Raw data processing techniques for material classification of objects in dual energy X-ray baggage inspection systems

Dual energy based X-Ray baggage inspection system (XBIS) are widely deployed for the detection of objects for security applications. We have presented a comprehensive approach for processing of the raw image data obtained from X-ray detectors for material classification and for pseudo-color representation in three classes of materials. Monte Carlo simulation studies were carried out to understand the variation of R-values with atomic number of materials. Since the R-values show dependence on the thickness of materials, an overlap of R values has been observed for materials of different atomic number. Hence, for material classification, an algorithm was developed to process the raw image data obtained from the detector system hardware using test objects. The dual energy data, i.e. low energy and high energy data obtained was integrated as one data using a wavelet based data fusion algorithm. The fused image was represented as a pseudo-colored image to indicate material classification of various objects. Standard Test Piece image and other images obtained through various processing steps were analyzed for accuracy of material classification. The data processing methods presented in this paper show improved accuracy and speed for material classification of objects in XBIS.