Photon counting imaging detectors (PCD) has paved the way for the emergence of Spectral X-ray Computed Tomography (SCT), which simultaneously measures a material’s linear attenuation coefficient (LAC) at multiple energies defined by the energy thresholds. In previous work SCT data was analysed with the SIMCAD method for material classifications. The method measures system-independent material properties such as electron density, ρe and effective atomic number, Zeff to identify materials in security applications. The method employs a spectral correction algorithm that reduce the primary spectral distortions from the raw data that arise from the detector response: charge sharing and weighting potential cross-talk, fluorescence radiation, scattering radiation, pulse pile up and incomplete charge collection. In this work, using real experimental data we analyze the influence of the spectral correction on material classification performance in security applications. We use a vectorial total variation (L∞-VTV) as a convex regularizer for image reconstruction of the spectral sinogram. This reconstruction algorithm employs a L∞ norm to penalize the violation of the inter energy bin dependency, resulting in strong coupling among energy bins. Due to the strong inter-bin correlation, L∞-VTV leads to noticeably better performance compared to bin-by-bin reconstructions including SIRT and total variation (TV) reconstruction algorithms. The image quality was evaluated with the correlation coefficient that is computed relative to ground-truth images. A positive weighting parameter defines the strength of the L∞-VTV regularization term and thus controls the trade-off between a good match to spectral sinogram data and a smooth reconstruction in both the spatial and spectral dimension. The classification accuracy both for raw and corrected data is analyzed over a set of weighting parameters. For material classification, we used 20 different materials for calibrating the SIMCAD method and 15 additional materials in the range of 6 ≤ Zeff ≤ 15 for evaluating the classification performance. We show that the correction algorithm accurately reconstructs the measured attenuation curve, and thus gives higher detection rates. We show that using the spectral correction leads to an accuracy increase of 1.6 and 3.8 times in estimating ρe and Zeff, respectively
We present an algorithm that simultaneously computes optical flow and estimates illumination change from an image sequence in a unified framework. We propose an energy functional consisting of conventional optical flow energy based on Horn–Schunck method and an additional constraint that is designed to compensate for illumination changes. Any undesirable illumination change that occurs in the imaging procedure in a sequence while the optical flow is being computed is considered a nuisance factor. In contrast to the conventional optical flow algorithm based on Horn–Schunck functional, which assumes the brightness constancy constraint, our algorithm is shown to be robust with respect to temporal illumination changes in the computation of optical flows. An efficient conjugate gradient descent technique is used in the optimization procedure as a numerical scheme. The experimental results obtained from the Middlebury benchmark dataset demonstrate the robustness and the effectiveness of our algorithm. In addition, comparative analysis of our algorithm and Horn–Schunck algorithm is performed on the additional test dataset that is constructed by applying a variety of synthetic bias fields to the original image sequences in the Middlebury benchmark dataset in order to demonstrate that our algorithm outperforms the Horn–Schunck algorithm. The superior performance of the proposed method is observed in terms of both qualitative visualizations and quantitative accuracy errors when compared to Horn–Schunck optical flow algorithm that easily yields poor results in the presence of small illumination changes leading to violation of the brightness constancy constraint.