive Detection Accuracy

• Published in: IEEE transactions on human-machine systems 2006-11, Vol.36 (6), p.750-760
• Main Authors: Qiang Lu, Conners, R.W.
• Format: Article
• Language: English
• Subjects: Accuracy; Airports; Applied sciences; Artificial intelligence; Computer science; control theory; systems; Costs; Cybernetics; Density; Detection; Exact sciences and technology; Explosive detection; Explosives; Explosives detection; FAA; Illicit; Image processing; Image processing systems; Pattern recognition. Digital image processing. Computational geometry; Sensors; Terrorism; X-ray; X-ray detection; X-ray detectors; X-ray imaging; X-ray scattering; X-rays
• ISSN: 1094-6977; 2168-2291; 1558-2442; 2168-2305
• DOI: 10.1109/TSMCC.2005.855532

X-ray scanners are considered one of the best technologies for detecting illicit materials because of their ability to characterize a material at the molecular and atomic levels, and also because of their relatively inexpensive cost. Using X-ray technology, it is possible to determine a material's density and effective atomic number, or Z eff -related information. In theory, an illicit material can be identified using those two pieces of information. The R-L technology developed at Virginia Tech is the first true multisensing technology for explosive detection. It uses X-ray dual-energy transmission and X-ray scatter technologies to obtain characteristic values of an object; i.e., R and L. The material type of this object can then be determined using the R-L plane. R is related to Z eff and is obtained from dual-energy transmission signals. L is related to density, and is obtained using transmission and scatter signals. Compared to single-sensing technologies and pseudo-multisensing technologies, R-L technology should provide a much higher level of detection accuracy. However, the R and L values can only be computed from an object's true gray levels, which are defined as the measured gray levels of an object in different sensing modalities when it is not overlapped with any other objects. Because an object in a bag is always overlapped with many other objects, being able to identify the object of interest and remove the overlap effects becomes the key issue in determining the true gray levels of that object. This paper focuses on the development of an image processing system to determine an object's true gray levels in all the sensing modalities used in this work